nature electronics

Article https://doi.org/10.1038/s41928-023-01068-x

A conformable phased-array ultrasound

patch for bladder volume monitoring

Received: 19 January 2023 Lin Zhang1,8, Colin Marcus1,2,8, Dabin Lin 3,8, David Mejorado1,
Scott Joseph Schoen Jr 4, Theodore T. Pierce 4, Viksit Kumar 4,
Accepted: 12 October 2023
Sara V. Fernandez 1,5, David Hunt4, Qian Li4, Ikra Iftekhar Shuvo1, David Sadat1,
Published online: xx xx xxxx Wenya Du1, Hannah Edenbaum4, Li Jin 6, Weiguo Liu3, Yonina C. Eldar7, Fei Li6,
Anantha P. Chandrakasan 2, Anthony E. Samir 4 & Canan Dagdeviren 1
Check for updates

Ultrasound can be used to image soft tissues in vivo for the early diagnosis
and monitoring of disease progression. However, conventional ultrasound
probes are rigid, have a narrow field of view and are operator dependent.
Conformable transducers have been proposed, but they lack efficient
element localization and effective spatial resolution during mechanical
deformations. Here we report a conformable ultrasound bladder patch that
is based on multiple phased arrays embedded in a stretchable substrate and
can provide mechanically robust, conformable and in vivo volumetric organ
monitoring. The phased arrays use Sm/La-doped Pb(Mg1/3Nb2/3)O3–PbTiO3
ceramics as the piezoelectric material, which offers superior properties
(d33 = 1,000 pC N−1, εr = 7,500 and k33 = 0.77) than conventional piezoelectric
ceramics. We use the conformable ultrasound patch in a pilot clinical study of
bladder monitoring. Bladder volume estimation with the patch is comparable
(relative errors of 3.2 ± 6.4% and 10.8 ± 8.2% with and without ultrasound gel,
respectively) to that obtained using standard clinical ultrasound equipment,
and not requiring manual translation or rotation by an operator.

Advances in material fabrication technology and mechanical engineer- Ultrasound is widely used in clinical practice because it is safe,
ing have led to the development of high-performance piezoelectric non-invasive, non-ionizing, low cost and provides real-time imaging,
systems in conformable formats1–5. Such piezoelectric conformable monitoring and therapy. Replacing rigid transducers with conform-
electronics can exhibit real-time and continuous signal monitoring and able ultrasound transducer arrays can allow image acquisition on
decoding for use in various biomedical and healthcare applications, curved body parts, improve image quality and enable functions such
including the monitoring of skin modulus6,7, heart motion8,9, facial as long-term monitoring and therapy15–28. However, it is challenging for
motion10,11, gastrointestinal motility12, tactile stimuli13, blood pres- current conformable piezoelectric ultrasonic arrays (Supplementary
sure14,15 and blood flow16. Such technology can also be used to create Note 1 and Supplementary Table 1) to simultaneously achieve high ultra-
piezoelectric ultrasound transducers that can provide insight into deep sound image quality and mechanical stretchability due to the unpre-
tissues and curvy organs through quantitative imaging, broadening dictable change in the pitch of elements when an intrinsically stretchy
our understanding of soft-tissue disorders. surface is deformed (Supplementary Fig. 1a). Localizing the relative

1
Media Lab, Massachusetts Institute of Technology, Cambridge, MA, USA. 2Department of Electrical Engineering and Computer Science, Massachusetts
Institute of Technology, Cambridge, MA, USA. 3School of Opto-electronical Engineering, Xi’an Technological University, Xi’an, China. 4Center for
Ultrasound Research & Translation, Department of Radiology, Massachusetts General Hospital, Boston, MA, USA. 5Department of Materials Science and
Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA. 6Electronic Materials Research Laboratory, Key Laboratory of the Ministry
of Education, School of Electronic Science and Engineering, Xi’an Jiaotong University, Xi’an, China. 7Faculty of Math and Computer Science, Weizmann
Institute of Science, Rehovot, Israel. 8These authors contributed equally: Lin Zhang, Colin Marcus, Dabin Lin. e-mail: canand@media.mit.edu

Nature Electronics
Article https://doi.org/10.1038/s41928-023-01068-x

positions of transducer elements requires the use of computationally retention, long-term monitoring of at-risk patients and an adjunct tool
complex compensation algorithms, optical fibres or cameras. Stretch- for paraplegic patients who could benefit from automatic and accurate
ability between elements also requires a larger pitch (usually >0.5 mm), bladder volume measurements. The use of a stretchable substrate with
which necessitates the use of low working frequencies to avoid grating an elastic modulus comparable with the human epidermis provides
lobe artifacts28–31. Conventional ultrasound transducers also require the conformable and seamless adhesion to the human body as well as an
application of pressure to achieve sufficient acoustic contact for imag- acoustically thin layer, allowing operation without the uncomfortable
ing when placed on curved body surfaces. Therefore, an engineering ultrasound gel frequently used in ultrasound imaging.
approach is necessary to create conformable ultrasound devices that
can achieve high spatial resolution, operate at higher, more diagnosti- cUSB-Patch design for large-area imaging
cally useful ultrasound frequencies and image over a broad field of view Recent stretchable devices have used an island–bridge design to connect
(FOV), particularly in the shallow near-field region. elements by serpentine electrodes (Supplementary Fig. 1, strategy I), but
The active piezoelectric material has a vital role in the performance these face obstacles to large and precise imaging (Supplementary Note 1
of ultrasound transducers32. The majority of conformable ultrasound and Supplementary Table 1)16,20. We instead use a multi-phased-array
devices use commercial lead zirconate titanate (PZT) (Supplemen- design (Supplementary Fig. 1, strategy II), which allows the technology
tary Table 1)15–22,25,28. Compared with traditional PZT-5H ceramic with to be integrated into a wearable format and provides a spatiotemporally
relatively low piezoelectric properties (d33 < 700 pC N−1, k33 < 0.6), accurate image reconstruction technique on a larger FOV for monitoring
Pb(Mg1/3Nb2/3)O3–PbTiO3 (PMN-PT) ceramics doped with single deeper body structures. In this approach, adjacent elements are merged
rare-earth elements (La3+, Sm3+, Pr3+, Yb3+ and Eu3+) have attracted the closer to consolidate them in the form of phased arrays to achieve a
interest of researchers due to their ultrahigh longitudinal piezoelectric balance between the ultrasound imaging resolution and mechanical
coefficient and electromechanical coupling coefficients (Supple- stretchability. The phased array can be either a one-dimensional (1D)
mentary Table 3)33–35. Among reported rare-earth-doped ceramics, array or two-dimensional (2D) matrix array, with the aperture designed
Sm-doped PMN-PT has the highest piezoelectric coefficient of the to achieve a larger FOV and cover the area between adjacent arrays. The
ferroelectric ceramics34. Rare-earth-doped PMN-PT-based transduc- design exhibits both local rigidity and global malleability, maintaining
ers have been developed for piezoelectric transducer design36,37, but the imaging performance of the phased arrays and simultaneously allow-
rare-earth-doped PMN-PT ceramics have not been previously used to ing extensive surface-area coverage and mechanical deformation. The
make ultrasound transducers for conformable ultrasound devices. final image reconstruction is facilitated by the partial overlap between
Interest in using ultrasound technology for monitoring the urinary images from each phased array. Using strategy II, we recently reported
bladder through the lower abdomen has led to various wearable or a nature-inspired honeycomb patch in junction with a 1D phased array,
portable ultrasound devices being proposed (Supplementary Note 2 which offered real-time, user-friendly and continuous monitoring of the
and Supplementary Table 2)38–44. However, most of these devices use curved breast tissue for early breast cancer screening26,45.
limited numbers of elements to detect anterior–posterior organ dimen- As a proof of concept, the cUSB-Patch is designed for real-time
sion (A mode), resulting in imprecise volume estimation, difficulty with bladder volume monitoring, an assessment that requires large-area
positioning the transducer and a limited view of the target organ. In continuous monitoring with high resolution, using novel Sm/
addition, most portable systems still have a rigid, planar configuration, La-PMN-PT ceramics. As shown in Fig. 1d–f and Supplementary Fig. 2,
resulting in poor contact over curved body surfaces. For handheld the cUSB-Patch constitutes a set of five 1D phased arrays: three arrays
probes, although they can obtain real-time bladder images, it is still positioned in the transverse (horizontal) plane with a 5 cm array-to-array
required to rotate in two orientations (transverse and sagittal planes) pitch and two arrays in the sagittal (vertical) direction with the same
to achieve a comprehensive view of the bladder (Fig. 1a). Despite high pitch. All the transducers in the transverse plane are oriented in the same
levels of training and experience, operator-to-operator measurement direction, whereas the transducers in the sagittal plane are rotated 90°
variability remains problematic. For these reasons, a conformable away from the centre array. Each array acquires a 2D slice of the tissue
ultrasound device that can seamlessly conform to the body and decode volume, with the slice width approximately equal to the array width.
the signals from both skin surface and deep tissue is desirable. The imaging planes of several arrays can be combined to produce a
In this Article, we report a conformable ultrasound bladder unified image with improved FOV and resolution, especially close to
patch (cUSB-Patch) that is mechanically adaptive and can provide the the surface. Assuming the beam steering angle is greater than 30°,
real-time in vivo monitoring of bladder volume (Fig. 1b,c and Supple- three arrays can cover a greater area than a conventional commercially
mentary Fig. 2). We develop a Sm/La-doped PMN-PT ceramic material available curved C5-2v or C1-6-D convex probe (Supplementary Fig. 2b).
with high piezoelectric properties (d33 = 1,000 pC N−1, εr = 7,500 and According to the patch design (Supplementary Fig. 3), the mould for
k33 = 0.77) and use it to fabricate a 64-element phased-array transducer the substrate consists of five elevated platforms with dimensions of
with an operational frequency of 2–5 MHz (Fig. 1g–j). The patch offers 20.0 mm × 20.0 mm × 4.0 mm, which matches the size of each array. The
mechanically adaptive and real-time in vivo monitoring of bladder thin layer of biocompatible silicone rubber (Ecoflex 00-30, Smooth-On)
volume (Fig. 1d–f). To avoid the requirement for manual rotation, the on top of the matching layer, prepared by dip coating, has two functions:
cUSB-Patch can simultaneously obtain images in two orientations. The one is to hold each array and the other is to provide acoustic coupling
device can image the urinary bladder at a depth up to 15 cm with an axial between the transducer and the target surface, which minimizes the
resolution of less than 0.6 mm. Theoretical and in vitro experimental need for the ultrasound gel. The overall thickness of the patch is less
studies establish that the cUSB-Patch with five such arrays can provide than 4.5 mm, but it is thicker than many Ecoflex-based electronic skins
accurate and reproducible imaging on different phantoms (planar and (<2 mm)24,28,46, allowing for embedding the entire array and still achiev-
curved), with and without an ultrasound gel. ing enough stretchability and stress recovery from mechanical deforma-
In a pilot clinical study, the cUSB-Patch centre transducer is shown tion. When the patch is subjected to twisting, bending and lamination
to reliably locate the initial centre position of the bladder, whereas its onto the surfaces of a curved phantom (Supplementary Fig. 4 and
five arrays image the bladder from multiple angles, enabling real-time Supplementary Video 1), the relative position of elements per array
imaging and measurement without the need for transducer motion remains stable, guaranteeing a consistent ultrasound beamforming
or rotation. The cUSB-Patch can image other deep tissues without from each array and operator independence because it does not neces-
modification, and the architecture supports imaging shallower tis- sitate manually applied pressure or scanning. In addition, when applied
sues by a simple adjustment of design parameters. Candidate clinical on the subject, the patch showed robust mechanical performance dur-
applications include the diagnosis and quantification of acute urinary ing bending with connected cables and when attaching and detaching

Nature Electronics
Article https://doi.org/10.1038/s41928-023-01068-x

a b c

Ultrasound be
am

h
-Patc
Handheld probe

cUSB
cUSB-Patch

l
wal
Bladder

rior

l
al
e

rw
Ant

io
er
Transverse plane

st
(Trans) Trans Sag

Po
Sagittal plane
(Sag)

d e f

g i

Second ML
Flexib
First ML le ca

Flexible cable
ble
Top electrodes Top electrodes
(64 traces)
Ceramic elements
Bottom electrodes

BL
j

Second ML
h First ML

Elements

BL

Fig. 1 | Overview of the phased array and cUSB-Patch. a, Schematic of the with the view of the backing layer’s side (e) and the image of the cUSB-Patch
operation of a handheld probe on the human’s lower abdomen for bladder under twisting (f). g, Schematic of a single, ultrasound phased array with the key
imaging. To obtain a comprehensive view of the bladder in standard clinical components labelled. The 1D linear high-performance piezoelectric elements are
practice, the handheld probe has to image in two orientations (transverse embedded in the epoxy matrix with a shared bottom electrode and independent
plane and sagittal plane) by rotating the ultrasound probe. b, Schematic of the top electrodes. A customized flex PCB cable (fabricated by PCBWay) is bonded
cUSB-Patch on the human’s lower abdomen for bladder imaging. To avoid the with the electrodes. The backing layer (BL) and two matching layers (MLs) are
requirement for manual rotation, the cUSB-Patch can simultaneously obtain bonded on the bottom and top sides, respectively. h, Optical image of a single
the images in two orientations. c, Schematic of a cross-sectional view of the phased array with the flexible cable. i, Optical image of the top view of the array
cUSB-Patch on the lower abdomen. The phased arrays can locate the anterior with electrodes. The top electrode (64 traces) is well aligned with traces on the
and posterior walls of the bladder and cover the entire FOV of the bladder flexible cable. j, Optical image of the cross-section of the array, including diced
using multiple ultrasound beams. d–f, Natural and flexible form factor of the La/Sm-PMN-PT ceramic elements with filled epoxy, a backing layer and two
cUSB-Patch: on hand with the view of the matching layer’s side (d), on hand matching layers. Scale bars, 1 cm (d–f, h); 2 mm (i); 200 µm (j).

several times on the subject’s lower abdomen (Supplementary Fig. 4 a high-temperature solid-state process with a hot-pressing technology
and Supplementary Video 2). (Methods). The selection of La as the second doping element is moti-
vated by the discovery that the width of the micro–macro polar tran-
Piezoelectric performance of Sm/La-doped sition range in PMN-0.15PT ceramics increased after La modification,
PMN-PT ceramics resulting in large electromechanical responses over a broader tempera-
In this work, PMN-PT ceramics doped with two rare-earth elements Sm ture range47. To achieve both high piezoelectric properties and a com-
and La (0.5–1.0 mol%) (termed Sm/La-PMN-PT) were synthesized using prehensive range of temperature stability, especially for compositions

Nature Electronics
Article https://doi.org/10.1038/s41928-023-01068-x

a b c d
Sm/La-PMN-PT 200 180
Tc 1,000
Cubic
150

Temperature (°C)
d33
150 800

d33 (pC N–1)

120

Tc (°C)
600
100 90
Tetragonal
60 400
50
30 200
Pseudocubic
0 0 0
28 30 32 34 36 28 29 30 31 32 33 34 35 36

Pb content (%) Pb content (%)

e 90 f 400 90 g 0.4
1M 187 kHz 4.72 MHz Sm/La-PMN-PT
60 60 Sm-PMN-29PT

Impedance (‹)
Impedance (‹)

300 0.3 PMN-35PT

100k 30 30

Strain (%)
Phase (°)

Phase (°)
PZT-5H
10k 0 200 0 0.2

–30 –30
1k 100 0.1
–60 –60
123 kHz 3.90 MHz
100 –90 0 –90 0
100 150 200 3 4 5 6 0 5 10 15 20

Frequency (kHz) Frequency (MHz) Electric field (kV cm–1)

h i 0.25
0 °C 20 °C 40 °C 60 °C 80 °C 100 °C j kp kt k33 d33
1,200
Sm/La-PMN-PT
0.8
1.5 Sm-PMN-29PT 0.20
Variations in εr

PMN-35PT ∆ > 100.0% 1,150

d33 (pC N–1)

0.7
Strain (%)

kp, kt, k33

PZT-5H 0.15
1.0
1,100
∆ = 69.8% 0.6
0.10
0.5 ∆ = 47.1%
0.5 1,050
0.05
∆ = 14.7%
0 0 0.4 1,000
0 20 40 60 80 100 0 15 0 15 0 15 0 15 0 15 0 15 0 20 40 60 80 100
–1
Temperature (°C) Electric field (kV cm ) Temperature (°C)

Fig. 2 | Characterization of morphological, piezoelectric and dielectric e, Frequency dependence of impedance and phase for a longitudinal bar
properties of Sm/La-PMN-xPT ceramics. a, Scanning electron microscopy (2 mm × 2 mm × 8 mm). f, Frequency dependence of impedance and phase for
(SEM) image of the cross-section of Sm/La-PMN-PT (Sm = La = 0.75%) ceramics. a flat plate (12 mm × 12 mm × 1 mm). g, Comparison of electric-field-induced
The inset shows the optical photograph of the ceramic disc with a high density strain of Sm/La-PMN-PT (Sm = La = 0.75%), Sm-PMN-29PT, PMN-35PT and PZT-5H
of 7.92 g cm−3. b, Zoomed-in SEM image of the ceramic grains. The red dashed ceramics, measured at 1 Hz at room temperature. h, Temperature dependence of
line indicates the zoomed-in area marked in a. The selected SEM images are the variation in relative dielectric constant (εr) for Sm/La-PMN-35PT, Sm-PMN-
representative of the cross-section observation of five different specimens 29PT, PMN-35PT and PZT-5H ceramics. i, Electric-field-induced strain behaviour
with the same doping amount and composition. c, Phase diagram of Sm/La- of the Sm/La-PMN-PT (Sm = La = 0.75%) ceramic at selected temperatures: 0,
PMN-xPT. The red dashed line indicates the composition (x = 35%) selected 20, 40, 60, 80 and 100 °C. The black dashed line indicates the variation (15%) in
in this work. d, Curie temperature Tc and piezoelectric coefficient d33 of Sm/ the maximum strain value with temperature. j, Temperature dependence of the
La-PMN-xPT (x = 28–36). The composition with x = 35 (red dashed line) shows piezoelectric coefficient d33 and electromechanical coupling factors kp, kt and k33
the best performance and was selected for transducer fabrication in this study. for Sm/La-PMN-PT (Sm = La = 0.75%) ceramics. Scale bars, 0.5 cm (a); 20 µm (b).

near the morphotropic phase boundary48, Sm/La-PMN-xPT ceramics and electromechanical coupling factors (kp and kt) increase to 7,500,
with different contents of PT (x = 28–36) and Sm/La doping (0.50%, 0.62 and 0.58, respectively, whereas the dielectric loss (tanδ) value
0.75% and 1.00%) were systematically investigated (Supplementary remains relatively low (<0.02) (Supplementary Fig. 7). Additionally,
Note 3). As shown in Supplementary Fig. 5, and with the micropores Fig. 2e,f demonstrates the resonance and antiresonance frequency
and grain size also playing important roles in shaping the properties features of the impedance and phase in the longitudinal bar and flat
of ceramics, the enhanced piezoelectric performance is ascribed to the plate, respectively, validating both enhanced electromechanical and
dense and uniform microstructure (Fig. 2a,b). Due to La modifications, piezoelectric properties as expressed by larger d33. The electrome-
Sm/La-PMN-PT (Sm = La = 0.75%) ceramics exhibit well-developed chanical coupling factors (k33 = 0.77 and kt = 0.58) are calculated from
density (7.92 g cm−3), fine-grain microstructures, homogeneous ele- these obtained data, indicating substantially larger electromechanical
ment distribution (Supplementary Fig. 6) and the best performance coupling factors compared with PZT-5H and undoped PMN-PT ceramics
compared with other doping amounts (Supplementary Table 3), which (Supplementary Table 3). To illustrate the comparative piezoelectric
would benefit device performance for different applications such as response, electric-field-induced strain measurement with different
actuators and piezoelectric transducers. According to the phase dia- ceramics was carried out (Fig. 2g). The Sm/La-PMN-PT (Sm = La = 0.75%)
gram shown in Fig. 2c, the pseudocubic–tetragonal phase boundary of ceramic exhibited a larger strain (21%) than those of the widely used
the PMN-PT system is shifted to a lower PT content by doping rare-earth PMN-35PT and PZT-5H ceramics34.
elements, and the maximum value is observed on the tetragonal side of Comparing between the four representative ceramics, the
the morphotropic-phase-boundary region. For this reason, the Curie results listed in Supplementary Table 3 indicate that the dielectric
temperature (Tc) value increases as the Pb content increases, whereas and piezoelectric performances of the Sm/La-PMN-PT (Sm = La = 0.75%)
the d33 value first increases and reaches the peak value of 1,000 pC N−1 ceramic are better than both commercial PZT-5H and PMN-32PT ceram-
at room temperature at x = 35, and then decreases (Fig. 2d). In addi- ics but slightly lower than the performance of the state-of-the-art
tion, around the Pb content of 35%, the relative dielectric permittivity Sm-PMN-29PT (Sm = 2.5%) ceramics. However, the co-doped ceramic

Nature Electronics
Article https://doi.org/10.1038/s41928-023-01068-x

shows a much higher Tc than the Sm-doped one, indicating better ther- firmly bonded by the epoxy solution under high-temperature pressure
mal stability (Supplementary Fig. 8). Compared with Sm-PMN-29PT (Fig. 1j). Supplementary Note 4 and Supplementary Table 6 discuss the
(Sm = 2.5%), the Curie peak shifts to a higher temperature and sharpens detailed simulation information and acoustic properties of each layer in
due to La modifications and higher PT content. The Sm/La-PMN-PT the stack, and Supplementary Note 5 describes the fabrication process
(Sm = La = 0.75%) exhibits a low dielectric variation (14.7%) over the and characterization methods. Consistency of array performance is
temperature range of 0–100 °C (Fig. 2h) and much improved thermal essential for high-quality ultrasonic imaging. The array’s electrical
stability of the dielectric permittivity compared with other selected impedance and acoustic performances were measured to investigate
ceramics34, which was also confirmed by the 15% change in the the uniformity of the array elements. The piezoelectric and acoustic
electric-field-induced strain in the same broad temperature range (Fig. 2i performances of all the elements of both Sm/La-PMN-PT-based and
and Supplementary Fig. 9). To elucidate the phase transition state at var- PZT-5H-based arrays are shown in Fig. 3a–c and Supplementary Fig. 19.
ious temperatures, temperature-dependent X-ray diffraction (XRD) was Supplementary Note 5 discusses the detailed simulation and experi-
evaluated at temperatures from 0 to 100 °C (Supplementary Fig. 10). mental analyses. Clearly, the Sm/La-PMN-PT array not only has the bet-
The sample exhibited mixed tetragonal (P4mm) and rhombohedral ter electromechanical properties (larger effective electromechanical
(R3m) phases according to the Rietveld refinement of the XRD data coupling coefficient keff = 0.62) compared with the PZT array (keff = 0.52)
(Supplementary Fig. 11). The tetragonal a and c parameters showed but also has a higher bandwidth (12% improvement), suggesting that
slight changes with increasing temperature (Supplementary Fig. 12a). the Sm/La-PMN-PT array transducer has higher acoustic performance
The tetragonal phase fraction (94%) and rhombohedral phase fraction for ultrasonic imaging.
(6%) remained almost constant in temperatures ranging between 0
and 100 °C (Supplementary Fig. 12b), demonstrating the notable ther- Ultrasound imaging on phantoms
mal stability observed in Sm/La-PMN-PT (Sm = La = 0.75%) ceramics. To achieve real-time imaging, a data acquisition interface linked the
Consequently, the electromechanical coupling factors (kp, kt and k33) array and cUSB-Patch with a Vantage 256 ultrasound system (Verason-
exhibited a minimal change (<5%) over the temperature range of ics). Before performing human clinical trials, the fabricated arrays
0–100 °C (Fig. 2j). In summary, the full matrix of dielectric, elastic and and cUSB-Patch were tested on three distinct phantoms to evaluate
piezoelectric constants of 0.75%Sm/La-PMN-35PT ceramic was meas- their acoustic performance and imaging capability. A phantom with
ured by the resonance technique and used to construct the complete fishing lines in a water tank was used to examine the resolution at vari-
property matrices (Supplementary Note 3 and Supplementary Table 4). ous depths (Supplementary Fig. 25 and Methods). The simulation and
As demonstrated in PMN-PT ceramics with a composition close to the experimental findings of the ultrasound point-spread function from
morphotropic phase boundary, adding La can reduce the grain size a single array placed at the top of the wire targets are depicted in Sup-
and enhance the density49. The results indicate that the co-doped Sm/ plementary Fig. 26 and described in Supplementary Note 6. Images
La-PMN-PT (Sm = La = 0.75%) ceramics (abbreviated as Sm/La-PMN-PT are scaled with a 30 dB dynamic range from the brightest point in each
hereafter) possess a large piezoelectric coefficient (d33 = 1,000 pC N−1), image. The resolution is obtained from the measured 6 dB region, cor-
high dielectric permittivity (εr = 7,500), excellent electromechani- responding to the normalized curve’s full-width at half-maximum.
cal coupling factor (k33 = 0.77), high phase transition temperature Comparing the simulation and experimental results (Supplementary
(Tc = 150 °C) and high temperature stability. Thermal stability can Fig. 27) in both axial and lateral dimensions, the phased array demon-
guarantee less performance degradation during device processing, strated a relatively high degree of resolution. For instance, it could
packaging and transportation, as well as broad temperature usage in still achieve 0.6 mm of axial resolution at −18.2 cm and the same lat-
different environments. eral resolution at −10.7 cm, which is suitable for imaging deep tissues
and organs. However, the lateral resolution has a large difference
Design and performance of ultrasound arrays between the experimental results and the simulation results at lower
The design of the 1D phased array is based on the traditional trans- depth targets due to the split lobe. To resolve this issue, the strategy
ducer three-port network32 (Fig. 1g), including the active piezoelectric of multiple-phased-array design is proposed to keep similar lateral
element, backing layer and two matching layers. The Sm/La-PMN-PT resolution during the increased depth.
ceramic was chosen for phased-array transducer fabrication, whereas To illustrate the FOV and resolution, an ultrasound phantom with
another commercially available piezoelectric material, namely, PZT-5H, a planar surface containing both wire and cylinder targets (model
was also selected (Baoding HongSheng Acoustics Electron Apparatus 040GSE, Computerized Imaging Reference Systems) was used. This
Co., Ltd, China) as a benchmark comparison (Supplementary Table 5). phantom includes near-field targets, resolution targets, hyperechoic
The suggested transducer frequency for bladder imaging from the and hypoechoic targets, as well as other types of target (Supplementary
literature is around 2–5 MHz (Supplementary Table 1). Supplementary Fig. 28). The commercial Verasonics C5-2v curved array probe, com-
Fig. 13 shows the geometry of the 1D phased array with the major com- monly used for imaging deep tissue, is used here as a point of reference
ponents. The 1D phased array with 64 elements and a working frequency (Fig. 3d,e). This figure shows that a single phased array has a larger FOV
of 3.5 MHz is proposed to balance the spatial resolution and depth for and similar imaging depth of ~14 cm compared with the C5-2v probe
bladder imaging. The corresponding ultrasound wavelength (λ = c/f, when not applying pressure and scanning across the 040GSE phantom
c = 1,540 m s−1 in water) in the soft tissue for 3.5 MHz is calculated as (Fig. 3f,g and Supplementary Video 3), which is probably sufficient for
440 µm. To meet the maximum half-wavelength requirement for the bladder imaging in most people. The cUSB-Patch can offer a larger FOV
phased array to avoid side lobes50, the pitch is selected as 220 µm and (Fig. 3h) than the C5-2v probe and single array. Because the primary
it equals the element width (170 µm) plus the kerf (50 µm). During the regions of interest scanned across all the transducers are the vertical/
dicing process, the 30-µm-thick synthetic diamond blade achieved horizontal distance targets and axial/lateral resolution targets, each
50 µm space due to the vibration of the dicing blade. The aspect ratio array acquired three typical images of the targeted areas and recon-
of width/thickness is smaller than 1/3, which meets the requirement for structed a complete 2D image (Fig. 3h and Supplementary Fig. 29).
desired thickness vibration37. The element length is selected as 14 mm The resolution of the vertical/horizontal distance targets are listed
(64 times the pitch) to obtain an acoustic beam with minimal spreading in Fig. 3i. It is clear that the axial resolution is less than 1.0 mm at both
within the planned imaging depth. The flexible cable with the same 40 cm and 90.0 mm depth and the minimum value is around 0.6 mm
electrode traces is aligned to the electrode on the array and placed (Fig. 3j,k), which is close to the results obtained from the phantom
between the elements and matching layer (Fig. 1i). The cross-section with fishing lines. When we check the axial/lateral resolution targets
view shows that all the components have a uniform thickness and are at 3.0 cm depth (Fig. 3i(iii)) and 6.5 cm depth (Fig. 3i(iv)), the arrays

Nature Electronics
Article https://doi.org/10.1038/s41928-023-01068-x

a –20 b 6 c 0.8
Resonance freq. Sm/La-PMN-PT array
300 Antiresonance freq. PZT-5H array

Frequency (MHz)
0.7

Impedance (—)
–40 5

Phase (°)
fa

keff
0.6
200 –60 4

0.5
fr
–80 3

100 0.4
1 2 3 4 5 6 7 0 10 20 30 40 50 60 0 10 20 30 40 50 60

Frequency (MHz) Element number Element number

C5-2v on 040GSE Arrays on 040GSE

d f Phased arrays h
C5-2v
3
40 mm

e g 90 mm

1 2

i j Horizontal targets at 40 mm k Horizontal targets at 90 mm

6 6
(i) (iii) dB Axial Axial
30 5 5
Lateral Lateral
Resolution (mm)

Resolution (mm)
24 4 4

3 3
18
2 2
(ii) (iv)
12
1 1

6 0 0
–30 –20 –10 0 10 20 –80 –60 –40 –20 0 20 40

0 Lateral distance (mm) Lateral distance (mm)

Fig. 3 | Piezoelectric and acoustic performances of phased-array transducers. on the top surface of the planar phantom with ultrasound gel. h, Reconstructed
a, Frequency dependence of the measured impedance and phase angle of the image from three arrays on the top surface of the planar phantom with
single element of the Sm/La-PMN-PT array. b, Resonance and antiresonance ultrasound gel. i, Zoomed-in images from h: 1 and 2 are the lower horizontal
frequencies of 64 elements of the Sm/La-PMN-PT array. c, Comparison of keff targets and 3 and 4 are the axial/lateral resolution targets. j,k, Axial and lateral
of the Sm/La-PMN-PT array and PZT-5H array. d,e, Optical photograph (d) and resolutions of the horizontal targets (distance between array, 5 cm): horizontal
image (e) of the C5-2v probe on the top surface of the planar phantom with targets at 40 mm (j); horizontal targets at 90 mm (k). Scale bar, 5 cm (d, f),
ultrasound gel. f,g, Optical photograph (f) and image (g) from the single array 1 cm (e, g, h), 2 mm (i).

separated targets with gaps as small as 0.5 mm in the axial direction gel greatly impacts daily usage and cleaning of the device, and a lower
and 2.0 mm in the lateral direction at 3.0 cm depth, and 0.5 mm in the applied voltage is crucial for a portable device. When we replaced
axial direction and 3.0 mm in the lateral direction at 6.5 cm depth. If the ultrasound gel with a thin Ecoflex layer (180–200 µm) for acous-
numerous arrays are positioned and aligned at measured intervals on tic coupling, the array produced images of comparable quality with
the phantom, aggregated images of the entire phantom can be obtained those obtained using gel (Supplementary Fig. 33). Eliminating the
(Supplementary Fig. 30). This was accomplished by sliding a single requirement for the acoustic coupling gel will save time, reduce costs
array across the surface in ~1 cm increments, followed by manual image and improve accessibility of the technology. We varied the applied
registration (Supplementary Fig. 31). In addition, when we change the voltage from 5 to 50 V to compare the quality of the scan image (Sup-
distance between arrays from 4 to 5 cm, there is no obvious difference plementary Fig. 34). Targets at depths of 7 cm can still be observed at
in the imaging resolutions (Fig. 3j,k and Supplementary Fig. 32), which 5 V, demonstrating the potential applicability of this technology to the
indicates that the patch can still maintain a similar image quality during point-of-care setting, where energy efficiency and low-power require-
short-distance stretchability. ments are key for successful clinical development.
Increasing the number of arrays expands the FOV and improves To evaluate the imaging quality of cUSB-Patch on a curved surface,
the wide-angle resolution, suggesting that the cUSB-Patch technology the US-18 fundamental ultrasound phantom (model US-18, Kyoto
has a pathway to clinical utility for additional applications that require Kagaku) was used (Supplementary Fig. 35). We first evaluated the C5-2v
wide-FOV imaging. In addition, we examined different scan settings probe using this phantom. With the C5-2v probe, achieving a broad FOV
(with/without gel, an applied voltage) for practical applications, as required applying a substantial amount of force (Supplementary Fig. 36)

Nature Electronics
Article https://doi.org/10.1038/s41928-023-01068-x

a b c d

C5-2v C5-2v

US-18

e f g h
Multiplexer box

4 2 5

Fig. 4 | Real-time ultrasound imaging by the C5-2v commercial probe and and horizontal orientations, respectively. g,h, The 2D reconstructed images on
the cUSB-Patch on the oval phantom. a,b, Optical photograph (a) and image the oval phantom. The orange dashed line indicates the constructed image along
(b) of the C5-2v probe on the top surface of the oval phantom with an ultrasound the vertical orientation and the blue dashed line indicates the constructed image
gel along the vertical direction. c,d, Optical photograph (c) and image (d) of the along the horizontal orientation, respectively. In all the ultrasound images, the
C5-2v probe on the top surface of the oval phantom with ultrasound gel along white dashed lines indicate the FOV. The red dashed circle indicates the 2D view
the horizontal direction. e, Optical photograph of the cUSB-Patch on the oval of the spherical object, the yellow dashed triangle indicates the 2D view of the
phantom. f, Schematic of the top view of the oval phantom and embedded 3D tetragonal object and the green dashed rectangle indicates the 2D view of the
objects. The five black squares indicate the location of each array, and the black rectangular object. The ultrasound gel was not applied to the US-18 fundamental
thin-line pattern in each black square indicates the orientation of 64 elements. ultrasound phantom for cUSB-Patch imaging. Scale bar, 5 cm (a, c, e, f),
The orange dashed line and blue dashed line indicate imaging along the vertical 1 cm (b, d, g, h).

and manual rotation of the probe to capture images in different orien- full, after partial voiding and after complete voiding. The full and par-
tations (Fig. 4a–d). When we used the cUSB-Patch to cover a partially tially empty bladder measurements are referred to as case I and case
curved surface (Fig. 4e), several three-dimensional (3D) objects were II, respectively, and were independently analysed. For each case, the
covered by the FOV of the patch (Fig. 4f and Supplementary Fig. 37). Five imaging sequence was as follows: (1) C1-6-D probe without ultrasound
individual images can be simultaneously obtained to detect distinct gel, (2) cUSB-Patch without gel, (3) cUSB-Patch with gel and (4) C1-6-D
3D objects (Supplementary Fig. 38). The spherical object is visible in probe with gel. The volume estimated by the conventional probe with
all the five images, whereas the tetragonal and rectangular objects gel was considered as the ground truth for error calculations. After the
are only visible in the left and right images, respectively. Arrays with final measurement in each case, a wipe (Sani-Cloth AF3 Germicidal
mutual image planes were registered and combined, and the resulting Disposable Wipe) was used to wipe away any gel residue on the subject
larger images were orthogonally combined to produce a 3D image and clean the surface of the cUSB-Patch. Methods and Supplementary
(Supplementary Fig. 39 and Supplementary Video 4). The 3D image Table 10 show the detailed scanning protocol.
rendering was performed using Napari, a Python library that provides Subjects A, B, C and D are shown as representative samples (Fig. 5
capabilities for volumetric image visualization and graphics process- and Supplementary Figs. 42–63), comprising two males and two
ing unit acceleration. Comparisons between the imaging operation of females within a BMI range of 18.19–27.98 kg m−2 and an age range of
the C5-2v probe and cUSB-Patch demonstrate that the cUSB-Patch has 24–45 years. The width (W), length (L) and height (H) of the bladder
several distinct advantages when imaging on curved surfaces, includ- were measured in each case, and the volume was calculated according
ing operator independence (as the cUSB-Patch does not necessitate to the equation51,52 V = W × L × H × 0.7, matching the method used with
manually applied pressure or scanning), less image distortion and the C1-6-D probe on LOGIQ E10. Supplementary Figs. 42–63 show the
larger FOV by image reconstruction. The phantom data revealed that sonographic volume-estimating images from the C1-6-D probe and
our phased arrays and cUSB-Patch are suitable for clinical studies on cUSB-Patch for four subjects and Supplementary Videos 5–8 show
the human bladder. the representative videos for continuous monitoring on subject A.
The statistical results for these subjects are shown in Fig. 5e–p. The
Clinical real-time imaging of human urinary mean volumes recorded with the cUSB-Patch without gel are greatly
bladder improved compared with the commercial C1-6-D probe without gel but
A human clinical study (Institutional Review Board (IRB) #2021P001706) comparable with its performance with gel. The statistical results for
was performed to evaluate the performance of cUSB-Patch, compar- these subjects are shown in Fig. 6a,b. The LOGIQ E10 and C1-6-D probe
ing the bladder volume estimation with a reference clinical system produced images without gel in many cases by increasing the gain
(GE LOGIQ E10 with the C1-6-D probe), and to assess the performance setting from 25 to 35 dB up to 90 dB. Nevertheless, the mean volume
of each system with and without the use of ultrasound gel (Scan 11-08 measurement error without gel was found to be −47.5 ± 34.9%. This was
Ultrasound Gel, Parker Laboratories). Human subjects ages 18–64 years mainly due to a reduced FOV without gel, and in a few cases, no image
were recruited with body mass index (BMI) ranging from 17 to 30 kg m−2, could be obtained (treated as −100% error). When the cUSB-Patch was
and 60% of the subjects were women. The bladder was imaged when used with gel, the volume measurement was similar to the ground-truth

Nature Electronics
Article https://doi.org/10.1038/s41928-023-01068-x

a b c d

Multiplexer
box

Strap

Transverse plane Sagittal plane cUSB-Patch on the cUSB-Patch on the

(Trans) (Sag) female subject male subject
Front view Side view

Subject A (Female)

e Case I: cUSB-Patch with gel f Case I: cUSB-Patch w/o gel g Case I: comparison
15 C1-6-D with gel cUSB-Patch with gel
H H C1-6-D w/o gel cUSB-Patch w/o gel
12

Size (cm)
9
W L
L 6
W
3

0
Length Height Width
Trans Sag Trans Sag
Dimension

h Case II: cUSB-Patch with gel i Case II: cUSB-Patch w/o gel j Case II: comparison
15 C1-6-D with gel cUSB-Patch with gel
H H
C1-6-D w/o gel cUSB-Patch w/o gel
12

Size (cm)
W L L 9
W
6

0
Trans Sag Trans Sag Length Height Width

Dimension

Subject D (Male)

k Case I: cUSB-Patch with gel l Case I: cUSB-Patch w/o gel m Case I: comparison
15 C1-6-D with gel cUSB-Patch with gel
H C1-6-D w/o gel cUSB-Patch w/o gel
12
Size (cm)

9
W L W 6
L
3

0
Trans Sag Trans Sag H Length Height Width

Dimension

n Case II: cUSB-Patch with gel o Case II: cUSB-Patch w/o gel p Case II: comparison
15 C1-6-D with gel cUSB-Patch with gel
H H C1-6-D w/o gel cUSB-Patch w/o gel
12
Size (cm)

9
L
L
W W 6

0
Trans Sag Trans Sag Length Height Width

Dimension

Fig. 5 | Real-time ultrasound imaging on the bladders of different subjects. by the C1-6-D probe and cUSB-Patch for subject A in case II. k,l, Results for subject
a,b, Schematic of the working principle and FOV of the cUSB-Patch on the D (male). The ultrasound images of full bladders for case I: the cUSB-Patch with
bladder. Three arrays collect images along the transverse plane (red dashed line), gel (k) and without gel (l). m, Comparison of length, height and width of the
and two arrays collect images along the sagittal plane (blue dashed line). c,d, bladder obtained by the C1-6-D probe and cUSB-Patch for subject A in case I.
The cUSB-Patch, which is connected to the multiplexer box, was applied to the n,o, Ultrasound images of full bladders for case II: the cUSB-Patch with gel (n)
female (c) and male (d) subjects during the clinical study. e,f, Results on subject and without gel (o). p, Comparison of length, height and width of the bladder
A (female). The ultrasound images of the full bladders for case I: the cUSB-Patch obtained by the C1-6-D probe and cUSB-Patch for subject A in case I. The depth of
with gel (e) and without gel (f). g, Comparison of length, height and width of the all the B-mode images is 15 cm. On all the B-mode images, the white solid closed
bladder obtained by the C1-6-D probe and cUSB-Patch for subject A in case I. h,i, loop shows the bladder wall, and the white dashed lines show the measurement
Ultrasound images of full bladders for case II: the cUSB-Patch with gel (h) and of dimensions. The red dashed line in g, j, m and p indicates the value obtained
without gel (i). j, Comparison of length, height and width of the bladder obtained from C1-6-D with gel as the baseline. Scale bar, 1 cm (e, f, h, i, k, l, n, o).

Nature Electronics
Article https://doi.org/10.1038/s41928-023-01068-x

a
C1-6-D with gel cUSB-Patch with gel

1,000 C1-6-D w/o gel cUSB-Patch w/o gel

Bladder volume (ml)

800 Subject A (female) Subject B (female) Subject C (male) Subject D (male)

600

400

200

0 Case I Case II Case I Case II Case I Case II Case I Case II

b C1-6-D cUSB-Patch
c C1-6-D probe cUSB-Patch
90 120
A (case I)

Average acoustic loss (dB)

A (case II)
60
90
Normalization (%)

B (case I)
30 +20%
B (case II)
0 C (case I) 60
C (case II)
–30 –20%
D (case I) 30
–60
D (case II)
–90
0
Case I Case II Smooth Hair Hair
skin (average) (array 5)
l

el

el
ge

ge
tg

tg
ith

ith
ou

ou
W

W
ith

ith
W

Fig. 6 | Summary and comparison of the calculated bladder volumes of eight different tests. The volume of the C1-6-D probe with gel on the full bladder
four subjects. a, Comparison of the calculated bladder volumes. The volume was used as the reference. c, Comparison of average acoustic loss in several no-
was reported as zero when it could not be measured. There are eight tests for gel imaging scenarios. The C1-6-D probe in case I shows the highest loss, followed
each subject, including C1-6-D probe with gel on a full bladder, C1-6-D probe by case II, which may be lower due to gel residue. For the cUSB-Patch, the lowest
without gel on a full bladder, C1-6-D probe with gel on a partially voided bladder, loss occurred for subjects with a smooth skin, followed by subjects with hair on
C1-6-D probe without gel on a partially voided bladder, cUSB-Patch with gel on their skin. Array 5 encountered the most hair due to being lowest on the abdomen
a full bladder, cUSB-Patch without gel on a full bladder, cUSB-Patch with gel and correspondingly shows the highest loss. The data are presented as mean
on a partially voided bladder and cUSB-Patch without gel on a partially voided values ± standard deviation (C1-6-D probe, n = 12 different areas; cUSB-Patch,
bladder. b, Normalization of the calculated bladder volumes of four subjects in n = 5). The circular dots show all the data points in each case.

value with a mean accuracy of −3.2 ± 6.4%. When used without gel, use a rare-earth-element-doped ceramic, namely, Sm/La-PMN-PT
the cUSB-Patch demonstrated better performance than the C1-6-D (Sm = La = 0.75%), with high dielectric and piezoelectric properties
probe without gel, achieving a mean accuracy of −10.8 ± 8.2%. This and a thermally stable range that make it suitable for the high fre-
result is particularly notable because unlike the LOGIQ E10 system, the quencies used in clinical ultrasound imaging. These performance
cUSB-Patch was used with a constant 40 dB gain setting, demonstrating characteristics also make the ceramic promising for other electro-
the robustness and repeatability of the imaging without gel. mechanical applications at room and elevated temperatures. The Sm/
Another objective of the study was to evaluate and compare the La-PMN-PT-ceramic-based phased array exhibited larger bandwidth
acoustic coupling performance of the C1-6-D probe and cUSB-Patch compared with PZT-5H-based phased arrays, which improves the detec-
with and without ultrasound gel on the skin. This was performed by tak- tion depth and resolution. Designing multiple arrays embedded in a
ing the corresponding gel and no-gel images and measuring the change soft substrate allowed imaging over a wide FOV and simultaneously on
in image brightness (Methods). As shown in Fig. 6c and Supplemen- multiple axes, allowing full organ visibility without manual scanning
tary Table 12, the acoustic loss was low when coupling to the flat-top or rotation by an operator.
phantom (<5 dB; Supplementary Fig. 31), moderate when coupling to The bladder patch, when used with ultrasound gel, measured
a smooth skin surface (15.3 ± 7.2 dB) and slightly higher for individuals a bladder volume in close agreement (error of −3.2 ± 6.4%) with the
with a notable amount of abdominopelvic body hair (17.6 ± 10.0 dB). volume obtained using a clinical C1-6-D probe, indicating comparable
In particular, array 5 was the most probable to encounter pubic hair image quality. It also demonstrated the capability to acquire ultra-
(21.7 ± 10.8 dB). In contrast, the acoustic loss in case I for the C1-6-D sound images without gel because the acoustic loss without gel was
probe without gel was much higher at 77.3 ± 22.5 dB, whereas that for improved by approximately 60 dB compared with the C1-6-D probe.
case II was 45.2 ± 16.0 dB. The difference between the loss values in cases Without ultrasound gel, the C1-6-D probe showed a measurement
I and II may indicate that a minute quantity of gel remained on the skin error of −47.5%, whereas the cUSB-Patch had an error of −10.8%, com-
even after attempting removal. parable with the error with gel. In addition to precise bladder volume
estimation, the cUSB-Patch was easily attached and detached multiple
Conclusions times during the clinical study and could be cleaned by a wipe (Sup-
We have reported a cUSB-Patch for non-invasive, real-time and con- plementary Video 2).
tinuous monitoring of soft tissues. Compared with clinical ultra- Though the bladder was chosen as an illustrative example, the
sound probes, it shows a large FOV and is operator independent. cUSB-Patch technology is suited for identifying and characterizing
The patch consists of multiple phased arrays of transducers that disease in any deep tissue, such as the liver and kidneys, via deep tissue

Nature Electronics
Article https://doi.org/10.1038/s41928-023-01068-x

ultrasonography. In addition, the architecture of the cUSB-Patch can be electric field was measured using a linear differential transducer driven
adjusted to provide a FOV and scan depth suitable for shallower tissues by a lock-in amplifier (Stanford Research Systems, model SR830). Sup-
(such as breast tissue)45, enabling wearable ultrasound devices that plementary Note 3 discusses the method and characterization of the
can adapt to image tissue at multiple depths (Supplementary Fig. 64). complete set of material constants for Sm/La-PMN-PT.
Additional advantages of this system include reusability, high imaging
performance without gel, operator independence without rotation Fabrication and characterization of 1D array transducers
and pressing, ease of cleaning and suitability for home-care settings. The whole device fabrication was achieved in the Center for Nanoscale
Future studies will focus on several improvements. First, the simu- Systems, Harvard University, and the YellowBox, Media Lab, Massachu-
lation of mechanical deformation and acoustic performance of the setts Institute of Technology. The fabrication included four steps: (1)
adhesive layer will be investigated to meet the requirements of strong fabrication of a 1D array with 64 elements, (2) electrode deposition and
adhesion around the array–skin interface, robustness during mechani- patterning, (3) flexible cable bonding and (4) fabrication of matching
cal stretching, lower attenuation and improved wave penetration for and backing layers. The characterization of the 1D array included (1)
imaging. Second, the incorporation of 2D phased arrays would enable frequency dependence of electrical impedance and (2) pulse-echo test
superior 3D imaging reconstructions to the 2D ultrasound images for acoustic performance. Supplementary Note 5 with Supplementary
used here. Third, incorporating further post-processing steps such Figs. 14–19 describe the detailed fabrication steps and characteriza-
as coherence imaging, which uses adaptive weighting based on the tion methods.
cross-correlation between received signals to improve image contrast
would enhance image quality, especially during long-term monitor- Fabrication of cUSB-Patch
ing. Fourth, the addition of a portable data acquisition system with Supplementary Fig. 3 shows the design and fabrication of the conform-
wireless data communication and energy-efficient machine learning able substrate. First, the mould was fabricated by a 3D printer (Prusa
inference would help further translate this technology into clinical use. i3 MK3S) and sanded to prevent the texture from the 3D printing pro-
Further development of the cUSB-Patch could allow wearable ultra- cess from creating an imprint on the patch. After cleaning the mould
sound devices to be deployable across a range of health-monitoring with isopropyl alcohol and deionized water, a thin layer of silicone
applications that require accurate, real-time and autonomous tissue rubber (Ecoflex 00-30, Smooth-On) was coated onto the mould by
monitoring. dip coating to achieve a thickness of 180–200 µm (smaller than the
half-wavelength). Then, 50 g of Ecoflex (A and B) was mixed and poured
Methods into the mould (enough to fill the mould without overflowing). After
Preparation of piezoelectric ceramics curing for 4 h at room temperature, the Ecoflex substrate was carefully
Pb(1–1.5y–1.5z)SmyLaz[(Mg1/3Nb2/3)(1–x)Tix]O3 (Sm/La-PMN-PT) with x = 0.28– released from the mould. Then, the arrays were placed and embedded
0.36, y = 0.005–0.010 and z = 0.005–0.010 polycrystalline were pre- within their respective cavities and pressed to eliminate any air bub-
pared using a B-site-cation precursor method. The MgNb2O6 precursor bles that might have been trapped between the silicone and matching
materials were synthesized using Nb2O5 and MgCO3 powders at an layer. If air bubbles remain, it will diminish the overall acoustic energy
ambient temperature of 1,200 °C for 4 h. First, the powders were wet delivered and cause bright artifacts in the acquired images. Finally, the
mixed using an alcohol solvent for 24 h by a zirconium ball milling pro- additional Ecoflex was poured onto the backing layer to seal the gap
cess. Subsequently, the powder mixture was calcined at 850 °C for 2 h, between the arrays and substrate.
followed by a binder-induced alcoholic vibratory milling process for
24 h. Next, the mixture was dried at 80 °C for 10 h and uniaxially pressed Data acquisition interface for single array and multiple arrays
under 100 MPa to form fine thin pellets (thickness, 10 mm; diameter, Data acquisition was performed using the Vantage 256 ultrasound
50 mm). A compaction step was introduced, which involved cold iso- system, which provides an analogue front-end sufficient to simulta-
static pressing under 100 MPa. Following compaction, the specimens neously transmit and receive on up to 256 channels. To interface the
were baked at 550 °C for 2 h and sintered at 1,200–1,250 °C for 2 h inside fabricated transducers with the Vantage 256 system, an adaptable
sealed corundum crucibles in an oxygenated atmosphere. Finally, in connector was designed and built to be compatible with a universal
the presence of zirconia powders, the specimens were hot pressed transducer adapter (UTA). Supplementary Fig. 21 describes the overall
at 900–1,100 °C for 2 h under 30 MPa inside a silicon carbide mould. design and signal path from a transducer to the system. First, a Philips
ATL C5-2v ultrasound transducer probe was deconstructed, which
Characterization of piezoelectric ceramics can support up to 128 individual elements. After detaching from the
XRD (Bruker D8) was employed for determining the crystal struc- original printed circuit board (PCB), each individual signal line was
ture and lattice parameters. The patterns were analysed using checked by the multimeter to guarantee the connection to the cor-
TOPAS-Academic software version 5.0 (Bruker AXS). The cross-sectional responding pin in the UTA. It should be mentioned that the ordering
morphology and elemental energy-dispersive X-ray spectroscopy maps of the UTA connector is not sequential and therefore does not follow
of the specimens were measured by a field-emission SEM instrument the physical structure of a transducer. The order of the elements in
(Zeiss Gemini 500). For further electrical characterization, a silver the UTA connector was noted, and each cable was soldered onto PCB
paste was fired on both sides of the specimens at 600 °C for 10 min II accordingly. In PCB II, the final design grouped these cables into
to form the electrodes. The specimens were poled in silicone oil at eight groups of 16 lines, took a sequential ordering at its inputs and
25 °C for 10 min using a 30 kV cm−1 d.c. electric field. The piezoelectric rerouted the signals to match what the UTA connector would expect.
coefficients were determined by a combination of impedance method To allow for the transducer to be interchangeable, PCB II interfaced
and a quasi-static d33 meter. The resonance and antiresonance fre- via ribbon cables to the transducer’s complementary PCB I. Due to
quencies were measured using an HP 4194A impedance analyser. The each transducer consisting of 64 elements, two transducers could be
temperature-dependent dielectric properties were determined using connected in parallel with this interface.
an LCR meter (HP 4284A) connected to a computer-controlled cooling– For the cUSB-Patch with multiple arrays, each transducer array
heating stage. According to the IEEE Standard on Piezoelectricity (176- consists of 64 elements, and five arrays were used, making it necessary
1987), the temperature-dependent electromechanical coupling factors to develop an analogue multiplexer to expand the input/output capa-
k33, kt and kp were determined by the resonance method employing an bility of the Vantage 256 system. The multiplexer (Supplementary Fig.
impedance analyser (HP 4194A) connected to a computer-controlled 22) provides the capability to interface the Vantage 256 system with up
cooling–heating stage. The unipolar strain induced by an external to eighteen 128-channel arrays, for a total capacity of 2,304 channels

Nature Electronics
Article https://doi.org/10.1038/s41928-023-01068-x

(256 simultaneously). Each array is connected to a multiplexer card measured inside a ten-gallon water tank filled with degassed water. The
(Supplementary Fig. 23a), which contains 128 high-voltage analogue transducer was fixed to the side of the tank using double-sided tape and
single-pole/single-throw switches. The multiplexer card plugs into was driven by the Vantage 256 system. The previously described hydro-
a backplane, which wires all the corresponding analogue channels phone was mounted on a three-axis motion stage (DTS50, Thorlabs)
in parallel. In normal operation, one multiplexer card will enable all used to measure the acoustic pressure. The hydrophone’s generated
its switches and connect to the backplane analogue bus, whereas all voltage is fed through a preamplifier (Onda AG-2010 Hydrophone
the other multiplexer cards on the bus are disabled. The Vantage 256 Preamplifier), which is then read and captured using an oscilloscope
system connects to a pickoff card (Supplementary Fig. 23b), plugging (Tektronix 3 Series MDO32). Finally, a trigger signal was also output
into the backplane to access the analogue bus. The analogue signals from the Vantage 256 system into the oscilloscope to facilitate data
travel on ribbon cables from the pickoff card to an adaptor card that capture.
provides the conversion to the native Vantage 256-compatible cable. The measured results (ISPTA = 0.08 mW cm−2, ISPPA = 10.10 W cm−2 and
The multiplexer is designed to support computer control, providing MI = 0.27) below the FDA’s acoustic output limits (ISPTA = 0.72 mW cm−2,
external digital interfaces including Inter-Integrated Circuit (I2C) and ISPPA = 190.00 W cm−2 and MI = 1.90) for ultrasound diagnosis (Sup-
a high-speed serial peripheral interface (SPI) transmitted using low plementary Fig. 40). To assess tissue heating, a piece of bovine tissue
voltage differential signaling (LVDS). Each multiplexer card contains a was used to mimic human tissue and was imaged using five arrays
microcontroller and obtains a unique address according to which slot positioned on the tissue surface with converging focal regions to
it is plugged into on the backplane. For convenience, an Arduino Due achieve the maximum temperature rise at a single point. After 30 min
board was used to implement a USB-to-I2C bridge, allowing a personal of continuous operation, the temperature difference between the
computer to control the multiplexer using serial commands. It should focused point and another area in the meat was negligible (~0.3 °C)
be noted that we did not perform any electrical impedance matching (Supplementary Fig. 41), consistent with the measured very low thermal
on the array with the Verasonics system. The transducer impedance index (2.7 × 10−3) implied by the output levels (Supplementary Note 8).
was approximately 200 Ω, and the Verasonics internal impedance These experiments indicate that the cUSB-Patch is unlikely to cause any
matching was set to 250 Ω. harmful bioeffects due to acoustic power or heat generation.
The imaging acquisition system includes (1) a Vantage 256 system
with a UTA, (2) a multiplexer switch box with an array-to-system con- In vivo clinical study
nection card, (3) multiple arrays of the cUSB-Patch, (4) blue-coloured The Mass General Brigham IRB (#2021P001706) approved the pro-
electrostatic discharge wrist straps for both operator and subjects to spective pilot study of twenty subjects and served as the central IRB
protect the system electronics from static electricity, (5) a monitor for the study. Written informed consent was obtained from each sub-
to display the software and images, (6) a power cable for the switch ject before initiating the study procedures. All the procedures were
and (7) a USB cable for communication (Supplementary Fig. 24). The performed in accordance with the Health Insurance Portability and
whole system and cUSB-Patch passed the electrical inspection by the Accountability Act and the Declaration of Helsinki.
biomedical engineering team at Massachusetts General Hospital (no. Inclusion criteria were age of 18–65 years and BMI between 17 and
40200450). Arrays were connected to the Vantage 256 system using the 30 kg m−2. Subjects were excluded for confirmed or suspected preg-
custom UTA connector and were driven with a single-cycle sine pulse nancy; presence of an implanted electronic device; history of bladder
at 50 V with a 3.5 MHz centre frequency. During image acquisition, a outlet obstruction, urinary retention or benign prostatic hyperplasia;
phased-array beamformer was used to obtain a real-time image to visu- history of bladder, prostate or urethral surgery; and the presence of
ally align the transducer over the desired region on the phantom. After an indwelling urinary catheter. We recruited 20 subjects with a BMI
alignment was completed, a synthetic array transmit/receive sequence between 18.19 and 27.98 kg m−2. Subjects were requested to arrive
was used, and the raw received data were stored and beamformed in with a full bladder before the start of imaging. Imaging was performed
post-processing. using a clinical GE LOGIQ E10 system (GE Healthcare) with a curvilinear
transducer (GE C1-6-D curved array probe, 192 elements; FOV, 70°)
Acoustic imaging on different phantoms and the Vantage 256 system with the custom conformable transducer
The acoustic performance of the single array was initially evaluated (cUSB-Patch). Scan 11-08 Ultrasound Gel (Parker Laboratories) was used
using a wire phantom consisting of two parallel stable plates and eight for acoustic coupling. The two operators from Mass General Hospital
fishing lines with a diameter of 0.2 mm (around the half-wavelength) included an ultrasound radiologist with 15 years of experience and a
(Supplementary Fig. 25). All the fishing lines were submerged in deion- sonographer with 12 years of experience. Both completed hands-on
ized water in the tank with a one-inch line-to-line spacing. A three-axis training with the cUSB-Patch and demonstrated proficiency with the
stage (DTS50, Thorlabs) was used to adjust the height and position of GE LOGIQ E10 device before clinical trials.
the array. During the test, the array was submerged in deionized water The initial 11 subjects were utilized for the technical optimization
and positioned at three different locations to collect the ultrasound of the bladder patch performance in humans, including optimization of
images of fishing lines using Vantage 256. Then, the multipurpose, mul- the transmit–receive pulse sequences, procedures and data-recording
titissue ultrasound phantom (model 040GSE Computerized Imaging methods. The following four consecutive patients were analysed for
Reference Systems) was used for performance and quality assurance the purposes of proof of concept. Subjects were asked to lay supine on
testing of the phased array. For this phantom, the gel was used to ensure an examination table and adjust their clothing for ultrasound scanning
the acoustic coupling (Aquasonic 100 Ultrasound Transmission Gel, access to the lower abdomen. Each scanning sequence began without
Parker Laboratories). The US-18 fundamental ultrasound phantom the utilization of the acoustic coupling gel. First, the C1-6-D probe was
(Kyoto Kagaku) with an oval shape and a curved surface was used for used without gel to acquire sweep images in the transverse and sagittal
the cUSB-Patch imaging without the use of gel. The C5-2v convex ultra- planes. Additional sagittal and transverse still images were acquired
sound probe (Verasonics) was also used for comparison. to maximize the bladder size within the image for use in computing
the bladder volume. Next, both subject and clinician wore antistatic
Biologic safety study wrist bands (Croing) to protect the electronics from electrostatic
Before the clinical study, the spatial peak temporal average intensity discharge. The cUSB-Patch was placed on the subject’s lower abdo-
(ISPTA), the spatial peak pulse average intensity (ISPPA) and the mechanical men and positioned above the bladder using the central transducer
index (MI) were quantified to assess the biologic safety (Supplementary element to guide placement with the long axis aligned with the trans-
Note 7). The acoustic pressure generated by the entire phased array was verse plane of the body. Each array was connected to the Vantage 256

Nature Electronics
Article https://doi.org/10.1038/s41928-023-01068-x

machine by switching the multiplexer, and several seconds of video 6. Dagdeviren, C. et al. Conformal piezoelectric systems for clinical
were acquired using a focused beamformer. The synthetic array data and experimental characterization of soft tissue biomechanics.
were then acquired and saved (Supplementary Note 7). Gel was then Nat. Mater. 14, 728–736 (2015).
applied to the abdomen, and the same procedures were performed with 7. Guan, Y. et al. Soft, super-elastic, all-polymer piezoelectric
the cUSB-Patch first, followed by the C1-6-D probe. On the completion elastomer for artificial electronic skin. ACS Appl. Mater. Interfaces
of imaging, gel was removed from the skin with a wipe (Sani-Cloth AF3 15, 1736–1747 (2023).
Germicidal Disposable Wipe). Each subject was asked to partially void 8. Dagdeviren, C. et al. Conformal piezoelectric energy harvesting
and later entirely void their bladders, and the voided volumes were and storage from motions of the heart, lung, and diaphragm.
recorded. Imaging was repeated after voiding each time. As a result, Proc. Natl Acad. Sci. USA 111, 1927–1932 (2014).
data acquisitions were performed on full, partial and empty bladder 9. Nayeem, M. O. G. et al. All-nanofiber–based, ultrasensitive,
volumes for each subject. The patch was disinfected between subjects gas-permeable mechanoacoustic sensors for continuous
using another wipe. long-term heart monitoring. Proc. Natl Acad. Sci. USA 117,
Measurement and calculation of the bladder volume were per- 7063–7070 (2020).
formed using the standard clinical method of manually selecting 10. Sun, T. et al. Decoding of facial strains via conformable
high-quality images from each acquisition51,52, visually identifying the piezoelectric interfaces. Nat. Biomed. Eng. 4, 954–972 (2020).
bladder walls, and measuring each of the width, length and height val- 11. Amiri, N., Tasnim, F., Tavakkoli Anbarani, M., Dagdeviren, C. &
ues at their widest points (Fig. 5e–p). The bladder volume was estimated Karami, M. A. Experimentally verified finite element modeling
as V = W × L × H × 0.7. The same measurement and calculation methods and analysis of a conformable piezoelectric sensor. Smart Mater.
were used for the images from the C1-6-D probe and cUSB-Patch. The Struct. 30, 085017 (2021).
images from C1-6-D with gel were treated as the ground truth for all 12. Dagdeviren, C. et al. Flexible piezoelectric devices for
the volume error calculations. The mean and standard deviations of gastrointestinal motility sensing. Nat. Biomed. Eng. 1, 807–817 (2017).
the volume error were reported for several cases of interest: C1-6-D 13. Chun, S. et al. An artificial neural tactile sensing system. Nat.
probe without gel, cUSB-Patch with gel and cUSB-Patch without gel Electron. 4, 429–438 (2021).
(Supplementary Table 11). 14. Dagdeviren, C. et al. Conformable amplified lead zirconate
The acoustic loss was also measured to determine the performance titanate sensors with enhanced piezoelectric response for
of different acoustic coupling media. The measurement was performed cutaneous pressure monitoring. Nat. Commun. 5, 4496 (2014).
by comparing image brightness between the corresponding gel and 15. Wang, C. et al. Monitoring of the central blood pressure waveform
no-gel images, as well as considering any changes in the system gain. via a conformal ultrasonic device. Nat. Biomed. Eng. 2, 687–695
The acoustic loss was determined using a successive approximation (2018).
algorithm, where the gel image brightness was iteratively scaled and 16. Wang, F. et al. Flexible Doppler ultrasound device for the
thresholded according to the image dynamic range to match the mean monitoring of blood flow velocity. Sci. Adv. 7, eabi9283 (2021).
values of the images (in this case, ‘mean’ refers to the mean value of the 17. Kim, T. et al. Flexible 1–3 composite ultrasound transducers with
non-thresholded pixels, that is, pixels that are not black). An iterative silver nanowire-based stretchable electrodes. IEEE Trans. Ind.
approach was used because of the non-Gaussian statistical distribu- Electron. 67, 6955–6962 (2019).
tion of pixel brightness and the nonlinear behaviour of thresholding. 18. Pashaei, V. et al. Flexible body-conformal ultrasound patches for
image-guided neuromodulation. IEEE Trans. Biomed. Circuits Syst.
Reporting summary 14, 305–318 (2020).
Further information on research design is available in the Nature Port- 19. Sadeghpour, S., Lips, B., Kraft, M. & Puers, R. Bendable
folio Reporting Summary linked to this article. piezoelectric micromachined ultrasound transducer (PMUT)
arrays based on silicon-on-insulator (SOI) technology. J.
Data availability Microelectromech. Syst. 29, 378–386 (2020).
The raw data from human trials are available from the correspond- 20. Wang, C. et al. Continuous monitoring of deep-tissue
ing author, subject to approval from The Mass General Brigham IRB haemodynamics with stretchable ultrasonic phased arrays. Nat.
(#2021P001706). All other data that support the findings of this study Biomed. Eng. 5, 749–758 (2021).
are available from the corresponding author upon reasonable request. 21. Liu, W., Chen, W., Zhu, C. & Wu, D. Design and micromachining of
a stretchable two-dimensional ultrasonic array. Micro Nano Eng.
Code availability 13, 100096 (2021).
The codes based on Python used for the synthetic array beamforming, 22. Zhang, Q. et al. Transformable ultrasonic array transducer for
simulation and image analysis are available from the corresponding multiscale imaging and beamforming. IEEE Trans. Ind. Electron.
author on reasonable request. 69, 3078–3087 (2022).
23. Wang, C. et al. Bioadhesive ultrasound for long-term continuous
References imaging of diverse organs. Science 377, 517–523 (2022).
1. Dagdeviren, C. et al. Recent progress in flexible and stretchable 24. Yu, C.-C. et al. A conformable ultrasound patch for
piezoelectric devices for mechanical energy harvesting, sensing cavitation-enhanced transdermal cosmeceutical delivery. Adv.
and actuation. Extrem. Mech. Lett. 9, 269–281 (2016). Mater. 35, 2300066 (2023).
2. Wang, C., Wang, C., Huang, Z. & Xu, S. Materials and structures 25. de Oliveira, T. F., Pai, C. N., Matuda, M. Y., Adamowski, J. C. &
toward soft electronics. Adv. Mater. 30, 1801368 (2018). Buiochi, F. Development of a 2.25 MHz flexible array ultrasonic
3. Fernandez, S. V. et al. On-body piezoelectric energy harvesters transducer. Res. Biomed. Eng. 35, 27–37 (2019).
through innovative designs and conformable structures. ACS 26. Du, W. et al. Conformable ultrasound breast patch for deep tissue
Biomater. Sci. Eng. 9, 2070–2086 (2021). scanning and imaging. Sci. Adv. 9, eadh5325 (2023).
4. Dagdeviren, C., Li, Z. & Wang, Z. L. Energy harvesting from the 27. Hu, H. et al. A wearable cardiac ultrasound imager. Nature 613,
animal/human body for self-powered electronics. Annu. Rev. 667–675 (2023).
Biomed. Eng. 19, 85–108 (2017). 28. Hu, H. et al. Stretchable ultrasonic transducer arrays for
5. Fernandez, S. V. et al. Ubiquitous conformable systems for three-dimensional imaging on complex surfaces. Sci. Adv. 4,
imperceptible computing. Foresight 24, 75–98 (2022). eaar3979 (2018).

Nature Electronics
Article https://doi.org/10.1038/s41928-023-01068-x

29. Yang, Y. et al. A flexible piezoelectric micromachined ultrasound 49. Kim, N., Jang, S. J. & Shrout, T. R. Relaxor based fine grain
transducer. RSC Adv. 3, 24900–24905 (2013). piezoelectric materials. In Proc. 1990 IEEE 7th International
30. Zhu, Q. et al. A piezoelectric micro-machined ultrasonic Symposium on Applications of Ferroelectrics 605–609 (IEEE,
transducer array based on flexible substrate. In Proc. 13th Annual 1990).
IEEE International Conference on Nano/Micro Engineered and 50. Jiang, X.-J. et al. A microscale linear phased-array ultrasonic
Molecular Systems 345–348 (IEEE, 2018). transducer based on PZT ceramics. Sensors 19, 1244 (2019).
31. Wang, Z. et al. A flexible ultrasound transducer array with 51. Bala, K. G. & Chou, Y.-H. Ultrasonography of the urinary bladder.
micro-machined bulk PZT. Sensors 15, 2538–2547 (2015). J. Med. Ultrasound 18, 105–114 (2010).
32. Zhou, Q., Lam, K. H., Zheng, H., Qiu, W. & Shung, K. K. 52. Bih, L.-I., Ho, C.-C., Tsai, S.-J., Lai, Y.-C. & Chow, W. Bladder shape
Piezoelectric single crystal ultrasonic transducers for biomedical impact on the accuracy of ultrasonic estimation of bladder
applications. Prog. Mater. Sci. 66, 87–111 (2014). volume. Arch. Phys. Med. Rehab. 79, 1553–1556 (1998).
33. Li, C. et al. Atomic-scale origin of ultrahigh piezoelectricity in
samarium-doped PMN-PT ceramics. Phys. Rev. B 101, 140102 Acknowledgements
(2020). C.D. thanks C. Dagdeviren and H. E. Dagdeviren for fruitful
34. Li, F. et al. Ultrahigh piezoelectricity in ferroelectric ceramics by discussions throughout this project. This work was supported by the
design. Nat. Mater. 17, 349–354 (2018). National Science Foundation CAREER: Conformable Piezoelectrics
35. Zhou, S., Lin, D., Su, Y., Zhang, L. & Liu, W. Enhanced dielectric, for Soft Tissue Imaging (grant no. 2044688), 3M Non-Tenured
ferroelectric, and optical properties in rare earth elements doped Faculty Award, Sagol Weizmann-MIT Bridge Program, Texas
PMN-PT thin films. J. Adv. Ceram. 10, 98–107 (2021). Instruments Inc. and MIT Media Lab Consortium funding. W.D. was
36. Zhang, Z. et al. New Sm-PMN-PT ceramic-based 2-D array for supported by the National Science Foundation Graduate Research
low-intensity ultrasound therapy application. IEEE Trans. Ultrason. Fellowship Program (grant no. 2141064). Any opinions, findings
Ferroelectr. Freq. Control 67, 2085–2094 (2020). and conclusions or recommendations expressed in this material
37. Zhang, Z. et al. High-performance ultrasound needle transducer are those of the author(s) and do not necessarily reflect the views
based on modified PMN-PT ceramic with ultrahigh clamped of the National Science Foundation. T.T.P. was supported by the
dielectric permittivity. IEEE Trans. Ultrason. Ferroelectr. Freq. 2020 ARRS Scholar Award. This work was performed in part at the
Control 65, 223–230 (2018). Harvard University Center for Nanoscale Systems (CNS)—a member
38. Kristiansen, N. K., Djurhuus, J. C. & Nygaard, H. Design and of the National Nanotechnology Coordinated Infrastructure Network
evaluation of an ultrasound-based bladder volume monitor. Med. (NNCI), which is supported by the National Science Foundation
Biol. Eng. Comput. 42, 762–769 (2004). under NSF award no. ECCS-2025158. We thank P. Cao and C. Peng
39. Tanaka, R. & Abe, T. Measurement of the bladder volume for technical discussions; H. Park and J. Tresback for assistance
with a limited number of ultrasonic transducers. In 2010 IEEE on microfabrication; O. Drori, A. Mamistvalov, K. Brahma and A.
International Ultrasonics Symposium 1783–1786 (IEEE, 2010). Benjamin for imaging discussion; E. Suh for patch design discussion;
40. van Leuteren, P. G. et al. URIKA, continuous ultrasound monitoring and M.-C. O’Connell, M. Martin, E. Sands and Y. Gu for assisting on
for the detection of a full bladder in children with dysfunctional the clinical study.
voiding: a feasibility study. Biomed. Phys. Eng. Express 3, 017005
(2017). Author contributions
41. Kuru, K. et al. Intelligent autonomous treatment of bedwetting C.D. conceived the research idea, designed the research direction
using non-invasive wearable advanced mechatronics systems and directed all the research activities. A.E.S. directed the proposed
and MEMS sensors. Med. Biol. Eng. Comput. 58, 943–965 implementation of the technology and supervised the clinical study.
(2020). A.P.C. contributed to the overall system architecture. L.Z. and C.D.
42. Kuru, K., Ansell, D., Jones, M., De Goede, C. & Leather, P. conceived the research methodology and aims. L.Z., C.M. and C.D.
Feasibility study of intelligent autonomous determination of designed the experiments. D.L. processed and characterized the
the bladder voiding need to treat bedwetting using ultrasound piezoelectric ceramics. W.L. supervised the materials investigation.
and smartphone ML techniques. Med. Biol. Eng. Comput. 57, D.L. and L.J. studied the structure and morphology of ceramics.
1079–1097 (2019). D.L., L.Z. and F.L. analysed the performance of the ceramics. C.D.,
43. Fournelle, M. et al. Portable ultrasound research system for use L.Z. and D.M. designed the patch for the bladder. L.Z. designed
in automated bladder monitoring with machine-learning-based and fabricated the ultrasound arrays. D.M. and L.Z. characterized
segmentation. Sensors 21, 6481 (2021). the array and fabricated the patch. C.M. and D.M. designed the
44. Jo, H. G. et al. Forward-looking ultrasound wearable scanner electronic circuits for data acquisition. S.J.S., assisted by V.K.,
system for estimation of urinary bladder volume. Sensors 21, conducted the coding and ultrasound signal processing for image
5445 (2021). generation by the Verasonics Vantage 256 system. L.Z., C.M., S.J.S.
45. Dagdeviren, C. & Zhang, L. Methods and apparatus for imaging and D.M. conducted the in vitro ultrasound imaging. I.I.S., W.D. and
with conformable ultrasound patch. US patent 16/658,237 S.V.F. assisted with the device characterization. V.K. coordinated the
(2020). in vitro study at Mass General Hospital, and T.T.P. coordinated the
46. Yang, J. C. et al. Geometrically engineered rigid island array in vivo study, including overseeing recruitment and scheduling (with
for stretchable electronics capable of withstanding various H.E.), obtaining subject informed consent and monitoring subject
deformation modes. Sci. Adv. 8, eabn3863 (2023). safety. H.E. recruited the subjects and assisted in the clinical study.
47. Fielding, J. T., Shrout, T. R. & Jang, S. J. Increased operating L.Z., D.M., S.J.S. and V.K. conducted the safety testing and prepared
temperature range in La-modified Pb(Mg1/3Nb2/3)O3-PbTiO3 the related IRB document filings. D.H., T.T.P., L.Z. and C.M. designed
relaxor ferroelectric-based transducers. In Proc. 1994 IEEE the clinical-trial imaging protocols. D.H. and Q.L. performed the
International Symposium on Applications of Ferroelectrics image acquisition of the human subject. C.M., S.J.S., T.T.P., H.E., V.K.,
158–161 (IEEE, 1994). D.H. and Q.L. executed the in vivo study and analysed the data. Y.C.E.
48. Choi, S. W., Shrout, T. R., Jang, S. J. & Bhalla, A. S. Morphotropic assisted in the imaging process. W.D. and D.S. composed the layout
phase boundary in Pb (Mg13Nb23) O3-PbTiO3 system. Mater. Lett. 8, of Supplementary Videos 1–8 and formed the videos. All authors
253–255 (1989). contributed to the manuscript writing.

Nature Electronics
Article https://doi.org/10.1038/s41928-023-01068-x

Competing interests Reprints and permissions information is available at

A.P.C. is on the Board of Analog Devices. The other authors declare no www.nature.com/reprints.
competing interests.
Publisher’s note Springer Nature remains neutral with regard to
Additional information jurisdictional claims in published maps and institutional affiliations.
Supplementary information The online version contains supplementary
material available at https://doi.org/10.1038/s41928-023-01068-x. Springer Nature or its licensor (e.g. a society or other partner) holds
exclusive rights to this article under a publishing agreement with
Correspondence and requests for materials should be addressed to the author(s) or other rightsholder(s); author self-archiving of the
Canan Dagdeviren. accepted manuscript version of this article is solely governed by the
terms of such publishing agreement and applicable law.
Peer review information Nature Electronics thanks Yangzhi Zhu, Qifa
Zhou and the other, anonymous, reviewer(s) for their contribution to © The Author(s), under exclusive licence to Springer Nature Limited
the peer review of this work. 2023, corrected publication 2023

Nature Electronics
 nature portfolio | reporting summary
Corresponding author(s): Canan Dagdeviren
Last updated by author(s): Sep 1, 2023

Reporting Summary
Nature Portfolio wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency
in reporting. For further information on Nature Portfolio policies, see our Editorial Policies and the Editorial Policy Checklist.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.
n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement
A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly
The statistical test(s) used AND whether they are one- or two-sided
Only common tests should be described solely by name; describe more complex techniques in the Methods section.

A description of all covariates tested

A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons
A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient)
AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)

For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted
Give P values as exact values whenever suitable.

For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings
For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes
Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated
Our web collection on statistics for biologists contains articles on many of the points above.

Software and code

Policy information about availability of computer code
Data collection Data collection of the ultrasound signal was carried out with the Verasonics Vantage platform.

Data analysis All data collected during in vitro and in vivo analysis was analyzed and plotted in OriginPro 2018 64-bit software.
The KLM modeling was conducted in the PiezoCAD software (Pro. 4.0.1).
The custom codes based on Python 3.8 were used for SAR beamforming, simulation, and image analysis.
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and
reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Portfolio guidelines for submitting code & software for further information.

Data
March 2021

Policy information about availability of data

All manuscripts must include a data availability statement. This statement should provide the following information, where applicable:
- Accession codes, unique identifiers, or web links for publicly available datasets
- A description of any restrictions on data availability
- For clinical datasets or third party data, please ensure that the statement adheres to our policy

The data supporting the results in this study are available within the paper and its Supplementary Information. The raw data from human trials are available from
the corresponding author, subject to approval from The Mass General Brigham Institutional Review Board (IRB) (#2021P001706).

1
 nature portfolio | reporting summary
Human research participants
Policy information about studies involving human research participants and Sex and Gender in Research.

Reporting on sex and gender Subjects were recruited to ensure adequate representation of morphologic features unique to each gender. Individuals will
not be excluded from the proposed research study on the basis of gender.

Population characteristics We recruited 20 subjects with the age between 18-65 years and BMI between 18.19-27.98 kg/m2. The data of four subjects
were studied in this manuscript.
Healthy Subject A, Female, age 24, BMI: 22.67 kg/m2, no relevant health conditions.
Healthy Subject B, Female, age 32, BMI: 18.19 kg/m2, no relevant health conditions.
Healthy Subject C, Male, age 45, BMI: 24.78 kg/m2, no relevant health conditions.
Healthy Subject D, Male, age 25, BMI: 27.98 kg/m2, no relevant health conditions.

Recruitment We used Rally (https://rally.partner.org/) to recruit subjects. Healthy subjects were recruited to represent a variety of ages,
genders and cultural backgrounds. There is no self-selection or other bias.

Ethics oversight All procedures in the in vivo trials were performed in accordance with the experimental protocol approved by the Mass
General Brigham Institutional Review Board (IRB) (#2021P001706). The participants gave informed consent.

Note that full information on the approval of the study protocol must also be provided in the manuscript.

Field-specific reporting
Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.

Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf

Life sciences study design

All studies must disclose on these points even when the disclosure is negative.
Sample size 5 arrays with closed performance were selected from 10 arrays to make the cUSB-Patch. One patch with 5 arrays was used to perform in vivo
tests on four subjects.

Data exclusions No data were excluded.

Replication All attempts at replication were successful. For each subject and each case, the patch imaged bladder for several minutes using all five arrays
to achieve multiple images. The same patch was used for four subjects in different, scheduled times and dates.

Randomization One type of device was tested following same protocol for in vivo trial, but we tried to collect wide variety of subjects.

Blinding Data collection and analysis for each subject were performed by different researchers.

Reporting for specific materials, systems and methods

We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material,
system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response.

Materials & experimental systems Methods

n/a Involved in the study n/a Involved in the study
Antibodies ChIP-seq
March 2021

Eukaryotic cell lines Flow cytometry

Palaeontology and archaeology MRI-based neuroimaging
Animals and other organisms
Clinical data
Dual use research of concern