Volumetric B-mode ultrasound and Doppler Imaging: Automatic Tracking With One Single Camera

Abstract

Vascular diseases may occur in the upper extremities, and the lesions can span the entire length of the blood vessel. One of the most popular methods to identify vascular disorders is ultrasound Doppler imaging. However, traditional two-dimensional (2D) ultrasound Doppler imaging cannot capture the entire length of a long vessel in one image. Medical professionals often have to painstakingly reconstruct three-dimensional (3D) data using 2D ultrasound images to locate the lesions, especially for large blood vessels. 3D ultrasound Doppler imaging can display the morphological structure of blood vessels and the distribution of lesions more directly, providing a more comprehensive view compared to 2D imaging. In this work, we propose a wide-range 3D volumetric ultrasound Doppler imaging system with dual modality, in which a high-definition camera is adopted to automatically track the movement of the ultrasound transducer, simultaneously capturing a corresponding sequence of 2D ultrasound Doppler images. We conducted experiments on human arms using our proposed system and separately with X-ray computerized tomography (X-CT). The comparison results prove the potential value of our proposed system in the diagnosis of arm vascular diseases.

Keywords

volumetric imaging dual-modality imaging ultrasound imaging wide-range imaging

Introduction

Ultrasound Doppler imaging technology has been widely used in the field of medical detection and diagnosis due to its noninvasiveness and the ability to provide highly sensitive anatomical and functional mapping of blood vessels.^1–8 Compared to B-mode ultrasound imaging, Doppler ultrasound imaging can provide additional information about the dynamic functional characteristics of the target tissue, including blood flow velocity and direction.⁹ Therefore, Doppler ultrasound imaging shows significant potential in diagnosing vascular diseases such as peripheral arterial disease (PAD), chronic venous disease (CVD), and deep vein thrombosis (DVT) in the upper extremities. Generally, this type of disease occurs due to arterial blockage or the formation of blood clots in the deep venous system, which is known as the atherosclerosis.^7,10–12 Worldwide, PAD has already affected almost 200 million people until the year 2020.¹³ CVD can lead to venous ulcerations, resulting in cutaneous dystrophy, delayed healing, and recurrent ulceration. The overall prognosis for CVD is poor, with more than half of the cases requiring extensive therapy lasting over a year.^14–16 Symptoms of DVT include swelling, tenderness, and redness in the affected area. However, some patients with DVT may be asymptomatic, increasing the risk of potentially fatal complications.^12,17 Many experiments have been conducted to demonstrate the high sensitivity and specificity of ultrasound Doppler in detecting vascular diseases. O’Leary et al.¹⁸ conducted a study involving 50 subjects and found that ultrasound Doppler had a sensitivity of 88% and a specificity of 96% in detecting DVT, comparable to venography. In another review of 9 studies, Doppler sonography showed a sensitivity of 97% and a specificity of 96% in diagnosing clinically suspected venous thrombosis.¹⁹

Upper extremity venous thrombosis may also occur concomitantly with other diseases, such as diabetes and sarcoma. Luo et al.²⁰ found a high rate of upper extremity venous thrombosis related to peripherally inserted central catheters (PICC) in oncology patients, and regular weekly Doppler screening after PICC insertions could reduce the occurrence of symptomatic thrombosis. A convenient and comfortable method for arm vascular imaging is necessary for routine physical examinations. However, vascular detection is time-consuming and requires a highly trained and experienced operator, especially for oversized organs.²¹ While ultrasound Doppler imaging proves efficacious in the detection of vascular pathologies in the extremities, it remains confined to provide a 2D cross-sectional view. Three-dimensional (3D) ultrasound Doppler imaging presents a more feasible approach in aiding clinicians with intricate vascular diseases, eliminating the need for mental reconstruction of stereoscopic information from 2D slices,²² which shows potential applications in the diagnosis of vascular diseases related to limbs.²³

Several approaches for 3D ultrasound Doppler imaging based on 3D ultrasound imaging have been proposed and can generally be classified into three categories. One approach involves directly obtaining 3D images using 2D array transducers, while another approach involves acquiring the relative locations and orientations of the 2D image sequence using a positioning device to reconstruct the volumetric data. The third approach involves tracking speckle decorrelation during a smooth, preferably linear, transducer sweep to obtain the motion rate.²⁴

The use of 2D array transducers allows for direct acquisition of volumetric scans by emitting beams that cover a pyramidal-shaped volume in both the azimuth and elevation dimensions.²⁵ However, these transducers are typically expensive due to their complex manufacturing process, and although they can directly visualize targets in 3D, their imaging range and resolution are limited.

The second approach uses traditional linear array transducers and can be grouped into three categories: dedicated 3D transducers, mechanical scanning approach, and freehand scanning approach.²⁶ Dedicated transducers integrate a regular linear transducer with a motor inside a handheld instrument, allowing it to rotate, tilt, or translate under the control of the computer, providing real-time 3D images.²⁷ However, the transducer needs to be held still during scanning, making it unsuitable for visualizing large-volume organs in 3D.²⁸ On the other hand, mechanical scanners fix the linear array transducer and motorized mechanisms with an external fixture.²⁹ The scanning route of mechanical scanners is predefined to obtain precise position data but is limited by the scanning range.^28,30

Freehand 3D ultrasound Doppler imaging offers greater maneuverability and flexibility compared to the other two modalities. In freehand imaging, a handheld linear array ultrasound transducer is used to acquire images through multiple 2D scans, while a positioning system records the position of each scan. This information is then used to reconstruct the 3D volumetric data. This technique allows the operator to manipulate the transducer freely, enabling clinicians to choose the optimal imaging position and angle. Four types of positioning devices are commonly used to track the transducer, magnetic tracker, acoustic tracker, articulated arm positioner, and optical tracker. A magnetic tracker consists of a magnetic transmitter and a receiver. The transmitter is placed near the patients, while the receiver is attached to the transducer to track its movement by measuring the strength of the magnetic field. The magnetic tracker is relatively small and flexible, and it does not require unobstructed sight. However, electromagnetic interference and the presence of metallic objects may compromise tracking accuracy and cause distortion.³¹ In comparison, an acoustic tracker involves mounting sound-emitting devices on the transducer and placing the microphone array over the patient.³² The position and orientation can be calculated using knowledge of the speed of sound in the air. However, it is easily affected by environmental humidity. Additionally, occlusion between the sound generator and the microphone can affect signal reception.¹

The articulated arm positioner utilizes potentiometers on the joints to continuously monitor the angulation and position of articulated arms. This method is effective for calculating the spacing information of the transducer for 3D reconstruction. However, it is important for the articulated arms to be as short as possible in order to improve precision, which may result in a limited range of view.²³ A freehand transducer with an optical positioner system generally consists of passive or active targets fixed on the transducer, and at least two cameras are used to track targets.^33,34 The cameras receive the infrared light reflected from the passive markers or emitted by the active markers to calculate the position and orientation information. The space between the cameras and the transducer should be free of obstacles. Additionally, the markers should remain within the view range of the cameras during motion. Although this limitation can be overcome by using additional cameras, it would result in high costs and require a large amount of space, such as using eight cameras to cover all angles.³⁵ Apart from the four types of positioners mentioned above, there are also several novel methods of 3D reconstruction. For instance, Cai et al. utilized inertia measurement units (IMUs) to track the transducer. However, IMUs are unable to compensate for the significant drift from the integration of acceleration measurements, leading to inaccurate orientation information.³⁴ A 3D translating device and a linear sliding track with a position sensor were used to develop a 3D Ultrasound Strain Imaging.^29,36 In addition, a new robotic ultrasound system for spine imaging, which incorporates more anthropomorphic scanning manipulation compared to previously reported techniques, has also been developed.³⁷

Various freehand 3D ultrasound reconstruction algorithms have been reported and evaluated.³⁸ These algorithms can generally be classified into three categories: function-based methods (FBMs), voxel-based methods (VBMs), and pixel-based methods (PBMs).³⁹ FBMs establish a specific functional relationship between the pixels in the 2D ultrasound images and their spatial positions, and then calculate the voxel values in the volume grid based on this function. The precision of this method is high, but it typically requires several hours to complete the 3D reconstruction of approximately 200 ultrasound images.³⁹ Voxel nearest neighbor (VNN) is a basic method used in VBMs. It involves traversing the voxel, calculating the shortest distance between the voxel and the ultrasound image planes to find the nearest pixel, and assigning the value of the nearest pixel to the voxel.⁴⁰ While VBM is the most intuitive method and can preserve the original texture patterns from ultrasound images,⁴¹ it tends to generate large artifacts when the voxel is far from the image plane.²⁶ Pixel nearest neighbor (PNN), one of the PBMs, involves traversing each pixel in the ultrasound images and assigning the pixel value to its nearest voxel.⁴² Similar to VNN, it can preserve texture patterns, but it may also introduce artifacts. Some researchers utilized deep learning to generate a three-dimensional model of the human spine, and then reconstructed the entire spine by reconnecting all the vertebral models in a 3D space.⁴³ Considering the large amount of data collected in our experiment, we employed the computationally efficient PNN algorithm for reconstructing 3D volume data, including both B-mode ultrasound images and ultrasound Doppler images.

The system proposed in this work is based on an optical positioner. However, traditional optical positioners have limitations in terms of tracking range, making them unsuitable for oversize targets. Additionally, the camera’s ability to locate highly oblique targets is limited by the depth of view. To address these issues, we proposed a volumetric Ultrasound Doppler imaging system that enables wide-range volumetric freehand Ultrasound Doppler imaging with dual modality. Furthermore, we designed the system to automatically track the moving transducer using only one camera. A passive marker is attached to the transducer, eliminating the need for launching infrared light. Instead, the transducer’s movement is tracked on the camera screen. During the acquisition process, we can independently select the imaging area through real-time 2D imaging. The reconstruction results obtained from blood vessels in the arm, validated by X-ray computerized tomography (X-CT), demonstrated that our proposed system holds significant value and is expected to be beneficial in the diagnosis of vascular diseases.

Materials and Methods

The structure of the system is depicted in Figure 1. It comprises three components: a 2D Ultrasound Doppler imaging system with a linear array, a single-camera tracking system, and a dual-modality volumetric ultrasound imaging system. Each component is explained in detail as follows.

System Setup

Figure 2 illustrates the components of the entire system utilized for this demonstration. The ultrasound Doppler imaging system comprises a linear array ultrasound transducer (9L-4) and a workstation (LOGIQ E9, GE Medical Systems Ultrasound and Primary Care Diagnostics, LLC 9900 Innovation Drive Wauwatosa, WI, USA) for acquiring, processing the radiofrequency (RF) data as well as reconstructing, visualizing the volume data. We adopted the power ultrasound Doppler imaging mode to acquire ultrasound Doppler images. The single-camera tracking system consists of three components, a high-resolution camera (MV-CH650-90XM, HIKVISION, Hangzhou, China), two markers, and a motor control module. One of the markers is affixed to the bracket fixed on the transducer, while the other is positioned in front of the target to track the camera, as shown in Figure 2. The motor control module comprises a power supply (AC, 220 V), a motion controller, a micro-step driver, and a bipolar step motor. Communication between the motor control module and the workstation occurs via an RS232 serial port, enabling the control of camera motion at the appropriate time to track the transducer.

Target Tracking

The camera is mounted on the slide block of the guide rail, as shown in Figure 2. When the transducer approaches the edge of the camera screen, a control signal is transmitted to the motion controller of the guide rail via the RS232 serial port to adjust the position of the camera. The control procedure is outlined in Algorithm (1). Consequently, it is necessary to track both the camera and ultrasound transducer simultaneously.

Algorithm (1).

The algorithm for controlling the movement of the camera.

Input: The distance from the ultrasound transducer to the left boundary of the image:

d_{t a r g e t}

Output: Send control signal through the RS232 serial port to drive or stop the camera
Step1. Set the parameters
1: Set the minimum distance from the ultrasound transducer to the image boundary:

d_{1} = \frac{1}{3} c o l s

, where

c o l s

is the horizontal resolution of the image
2: Set the threshold:

t_{t h}

to limit the movement time of the camera
Step2. Track the target
1: Detect the ultrasound transducer and calculate the distance from the ultrasound transducer to the left boundary of the image:

d_{t a r g e t}

2: if

d_{t a r g e t} < d_{1} o r d_{t a r g e t} > c o l s - d_{1}

then
3: Send a control signal to drive the camera
4: Record the movement time of the camera:

t_{p}

5: if

t_{p}

t_{t h}

then
6: Send a control signal to stop the camera
7: endif
8: endif
9: return

Tracking the camera

The marker we used to track the camera is depicted in Figure 3(a), which is a glass plate measuring 880 mm × 80 mm. It is printed with three Augmented Reality University of Córdoba (ArUco) Boards. The interval between each ArUco board is 400 mm, ensuring that only one ArUco board can be detected in the image. We designated the rightmost ArUco Board as the reference and calibrated the positional relationship between the reference and the other two ArUco Boards in advance. The corresponding rotation and translation matrixes are denoted as $R_{b 2 b}^{1}$ , $t_{b 2 b}^{1}$ and $R_{b 2 b}^{2}$ , $t_{b 2 b}^{2}$ .

Tracking the transducer

With these prior data, we can locate and track the movement of the ultrasound transducer over a long distance. A three-sided calibration plate is attached on the ultrasound transducer to accurately measure the position and orientation changes in real-time between each frame. The principle of calibration has been described in detail.⁴⁴ The design of the three-sided calibration plate is shown in Figure 3(b). Each plane of the calibration plate has a calibration mark attached, and the angle between each face is set to 120°. The pattern on each calibration plate consists of a circular asymmetric checkerboard and a pair of ArUco makers. The calibration method for these markers is explained in detail in Garrido-Jurado et al.⁴⁵ and Wang et al.⁴⁶ The algorithm for detecting the three-sided calibration plate is described in Algorithm (2).

Algorithm (2).

Detect the three-sided calibration plate.

Input: Ultrasound images:

{I_{1}, I_{2}, \dots, I_{N}}

Output:

{R_{o 1}, R_{o 2}, \dots, R_{o N}}, {t_{o 1}, t_{o 2}, \dots, t_{o N}}

Step1. Set the parameters
1: Detect the ArUco and set the parameters and the region size of the three-sided calibration board in pixels
Step2. Detect the initial frame
1: Detect the ArUco markers in the region set in Step 1
2: Estimate the marker poses:

{R_{1}, R_{2}, \dots, R_{N}}, {t_{1}, t_{2}, \dots, t_{N}}

3: for i = 1 → N do
4: Calculate

R_{i} t_{z}, t_{z} = {[0, 0, 1]}^{T}

to choose the board facing forward
5: end for
6: Set a region centered in the chosen board to detect the circle centers

{(x_{1}, y_{1}), (x_{2}, y_{2}), \dots, (x_{M}, y_{M})}

Step 3. Detect the interframe.
1: for i = 1 → N do
2: while (detect circles successfully) do
3: for j = 1 → M do
4: Set a series of segments centered in the circle center

(x_{j}, y_{j})

detected in the previous frame
5: end for
6: Detect circles in the segments using connected domain analysis
7: Record the offset of the three-sided calibration board
8: Estimate the camera pose
9: return

R_{o i}, t_{o i}

10: end while
11: if detect circles unsuccessfully then
12: Reset the calibration board region according to the offset and go to Step 2
13: end if
14: end for

The positional relationship between the marker and the transducer is calibrated in advance and denoted as $R_{I 2 w}$ , $t_{I 2 w}$ . During practical operation, the images of the transducer are collected in real time using a video capture card. These images are then used to calculate the positions and orientations, which are aligned with the acquired 2D ultrasound Doppler images. The rotation and translation matrixes between the marker attached to the transducer and the camera are calculated from the images and denoted as $R_{w 2 c} [i]$ , $t_{w 2 c} [i]$ , where $i$ represents the index of the images. The relative positional relationship of the coordinate systems is depicted in Figure 4. The coordinate system in an ultrasound image, denoted as $O - X Y Z$ , is positioned in the image plane, centered in the upper left corner of the image, and with the Z-axis pointing inward. The coordinate system of the calibration plate fixed on the transducer is denoted as $O_{w} - X_{w} Y_{w} Z_{w}$ , and the coordinate system of the camera is denoted as $O_{c} - X_{c} Y_{c} Z_{c}$ . The coordinate system in an ArUco grid board is positioned in the board plane, centered in the bottom left corner of the board, and with the Z-axis pointing outward. This coordinate system is denoted as $O_{b} - X_{b} Y_{b} Z_{b}$ . We place the long glass plate in front of the target. The rotation and translation matrixes between the camera and the long glass plate are denoted as $R_{c 2 b}$ , $t_{c 2 b}$ . A random point in the ultrasound image coordinate system $P_{I}$ is transformed into $P_{b}$ in the ArUco board coordinate system. The coordinate transformation relation is expressed as

P_{b} = R_{c 2 b} (R_{w 2 c} [i] (R_{I 2 w} P_{I} + t_{I 2 w}) + t_{w 2 c} [i]) + t_{c 2 b} .

(1)

Volumetric Reconstruction

PNN mainly consists of two steps: the pixel-distribution step (PDS) and the hole-filling step (HFS). A volumetric grid is constructed using two parameters: the size of the reconstructed volume data and the physical interval between voxels. The volume size, denoted as $s_{recon}$ , is predefined based on the imaging depth of the transducer and the section size of the target. The physical interval, denoted as $d_{inv}$ , indicates the physical size in millimeters of each voxel. This parameter determines the resolution of the volume data, the precision, and the reconstruction time.

In the PDS, the pixels in 2D ultrasound images are mapped to the voxels in the volume data. First, the center of the volumetric grid is set. The transformation can be expressed as

P_{s h i f t} = P_{b}^{m} - [\begin{matrix} s_{recon} / 2 \\ s_{recon} / 2 \\ s_{recon} / 2 \end{matrix}],

(2)

where the point $P_{s h i f t}$ represents the center of the volumetric grid, and the point $P_{b}^{m}$ is transformed from a random point $P_{I}^{m}$ using (1).

Then, traverse each pixel $P_{I}^{n}$ . The corresponding voxel $P$ can be calculated by:

P = (P_{b}^{n} - P_{s h i f t}) / d_{inv},

(3)

where the point $P_{b}^{n}$ is transformed from $P_{I}^{n}$ using (1). The voxel value $f (P)$ is calculated by

f (P) = I_{n},

(4)

where $I_{n}$ is the pixel value of $P_{I}^{n}$ .

Generally, there are often vacant voxels in the reconstructed volumetric grid due to the irregular and sparse scan. The HFS step is used to fill these gaps by utilizing surrounding known voxels. The procedure is illustrated in Figure 5. It traverses each voxel $Q$ to check whether the voxel value is empty. If it is empty, the voxel is taken as the center, and a spherical neighborhood with a radius of $r$ is determined. The collection of nonempty voxels in the neighborhood is denoted as ${X_{k}}$ . Then, the weighted average of the voxel values in the set is assigned to the voxel $Q$ . This is described by

f (Q) = \sum_{k \in N} ω_{k} f (X_{k}), ω_{k} = \frac{1}{N},

(5)

where $ω_{k}$ represents the distance weight of voxel $X_{k}$ , and $N$ is the number of voxels in ${X_{k}}$ . This step is not necessary if the scanning is done densely. After completing the above steps, the pixels in the ultrasound images are successfully mapped into the voxels in the volumetric grid.

Results and Discussion

To evaluate the performance of our tracking method, we made a phantom using a 12% gelatin solution and silicone hoses to mimic tissue and blood vessels. The water is pumped into the hose using an absorption pump to ensure a steady flow. As depicted in Figure 6(a), we put a hose with a 5 mm inner diameter into the phantom. The reconstruction results can be seen in Figure 6(b), and the measured results are presented in Table 1. The contours of the hose are well-defined, and it is evident that the overall shape and direction of the hoses are clear. The relative positional relationship between the two hoses at different depths is distinct.

Table 1.

The Measured Results of the Inner Diameter.

The inner diameter (mm)	Measured results (mm)			Mean (mm)
5.00	5.03	5.28	5.17	5.16

To compare our system with the optical positioner system that uses a single board, we measured the angular range at which the ultrasound transducer can be detected when it rotates along the Z-axis and the Y-axis of the camera, as well as the maximum target size. We affixed the calibration plate onto the displacement table to measure the angular range. The displacement table can rotate along the $Z_{d}$ -axis, as depicted in Figure 7(b), and its angular accuracy is $\pm 1 °$ . The results of these measurements are presented in Table 2, demonstrating improvements in both imaging size and range. Additionally, we conducted experiments on the left arm at different angles to showcase the advantages of utilizing multiple scanning angles. The reconstruction result with one angle is shown in Figure 8(a) and (c), and the result reconstructed from multiple angles is presented in Figure 8(b) and (d). For emphasis on the vessels, we only displayed the reconstruction results of the ultrasound Doppler images. It is evident that more details regarding the blood vessels can be detected when utilizing multiple angles.

Table 2.

The Measured Results of Different Systems.

	Angular range rotating along the Z-axis (°)	Angular range rotating along the Y-axis (°)	Target size (mm)
Single-board	74	60	300
Our system	131	114	1100

To verify the practicality of the system in vivo, we conducted an experiment on the left arm of a volunteer. The target is depicted in Figure 9(a). We scanned the blood vessels inside the elbow and moved the transducer along the arm while keeping the calibration plate fixed on the transducer. The length of the target is 350 mm, and the imaging depth is 30 mm with a reconstruction resolution of 0.5 mm per pixel. We acquired 5974 images to reconstruct the final volumetric image, which clearly depicts the direction and morphological changes of blood vessels. Considering that the vascular morphology varies significantly among individuals, we supplied an X-CT scan to the volunteer and rendered the result in 3D to obtain the accurate morphology of the blood vessels in the arm as the reference. The same volunteer lay on the X-CT table with her left arm stretched beyond her head and received a contrast injection to enhance blood vessel imaging, as shown in Figure 9(b). The rendering result is displayed in Figure 9(c). We measured the vessel diameter at points A and B, as indicated in Figure 9(c). The measured results are listed in Table 3. As we can observe, the reconstruction result from our proposed method is generally consistent with the X-CT result in terms of morphology.

Table 3.

The Measured Results of the Vessel Diameter.

	Our reconstruction result	X-CT rendering result
The vessel diameter (mm)	2.95	2.88

Conclusion

In this study, we introduce a wide-range volumetric ultrasound Doppler imaging system with dual modality to track the movement of the ultrasound transducer across a wide range and visualize the blood vessels in 3D. To accommodate large-size targets, such as a whole human arm, we utilize a guide rail to expand the visual space of the high-definition camera. Moreover, we propose a three-sided calibration plate that enhances the field of view when the ultrasound transducer rotates in the direction of the camera’s depth. We conducted several contrast experiments to evaluate our system, and the results demonstrate its capability to increase the angular range by $57 \circ$ in the Z-axis, $54 \circ$ in the Y-axis. Additionally, the target size is expanded by 800 mm, enabling the detection of more vascular details with a greater number of angles. If we disable the serial port and cover the other two sides of the calibration plate, this system can function as a single camera system with a one-sided calibration plate. Our proposed system enables the reconstruction of the 3D dual modality of large targets, which is advantageous for assisting clinicians in diagnosing arm vascular diseases. However, the continuity of the reconstructed vessels is constrained by the collection interval. Slow acquisition can produce dense reconstruction results, but it also leads to a longer and more labor-intensive acquisition process.

The imaging technology we propose is straightforward and completely noninvasive, eliminating the need for injections. This makes it a competitive option for regular check-ups compared to existing imaging techniques like angiography, which may be invasive. Regular check-ups are necessary to monitor the gradual buildup of fatty materials in the vessels, leading to artery blockages. Our proposed solution, dual-modality volumetric ultrasound Doppler imaging, could address this need. In conclusion, our system limitations associated with optical positioners and exhibits promising prospects for diagnosing arm vascular diseases.

Footnotes

Data Availability

All materials and data generated from this study are available upon request to the corresponding author.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was supported in part by the Natural Science Funds of China (No: 12027808). Natural Science Funds of Jiangsu Provinces of China (No: BK20231399).

ORCID iD

Jie Yuan

References

Kim

Kang

Shim

Seo

Kim

, et al. Versatile low-cost volumetric 3D ultrasound imaging using gimbal-assisted distance sensors and an inertial measurement unit. Sensors (Basel). 2020;20(22):6613.

Karim

Roberts

Salomon

Papageorghiou

AT.

Systematic review of first-trimester ultrasound screening for detection of fetal structural anomalies and factors that affect screening performance. Ultrasound Obstet Gynecol. 2017;50(4):429-41.

Drukker

Droste

Ioannou

Impey

Noble

Papageorghiou

AT.

Function and safety of SlowflowHD ultrasound Doppler in Obstetrics. Ultrasound Med Biol. 2022;48(6):1157-62.

Jia

Liu

Dou

Comparison of the value of color Doppler ultrasound and multislice spiral CT in the differential diagnosis of benign and malignant nodules in the liver. Contrast Media Mol Imaging. 2022;2022:5251966.

Hui

Liu

Deng

The clinical value of color Doppler ultrasonography in measuring the hemodynamics of liver cirrhosis patients’ portal and splenic veins. Am J Transl Res. 2021;13(3):1692-700.

Riggin

Chen

Gordon

Schultz

Soslowsky

Khoury

Ultrasound-guided dry needling of the healthy rat supraspinatus tendon elicits early healing without causing permanent damage. J Orthop Res. 2019;37(9):2035-42.

Shabani Varaki

Gargiulo

Penkala

Breen

PP.

Peripheral vascular disease assessment in the lower limb: a review of current and emerging non-invasive diagnostic methods. Biomed Eng Online. 2018;17(1):61.

Provost

Papadacci

Demene

Gennisson

Tanter

Pernot

3-D ultrafast Doppler imaging applied to the noninvasive mapping of blood vessels in vivo. IEEE Trans Ultrason Ferroelectr Freq Control. 2015;62(8):1467-72.

Poelma

Ultrasound Imaging Velocimetry: a review. Exp Fluids. 2017;58(1):3.

10.

Bergan

Pascarella

CHAPTER 4 - Venous anatomy, physiology, and pathophysiology. In: Bergan

, ed. The Vein Book. Cambridge: Academic Press; 2007, pp. 39-45.

11.

Bundens

WP.

6 - the chronically swollen leg: finding the cause: Theory and practice. In: Bergan

Shortell

, eds. Venous Ulcers. Cambridge: Academic Press; 2007, pp. 67-74.

12.

Office of the Surgeon General (US), & National Heart, Lung, and Blood Institute (US). The Surgeon General’s Call to Action to Prevent Deep Vein Thrombosis and Pulmonary Embolism. Rockville (MD): Office of the Surgeon General (US); 2008.

13.

Shahrulakmar

Omar

Johari

NH.

Brief review on recent advancement of computational analysis on hemodynamics in peripheral artery disease. In: Hassan

MHA

Zohari

Kadirgama

Mohamed

NAN

Aziz

, eds. Technological Advancement in Instrumentation & Human Engineering. Singapore: Springer Nature; 2023, pp. 555-72.

14.

Barron

Jacob

Kirsner

RS.

Dermatologic complications of chronic venous disease: medical management and beyond. Ann Vasc Surg. 2007;21(5):652-62.

15.

Eberhardt

Raffetto

JD.

Chronic venous insufficiency. Circulation. 2014;130(4):333-46.

16.

Scott

LaMorte

Gorin

Menzoian

JO.

Risk factors for chronic venous insufficiency: a dual case-control study. J Vasc Surg Cases. 1995;22(5):622-8.

17.

Barker

Marval

Venous thromboembolism: risks and prevention. Contin Educ Anaesth Crit Care Pain. 2011;11(1):18-23.

18.

O'Leary

Kane

Chase

. A prospective study of the efficacy of B-scan sonography in the detection of deep venous thrombosis in the lower extremities. J Clin Ultrasound. 1988;16(1):1-8.

19.

Di Nisio

Van Sluis

Bossuyt

Büller

Porreca

Rutjes

AW.

Accuracy of diagnostic tests for clinically suspected upper extremity deep vein thrombosis: a systematic review. J Thromb Haemost. 2010;8(4):684-92.

20.

Luo

Jing

Wang

Qin

Liang

Liu

SS.

Peripherally inserted central catheter-related upper extremity venous thrombosis in oncology patients: A prospective study based on Doppler sonography. J Ultrasound Med. 2016;35(8):1759-63.

21.

Eiberg

Grønvall Rasmussen

Hansen

Schroeder

TV.

Duplex ultrasound scanning of peripheral arterial disease of the lower limb. Eur J Vasc Endovasc Surg. 2010;40(4):507-12.

22.

Huang

Zeng

A review on real-time 3D ultrasound imaging technology. Biomed Res Int. 2017;2017:6027029.

23.

Mennitt

Deol

Gao

Emergency color Doppler sonography of the extremity artery: a pictorial essay. Clin Imaging. 2017;42:240-8.

24.

Yen

Steinberg

Smith

SW.

Sparse 2-D array design for real time rectilinear volumetric imaging. IEEE Trans Ultrason Ferroelectr Freq Control. 2000;47(1):93-110.

25.

Chen

Fowlkes

Carson

Rubin

JM.

Determination of scan-plane motion using speckle decorrelation: theoretical considerations and initial test. Int J Imaging Syst Technol. 1997;8(1):38-44.

26.

Wen

Zhu

Qin

Yang

Xie

, et al. An accurate and effective FMM-based approach for freehand 3D ultrasound reconstruction. Biomed Signal Process Control. 2013;8(6):645-56.

27.

Prager

Ijaz

Gee

Treece

GM.

Three-dimensional ultrasound imaging. Proc IMechE, Part H: J Engineering in Medicine. 2010;224(2):193-223.

28.

Huang

Zheng

Chi

ZR.

Development of a portable 3D ultrasound imaging system for musculoskeletal tissues. Ultrasonics. 2005;43(3):153-63.

29.

Huang

Xie

Chen

3-D ultrasonic strain imaging based on a linear scanning system. IEEE Trans Ultrason Ferroelectr Freq Control. 2015;62(2):392-400.

30.

Toonkum

Suwanwela

Chinrungrueng

Reconstruction of 3D ultrasound images based on cyclic regularized Savitzky-Golay filters. Ultrasonics. 2011;51(2):136-47.

31.

Jiang

Chen

A real-time freehand 3D ultrasound imaging method for scoliosis assessment. J Appl Clin Med Phys. 2022;23(8):e13709.

32.

Fenster

Downey

Cardinal

HN.

Three-dimensional ultrasound imaging. Phys Med Biol. 2001;46(5):R67-99.

33.

De Lorenzo

Vaccarella

Khreis

Moennich

Ferrigno

De Momi

. Accurate calibration method for 3D freehand ultrasound probe using virtual plane. Med Phys. 2011;38(12):6710-20.

34.

Cai

Peng

Prieto

Rosenbaum

Stringer

Jiang

A low-cost camera-based ultrasound probe tracking system: design and prototype. In: 2019 IEEE International Ultrasonics Symposium, pp. 997-99.

35.

Chung

Shih

Huang

CC.

Freehand three-dimensional ultrasound imaging of carotid artery using motion tracking technology. Ultrasonics. 2017;74:11-20.

36.

Chen

Huang

Development of a wireless and near real-time 3D ultrasound strain imaging system. IEEE Trans Biomed Circuits Syst. 2016;10:394-403.

37.

Huang

Wang

Luo

On mimicking human’s manipulation for robot-assisted spine ultrasound imaging. Robot Intell Autom. 2023;43(4):373-81.

38.

Rohling

Gee

Berman

A comparison of freehand three-dimensional ultrasound reconstruction techniques. Med Image Anal. 1999;3(4):339-59.

39.

Fan

Wen

Wang

Xie

, et al. The latest research progress of freehand 3D US volume reconstruction. Chin J Biomed Eng. 2017;36:92-102.

40.

Prager

Gee

Berman

Stradx: real-time acquisition and visualization of freehand three-dimensional ultrasound. Med Image Anal. 1999;3(2):129-40.

41.

Huang

Zheng

Chi

Speckle suppression and contrast enhancement in reconstruction of freehand 3D ultrasound images using an adaptive distance-weighted method. Appl Acoust. 2009;70(1):21-30.

42.

San José-Estépar

Martín-Fernández

Caballero-Martínez

Alberola-López

Ruiz-Alzola

A theoretical framework to three-dimensional ultrasound reconstruction from irregularly sampled data. Ultrasound Med Biol. 2003;29(2):255-69.

43.

Huang

Luo

Yang

Deng

Liu

, et al. Anatomical prior based vertebra modelling for reappearance of human spines. Neurocomputing. 2022;500:750-60.

44.

Fan

Yuan

Chen

Tao

, et al. Free scan real time 3D ultrasound imaging with shading artefacts removal. Ultrasonics. 2023;135:107091.

45.

Garrido-Jurado

Muñoz-Salinas

Madrid-Cuevas

Marín-Jiménez

MJ.

Automatic generation and detection of highly reliable fiducial markers under occlusion. Pattern Recognit. 2014;47:2280-92.

46.

Wang

Zhao

Jiang

Wang

Zhang

YD.

Zhang’s camera calibration method based on circular array calibration board. Trans Beijing Inst Technol. 2019;39(8):859.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB