Automating Compression Ultrasonography of Human Thigh Tissue and

Vessels via Strain Estimation

Rytis Jurkonis

1 a

, Rimvydas Eitminavi

cius

1 b

, Vaidotas Marozas

1 c

and Andrius Sakalauskas

2 d

Biomedical Engineering Institute, Kaunas University of Technology, K. Bar

sausko 59, Kaunas, Lithuania

TELEMED, Ultrasound Medical Systems, Savanoriu 178A, Vilnius, Lithuania

Keywords:

Deep Vein Thrombosis, Compression Quantiﬁcation, Operator Free, Tissue Displacement.

Abstract:

Despite the progress made in ultrasonic imaging, the current examination of vein structures by compression

is highly operator-dependent and is a time-consuming clinical routine. Current guidelines for the manage-

ment of deep vein thrombosis recommend compression ultrasonography follow-up for patients at risk of life-

threatening complications (pulmonary embolism, heart attack, or stroke). New methods are needed to allow

operator-free monitoring of vein structure at the point of care. This article presents the results of integrated

imaging with a tissue compression actuator and automated control of tissue deformation through strain es-

timation. The data for feedback control of the actuator is calculated from raw ultrasound radio-frequency

backscattered signals. The region-averaged strain curve (strain versus time) obtained during the tissue com-

pression cycle serves as input for the actuator. The mounting on the human thigh is made from rigid, pre-

shaped shells, which are adjusted to the circumference of the thigh with straps. The actuator facilitates a

novel, on-body-mounted, automated, operator-free examination of the human femoral vein.

1 INTRODUCTION

Compression ultrasonography is a method for vein

structure analysis, including testing of suspected deep

vein thrombosis (DVT). DVT is the formation of a

blood clot within the deep veins that blocks blood

ﬂow. In 50% of people with DVT, the clot even-

tually breaks off and travels to the central circu-

lation, causing life-threatening complications (pul-

monary embolism, heart attack, or stroke). Early di-

agnosis of DVT is crucial, and despite the progress

made in ultrasonic imaging, the current examination

of vein structures by compression is highly operator-

dependent and a time-consuming clinical routine.

In reviewing the history of venous ultrasound,

Cronan et al. (Cronan, 2003) mention that an arti-

cle by Talbot et al. (Talbot et al., 1982) highlights a

signiﬁcant breakthrough in venous clot detection that

occurred in 1982. Talbot noted that there were easily

recognizable differences between patent normal veins

and those containing clots. Using a combination of

https://orcid.org/0000-0002-9481-1773

https://orcid.org/0009-0009-4682-8878

https://orcid.org/0000-0002-6879-5845

https://orcid.org/0000-0002-3978-7301

B-mode real-time imaging and pulsed Doppler imag-

ing, the author indicated that (1) ﬂows in a normal

vein varied with respiration; (2) a Valsalva maneuver

augmented the size of a normal vein; (3) light pres-

sure exerted on a normal vein caused it to collapse;

and (4) Doppler signals were found in a patent vein.

Obstructed veins tend to be larger. Often, it contained

a speckled mass within it, and the diameter did not

change with respiration, a Valsalva maneuver, or light

pressure exerted on the overlying skin. The next par-

ticular paper (Raghavendra et al., 1986) proposed a

method called compression ultrasonography. Surpris-

ingly, quantitative characterization of compression or

criteria was described only in recent articles and only

in cases of residual vein thrombosis. With sonogra-

phy, vein diameters were evaluated during compres-

sion, and residual vein thrombosis (RVT) was deﬁned

as the persistence of thrombotic material resulting in

a diameter of 4 mm or more (Prandoni et al., 2002).

Criteria were deﬁned as ultrasound incompressibil-

ity of at least 4 mm in the common femoral and/or

popliteal vein after 3 months of RVT. This RVT cri-

terion was shown to be reproducible in other institu-

tions (Tan et al., 2012; Palareti et al., 2014; Liu et al.,

2023).

Jurkonis, R., Eitminavi

cius, R., Marozas, V. and Sakalauskas, A.

Automating Compression Ultrasonography of Human Thigh Tissue and Vessels via Strain Estimation.

DOI: 10.5220/0013264300003911

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 18th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2025) - Volume 1, pages 239-245

ISBN: 978-989-758-731-3; ISSN: 2184-4305

239

To the best of the author’s knowledge, no more

quantitative criteria about the limits of compression

are published. Therefore, there is a need for new ob-

jective criteria, especially from sonography images.

The study explores the feasibility of automated com-

pression ultrasonography of thigh tissues by propos-

ing a hardware actuator and compression control solu-

tion. Results of the investigation (intra- and intersub-

ject variability) by using the prototype are disclosed.

2 MATERIAL AND METHODS

2.1 Thigh Mount Supports

The prototype of the thigh mount was made from two

rigid C-shaped frames (Fig. 1). When designing these

C-shaped frames the anthropometric data (Heitmann

and Frederiksen, 2009) of human thigh size (median

circumference 56 cm) was considered. In one frame

the locking slot for ﬁxing the imaging transducer was

formed according to the transducer enclosing dimen-

sions. The ﬁrst frame to the second was ﬁxed with two

nonelastic straps that were adjusted to the individual

circumference of the thigh of the subject. After ﬁxa-

tion, the length of the straps was stable during the time

required for the whole session of data recording. The

second C-shaped frame has two ﬁxed bladders (Fig. 1

(b)). The ﬁxation was made with woven fabric, for

prototyping, while future support designs (including

the choice of fabric materials) will be improved for

better comfort of the user.

Figure 1: The tested design for mounting of compression

actuator and imaging transducer on human thigh: a) mount-

ing on the thigh; b) rigid frames with a mounted actuator

pressing tissues against the imaging transducer. The ap-

proximate location of the femoral artery and the compressed

femoral vein is indicated by “A” and “V”, respectively.

2.2 Pneumatic Actuator

Actuation on tissues was performed using bladders

that were in contact with the skin on the thigh. The air

bladders were rectangular with a size of 60 x 100 mm.

The required air volume to pressurize the bladders

to 100 mmHg was 200 mL (per bladder). Bladders

were pressurized by a silent piezoelectric air pump,

model UXPB5400000A (Lee Company, Westbrook,

CT, USA). This pump in combination with a smart

pump module is capable of producing pressures of up

to 190 mmHg, and a free ﬂow output of 1.35 L/min.

The pump and electronics were powered by a Li-Ion

battery.

2.3 Ultrasonic Imaging

Thigh tissue imaging was performed using the ArtUs-

1H beamformer (Telemed, Lithuania) that has raw

beamformed radio frequency (RF) data output. The

beamformer was equipped with a low-proﬁle linear

array (LF11-5H60-A3). The beamformer was inter-

faced to the computer via USB 3.0 and had syn-

chronization output connectors installed. Sync pulses

from the beamformer served as the activation signal

for the pneumatic actuator. The main parameters of

ultrasonic scanning were as follows: scanning depth

of 5 cm, ultrasound wave frequency in the range of 5

- 11 MHz, and transmission of a single focal depth of

5 cm. B-mode images (512 x 512) were acquired in

real-time into the host computer. The RF signal was

digitized with a 40 MHz sampling frequency and the

analog-to-digital converter resolution of 16 bits was

analyzed in real-time as described in the 2.3.2 section.

2.3.1 Software

Two real-time streams—B-mode ultrasound images

(for navigation) and RF data (for strain assess-

ment)—were transmitted to the MATLAB environ-

ment and MATLAB GUI application using a software

development kit (SDK) developed by TELEMED.

The functionality of the MATLAB GUI application

was sufﬁcient for research purposes and had an op-

tion to: adjust imaging depth, choose the frequency

of waves, adjust emission power and receiving gain,

preset the threshold of strain, and initiate the imag-

ing session. The DLL can be used to link C++

TELEMED SDK to any programming platform.

2.3.2 Estimation of Real-Time Tissue Axial

Displacements and Strain

Tissue displacements and strain parameters (calcu-

lated from RF data) are well-known in the commu-

BIOIMAGING 2025 - 12th International Conference on Bioimaging

240

nity of ultrasound elastography (Garra, 2015). Typ-

ically, these parameters serve for the assessment of

tissue hardening, because the parameters are highly

correlated with the stiffness of tissue. In this study,

tissue strain parameter was employed for the quantita-

tive assessment of compression ultrasonography with

implications to be used in the diagnosis of throm-

botic veins. Axial displacements were estimated by

using a window-based correlation technique proposed

by Zahiri and Salcudean (Zahiri Azar and Salcudean,

2006). The algorithm is one of the fastest proposed

up to date because it uses displacement value esti-

mated in previous, neighboring, correlation window

as a prior, to reduce the search for the next one to a

very small area of 3 samples (-1, 0, +1). Displace-

ment estimates are reﬁned by using parabolic interpo-

lation which is again very fast and efﬁcient. The time

delay was calculated by using the following parabolic

interpolation formula:

∆ = T

− y

2(y

− 2y

+ y

)

(1)

where T

– ultrasound RF data sampling period,

– the argument for maximal correlation (time delay

for the correlation peak), and y

, y

– nearest neigh-

bors of the peak.

Strain estimates were obtained from displace-

ments using a gradient operator combined with a

smoothing ﬁlter. The size of the kernel for axial dis-

placement evaluation was 107 (samples) × 7 (scan

lines). The median value of the strain matrix is cal-

culated from every frame of the RF data. Then the

strain curve is formed from the calculated median val-

ues. Real-time strain assessment served in a feedback

loop with a tissue compression actuator. The pneu-

matic actuator continuously applied the pressure onto

the thigh gradually closing the vein under investiga-

tion at the same time strain was estimated for each

received data frame of backscattered signals. The

operator sets the strain threshold value in the MAT-

LAB GUI application, which deﬁnes the end of the

ultrasonic imaging session. At the instance when the

imaging is stopped the beamformer provides a syn-

chronization pulse to the actuator to halt the pump and

initiate the release of the applied pressure.

2.4 Data Collection and Analysis

The subjects were instructed to relax their body mus-

cles, similar to the procedure during blood pressure

measurement with a brachial cuff. In addition, the

positioning of the body and the leg was standardized:

the subjects were placed on the testbed (Matsubara

et al., 2022) with their back elevated at a 30-degree

angle and the leg freely extended. Thigh tissue com-

pressions were performed in two different variants:

1) manually, by maneuvering the imaging transducer,

and 2) by using an automated tissue compression ac-

tuator. The automated compression was conducted

with a thigh-mounted imaging transducer and pres-

surized bladders, as depicted in Figure 1. Manual

compression was applied from the front side of the

thigh, in the same way as in routine clinical practice.

The dosing of compressions in both manual and auto-

mated variants was quantiﬁed using the preset strain

thresholds in MATLAB GUI. Manual compression

was stopped at 2, 5, 8, and 11% strain thresholds,

while the automated compression was halted at 1.5,

2.5, 3.5, 4.5, and 5.5%. For each threshold, three

compressions were recorded. These levels were cho-

sen empirically based on a current design. Image

sequences of tissue reactions were saved for off-line

analysis, which is described in the following section.

The initial and ﬁnal B-mode images from the sin-

gle subject thigh compression session are provided in

Fig. 2. In the initial image, the largest ellipse-shape

darker pattern represents the femoral vein in cross-

section. This pattern decreases in size as the thigh

tissues are compressed, either manually or by the acti-

vated actuator. The artery lumen appears as a smaller,

more circular pattern. The artery lumen begins to de-

form under a stronger compression, once the lumen of

the healthy vein is fully closed. The depicted B-mode

images are primarily used to initiate ofﬂine analy-

sis (intermediate images are provided in the appendix

Figure 9). M-mode imaging facilitated the analysis of

deformation over time. To generate the M-mode im-

age, the intensity of the B-mode image was extracted

along the blue line, which is centered on the main vein

width in the initial image (left panel in Fig. 2). The

same line location is indicated on the B-mode image

after maximal compression.













    













    



Figure 2: Cross-section B-mode images of femoral vessels

before compression (left) and at full lumen closure of the

veins (right).

The overview of tissue compression progress in

a sequence of B-mode images can be enhanced by

preparing M-mode images with a time dimension

Automating Compression Ultrasonography of Human Thigh Tissue and Vessels via Strain Estimation

241

Figure 3: Human thigh imaging data: a) representation of

vein lumen deforming in M-mode time diagram as prepared

from image intensity proﬁles indicated on B-mode images;

b) extracted position changes of vein walls (surfaces), and

c) calculated vein lumen changes during compression and

release of thigh tissues.

(Fig. 3 (a)). The transitions of tissue structures’ po-

sitions appear more continuous and smooth, making

tracking easier. Displacements of the upper and lower

walls of the vein are extracted from the M-mode im-

age, as exempliﬁed by the lines on the grayscale im-

age. These traces (Fig. 3 (b)) assist in calculating the

absolute size of the vein lumen. Figure 3 (c) shows the

initial size of the vein lumen, approximately 11 mm,

which was compressed to around 1 mm, resulting in

a relative decrease of the vein lumen by about 90%.

This is an example of the resulting reaction of tissues

and the femoral vein after manual compression on the

thigh. The subjective quantiﬁcation of the degree of

compression could be enriched with strain parameter.

The lumen decrease trace is exempliﬁed together in

strain–time dependency (Fig. 4).

The corresponding strain, as the quantity of the

whole image, averaged degree of deformation, in-

creases to more than 9% in the ﬁnal stage of the com-

pression. The strain increases while the lumen de-

creases, both peaking around 15 seconds in this exam-

ple. The strain appears more smooth during compres-

sion and the release of the thigh. Strain value mon-

itoring was done in real-time with ultrasonic imag-

ing, allowing the degree of deformation to be dosed

by the set threshold of strain. Real-time thresholding

0 5 10 15 20 25 30 35

Time (s)

-2

Strain in tissue (%)

-20

100

Lumen decrease (%)

Strain

Lumen decrease

Figure 4: Comparison of lumen relative decrease with strain

calculated from ultrasonic imaging RF data.

in strain trace was veriﬁed with vein lumen decrease

obtained from the M-mode image. An example of the

compression dosed according to the strain threshold

is provided in Figure 5. In this example, the preset

threshold for tissue strain was 8% which resulted in a

vein lumen change from the initial size of 11.2 mm to

7.3 mm (relative decrease of 35%).

0 100 200

B-scan intensity

Depth (mm)

11.2 mm

0 100 200

B-scan intensity

7.3 mm

0 5

Time (s)

M-scan

Figure 5: Analysis of the tissue compression image se-

quence: M-mode diagram is prepared for overview for over-

all progress; the B-mode image intensities proﬁle exempli-

ﬁes initial (left panel) and ﬁnal (right panel) dimensions of

vein lumen size.

This approach of estimation of only initial and ﬁ-

nal (strain threshold triggered) vein lumen sizes was

applied in the analysis of data from a group of healthy

subjects. In total three cases of thigh tissue reac-

tions to compression were recorded and analyzed to

estimate inter-subject variability. The possibilities of

monitoring tissue reactions in single subjects, or intra-

subject variability, were also evaluated.

BIOIMAGING 2025 - 12th International Conference on Bioimaging

242

3 RESULTS

In total 9 healthy volunteers were included in this

study, all participating in the intersubject investiga-

tion and one of them in the long-term (2.5 hours)

intrasubject investigation. Vein lumen responses to

manual and automated compressions were analyzed

ofﬂine in M-mode diagrams. Both inter-subject and

intra-subject variabilities were examined. Figure 6

illustrates inter-subject variability in vein lumen re-

actions to manual and automated thigh compressions

(intermediate images are provided in the appendix

Figure 10 and Figure 11). Each point represents the

mean result of three compression sessions, with lines

connecting points to show trends for each subject.

0 2 4 6 8 10 12

Strain threshold (%)

-100

-80

-60

-40

-20

Lumen decrease (%)

Manual

Automatic

Figure 6: Reactions of vein lumen in healthy subjects with

dosing the compression amount according to strain thresh-

old: manual (by-hand) maneuvering and automatic com-

pression.

Observed vein lumen reactions differ signiﬁcantly.

With the preset value of threshold of 11% for strain,

vein lumen reduction can range from 30% to 70%.

Meaning, that the same values of threshold can not be

applied to a whole group. In the manual variant, the

imaging transducer applies compression, while in the

automatic variant, tissues are pushed toward it, caus-

ing the differences.

Figure 7 compares the results using boxplots for

different strain thresholds in manual and automated

compression ultrasonography procedures. A strong

relationship between strain threshold and percentage

of lumen decrease can be observed. However, the in-

terquartile range of lumen decrease increases signiﬁ-

cantly at higher strain thresholds.

An automatic tissue compression actuator will be

used for prolonged intermittent monitoring of venous

conditions. Therefore, evaluating intra-subject vari-

ability in measurements is important. Figure 8 (a)

presents two series of automatic measurements ob-

tained using two different strain threshold settings

over 150 minutes of monitoring. Figure 8 (b) com-

pares the interquartile ranges of lumen decrease for

these two strain threshold settings. A higher strain

threshold setting results in signiﬁcantly greater intra-

subject variability in the estimated decrease in vein

lumen.

0 2 4 6 8 10 12

Strain threshold (%)

-100

-80

-60

-40

-20

Lumen decrease (%)

Manual

Automatic

Figure 7: Box-plot of vein lumen decrements in healthy

subjects during manual (by-hand) and automatic compres-

sion.

The boxplot in Figure 8 (b) provides a statistical

representation of the data in Figure 8 (a). The notches

do not overlap, which suggests a statistically signif-

icant difference in medians between the two groups.

The left boxplot (2.5% threshold) has a much tighter

interquartile range, indicating lower variability when

compared to the right boxplot (4.5% threshold).

0 50 100 150

Time (min)

-100

-80

-60

-40

-20

Lumen decrease (%)

Threshold 4.5%

Threshold 2.5%

2.5 4.5

Strain threshold (%)

Figure 8: Results of vein lumen response using a thigh-

mounted imaging transducer, with automated compression

doses set to strain thresholds of 2.5% and 4.5%. A single

actuator application produced 31 repeated automated com-

pressions at each threshold over a 2.5-hour period.

Automating Compression Ultrasonography of Human Thigh Tissue and Vessels via Strain Estimation

243

4 DISCUSSION

The tissue response to compression is subject-

speciﬁc. Deformation and accordingly calculated

strain depend on the site of actuation. If the actua-

tor is mounted on the side with varying layers of adi-

pose superﬁcial tissue, or varying stiffness of muscle

because of involuntary movements of the leg the re-

actions to compression will be different. So the intra-

subject peculiarities of tissues are proposed to be con-

sidered with the initial adjustment of the strain thresh-

old. The automatic compressions are performed from

the back side of the thigh and result in smaller mod-

ulation of strain. Despite low modulation of strain,

the actuation on the thigh from the opposite side and

the use of rigid pre-shaped frames to mount the imag-

ing transducer and actuating bladders contribute to re-

ducing uncertainty in tissue response (reducing shear

deformation).

The whole B-mode image was used for the strain

estimate calculation and not justiﬁed to regions of

thigh tissues. Adjusting region for strain estimation

could potentially increase strain modulation and the

accuracy in quantifying compression amount. Pro-

vided experimental results are from a limited num-

ber of healthy subjects. The personal body charac-

teristics were not examined in detail, thus further re-

search is needed including a bigger group of subjects

with higher variability in body composition. Addi-

tionally, individualized calibration should be devel-

oped to perform long-term accurate monitoring of tis-

sue response.

5 CONCLUSIONS

In this study, an ultrasonography tissue compression

actuator for automated long-term monitoring of ve-

nous vessels in the lower extremities is proposed. The

actuator is controlled by a tissue strain parameter that

regulates compression. The results show a negative

correlation between venous vessel lumen closure and

the tissue strain parameter, suggesting that using tis-

sue strain as a feedback mechanism is feasible. How-

ever, the variability at a higher strain threshold is rel-

atively big, leaving room for improvement. Inter-

subject variability could be addressed through initial

calibration, while intra-subject variability for moni-

toring applications could be managed by deﬁning a

more targeted region of interest.

ACKNOWLEDGEMENTS

This work is funded under the Horizon Europe In-

novation Action ThrombUS+ (Grant Agreement No.

101137227), co-funded by the European Union.

REFERENCES

Cronan, J. J. (2003). History of venous ultrasound. J Ultra-

sound Med, 22(11):1143–1146.

Garra, B. S. (2015). Elastography: history, principles, and

technique comparison. Abdom Imaging, 40(4):680–

697.

Heitmann, B. L. and Frederiksen, P. (2009). Thigh circum-

ference and risk of heart disease and premature death:

prospective cohort study. BMJ, 339:b3292.

Liu, Y., Deng, X., Zhu, F., Zhu, W., and Wang, Z. (2023).

High ﬁbrinogen and mixed proximal and distal throm-

bosis are associated with the risk of residual venous

thrombosis in patients with posttraumatic deep vein

thrombosis. Front Cardiovasc Med, 10:1003197.

Matsubara, S., Sugiyama, S., and Waki, N. (2022). Is long-

axis or short-axis pulsed wave doppler optimal for pre-

operative assessment of varicose veins in the lower

extremities? The Japanese Journal of Phlebology,

33:317–321.

Palareti, G., Cosmi, B., Legnani, C., Antonucci, E.,

De Micheli, V., Ghirarduzzi, A., Poli, D., Testa, S.,

Tosetto, A., Pengo, V., Prandoni, P., and DULCIS

(D-dimer and ULtrasonography in Combination Ital-

ian Study) Investigators (2014). D-dimer to guide

the duration of anticoagulation in patients with ve-

nous thromboembolism: a management study. Blood,

124(2):196–203.

Prandoni, P., Lensing, A. W. A., Prins, M. H., Bernardi, E.,

Marchiori, A., Bagatella, P., Frulla, M., Mosena, L.,

Tormene, D., Piccioli, A., Simioni, P., and Girolami,

A. (2002). Residual venous thrombosis as a predictive

factor of recurrent venous thromboembolism. Ann In-

tern Med, 137(12):955–960.

Raghavendra, B. N., Horii, S. C., Hilton, S., Subramanyam,

B. R., Rosen, R. J., and Lam, S. (1986). Deep venous

thrombosis: detection by probe compression of veins.

J Ultrasound Med, 5(2):89–95.

Talbot, S. et al. (1982). Use of real-time imaging in iden-

tifying deep venous obstruction: a preliminary report.

Bruit, 6(41-42):9.

Tan, M., Bornais, C., and Rodger, M. (2012). Interob-

server reliability of compression ultrasound for resid-

ual thrombosis after ﬁrst unprovoked deep vein throm-

bosis. J Thromb Haemost, 10(9):1775–1782.

Zahiri Azar, R. and Salcudean, S. (2006). Motion esti-

mation in ultrasound images using time domain cross

correlation with prior estimates. IEEE transactions on

bio-medical engineering, 53:1990–2000.

BIOIMAGING 2025 - 12th International Conference on Bioimaging

244

APPENDIX

Depth (mm)

10 20 30 40 50

Width (mm)

Lumen decr. (%), -4

Strain (%), 0.01

Depth (mm)

10 20 30 40 50

Width (mm)

Lumen decr. (%), -5

Strain (%), 0.5

Depth (mm)

10 20 30 40 50

Width (mm)

Lumen decr. (%), -13

Strain (%), 1

Depth (mm)

10 20 30 40 50

Width (mm)

Lumen decr. (%), -21

Strain (%), 2

Depth (mm)

10 20 30 40 50

Width (mm)

Lumen decr. (%), -36

Strain (%), 4

Depth (mm)

10 20 30 40 50

Width (mm)

Lumen decr. (%), -66

Strain (%), 6

Depth (mm)

10 20 30 40 50

Width (mm)

Lumen decr. (%), -82

Strain (%), 8

Depth (mm)

10 20 30 40 50

Width (mm)

Lumen decr. (%), -87

Strain (%), 9

Figure 9: Cross-sectional images of femoral vessels during compression session of the thigh tissues. The artery is the most

left circular pattern in all images and does not collapse at the end of the compression session.

Depth (mm)

10 20 30 40

Width (mm)

Lumen decr. (%), -7

Strain (%), 2.5

Depth (mm)

10 20 30 40

Width (mm)

Lumen decr. (%), -9

Strain (%), 3.5

Depth (mm)

10 20 30 40

Width (mm)

Lumen decr. (%), -18

Strain (%), 4.5

Depth (mm)

10 20 30 40

Width (mm)

Lumen decr. (%), -31

Strain (%), 5.5

Figure 10: Cross-sectional images of blood vessels during automated compression (up to strain 5.5%) of thigh tissues: inter-

mediate reactions of one of the subjects characterized in Figures 6.

Depth (mm)

10 20 30 40

Width (mm)

Lumen decr. (%), -5

Strain (%), 2

Depth (mm)

10 20 30 40

Width (mm)

Lumen decr. (%), -17

Strain (%), 5

Depth (mm)

10 20 30 40

Width (mm)

Lumen decr. (%), -36

Strain (%), 8

Depth (mm)

10 20 30 40

Width (mm)

Lumen decr. (%), -68

Strain (%), 11

Figure 11: Cross-sectional images of blood vessels during manual compression (up to strain 11%) of thigh tissues: interme-

diate reactions of one of the subjects characterized in Figures 6.

Automating Compression Ultrasonography of Human Thigh Tissue and Vessels via Strain Estimation

245