Fast volumetric ultrasound facilitates high-resolution 3D mapping of tissue compartments

Abstract

Volumetric ultrasound imaging has the potential for operator-independent acquisition and enhanced field of view. Panoramic acquisition has many applications across ultrasound; spanning musculoskeletal, liver, breast, and pediatric imaging; and image-guided therapy. Challenges in high-resolution human imaging, such as subtle motion and the presence of bone or gas, have limited such acquisition. These issues can be addressed with a large transducer aperture and fast acquisition and processing. Programmable, ultrafast ultrasound scanners with a high channel count provide an unprecedented opportunity to optimize volumetric acquisition. In this work, we implement nonlinear processing and develop distributed beamformation to achieve fast acquisition over a 47-centimeter aperture. As a result, we achieve a 50-micrometer -6-decibel point spread function at 5 megahertz and resolve in-plane targets. A large volume scan of a human limb is completed in a few seconds, and in a 2-millimeter dorsal vein, the image intensity difference between the vessel center and surrounding tissue was ~50 decibels, facilitating three-dimensional reconstruction of the vasculature.

Fig. 1.. Fast volumetric US imaging system.

(A) Schematic of the system. The imaging probe consists of eight linear transducers in octagonal geometry with 1024 elements and a circumcircle diameter of 162 mm. A 3D printed structure supports the octagon assembly that is attached to a motorized scanner and two flexible rubber bellows from the top and the bottom to translate the 1024-element aperture along the elevation direction in water, having the field of view of 130 mm in diameter by 250 mm in height. A programmable 1024-channel US platform drives the customized imaging probe. Block diagrams of (B) global beamformer and (C) partial beamformer. In the global beamformer, the full RF dataset is transferred from the data acquisition board to each node and then transferred from the secondary nodes to the primary node. In the partial beamformer, the RF data transferred to each node are half sized by decentralization, and the secondary nodes transfer only a single image to the primary node. ROI, region of interest; M, number of PWs.

Fig. 2.. Comparison between the global and partial beamformers.

(A) Execution time for the global and partial beamformers by process to acquire a one-frame tomographic image with a varied number of PWs. Total execution time was estimated by four consecutive processes: (i) Tx/Rx events, (ii) data transfer from the data acquisition board to the host computer (HWtoH), (iii) RDMA data transfer from a secondary system to the primary system, and (iv) image reconstruction. Other processes including synchronizing the four nodes, combining IQ data, and display do not unduly affect the frame rate. The field of view was 100 mm (X) × 100 mm (Z) with a pixel size of 1λ. The RF data size for each system was 3712 (samples) × 256 (channels) × (effective number of PWs) × 2 bytes, and the beamformed IQ data size was 350 (X) × 350 (Z) × 8 bytes (complex single precision). The pulse repetition frequency (PRF) was set to 5 kHz. The elapsed time for Tx/Rx events were determined by the PRF of insonation, which is the same for both global and partial beamformers. (B) Frame rate of the global and partial beamformers with a varied number of PWs.

Fig. 3.. Simulations and experiments on resolution for 1-view versus 8-view.

(A) Acquisition schemes of 1-view and 8-view. (B) B-mode images and (C) line profiles in x and z directions of the PSF for 1-view versus 8-view. A point target was located at the center of the arrays. For the 1-view acquisition, 40 PWs (−5° to 5°) were used. For the 8-view acquisition, five PWs (−5° to 5°) were transmitted for one view and the active arrays were rotated (40 PWs in total). (D) Simulated B-mode images of resolution targets for 1-view versus 8-view acquisitions. (E) Simulated and (F) experimental line profiles of two targets with 0.1-mm separation along the x and z axes. For the 1-view acquisition, 40 PWs (−5° to 5°) were used. For the 8-view acquisition, five PWs (−5° to 5°) were transmitted for one view and the active arrays were rotated (40 PWs in total). (G) Simulated B-mode images for 8-view acquisition with (a) 1 Rx, (b) 3 Rx, and (c) 5 Rx for each view. (H) The PSFs along the x direction for the varied number of receive channels. (I) Experimental B-mode images of the nylon wire targets for 8-view acquisition with (a) 1 Rx, (b) 3 Rx, and (c) 5 Rx for each view. (J) RF data of the nylon wire targets recorded by arrays 1 to 3 with array 1 as the transmitting array.

Fig. 4.. Quantitative analysis on a contrast phantom.

(A) B-mode images of a phantom containing hyperechoic and anechoic inclusions obtained from 1-view (24 PWs between −12.5° and 12.5°) and 8-view (3 PWs between −12.5° and 12.5° per view, 24 PWs in total) acquisitions reconstructed by DAS and DAS with averaged CF with a pixel size of 0.1λ. The hyperechoic and anechoic regions are marked by blue and red circles, respectively, and the background region corresponds to the green circle region excluding the hyperechoic and anechoic regions in (A). In the case of the CF-weighted images, the dynamic range was doubled to account for an implicit squaring effect. An array directed upward from the bottom of the images was used for 1-view acquisition. (B) Comparison of the probability density function of the intensity values in the hyperechoic, anechoic, and background regions for DAS-based images (top) and DAS-with-CF-based images (bottom). Quantification on the hyperechoic and anechoic cysts in (A): (C) SD, (D) sSNR, (E) CR, and (F) gCNR. bkg, background.

Fig. 5.. US tomography of rodent anatomy.

(A) Experimental comparison of cross-sectional images of the rat thorax obtained with the octagonal array and reconstructed using a single SOS value and using dual SOS values reconstructed by DAS and DAS with the averaged CF was achieved by correcting the delays for the pixels falling inside the segmented area (red dotted line in b) from the image reconstructed using a single SOS value. Dynamic range is set to cover the highest intensity 90% of pixels. (B) Zoom over the location of the vertebrae (yellow arrows) outlined in (A) in yellow. (C) Zoom over the location of the outer heart wall (green arrows) outlined in (A) in green. (D) Experimental comparison of the cross-sectional image of a rat abdomen imaged with the octagonal array with 1-view acquisition (left) and 8-view acquisition (right). For 1-view, the array was placed vertically at the bottom of the image. (E) The line profiles for 1-view versus 8-view extracted from images in (D). (F) Cross-sectional images of (a) upper and (b) lower thoracic cavity, (c) two lobes of the liver, (d) upper and (e) lower abdominal cavity, (f) abdominopelvic cavity, and (g) upper and (h) lower pelvic cavity in the rat. (G) Cross-sectional images of the (a) sagittal and (b and c) coronal view of the rat trunk. AT, aorta; BM, backbone muscles; CB, coxal bone; CC, cecum; DA, dorsal aorta; HT, heart; IC, ischium; IT, intestines; LF, left femur; LK, left kidney; LL, left lung; LV, liver; JV, jugular vein; RB, rib; RC, rib cage; RF, right femur; RK, right kidney; RL, right lung; RT, rectum; SC, spine; SCP, scapula; SM, stomach; SP, spleen; TC, trachea; UB, urinary bladder; VC, vena cava; VE, vertebra.

Fig. 6.. In vivo images of the human hand, wrist, and forearm.

(A) Schematic for the position of acquisition. (B) Cross-sectional images with 1-view (left column) versus 8-view (right column) for (a) distal hand (finger) section; (b) distal metacarpal section; (c) proximal metacarpal section; and (d) distal forearm section. All images were reconstructed using the dual SOS beamformer with CF weighting at a pixel size of 0.5λ. (C) The line profiles for 1-view versus 8-view extracted from images in (B, b). (D to F) Volume rendered images of the human (a and b) hand-wrist and (c and d) forearm region. The yellow bounding boxes represent the acquired volume of 120 mm (X) × 80 mm (Z) × 250 mm (Y) and 100 mm (X) × 100 mm (Z) × 250 mm (Y), respectively, which acquired in 12.5 s. (D) B-mode, (E) bone, and (F) vessel images are presented in gray, bone, and hot colormap, respectively. ADM, abductor digiti minimi muscle; ADPM, adductor pollicis muscle; APBM, abductor pollicis brevis muscle; DIM, dorsal interosseous muscle; DIPJ, distal interphalangeal joint; DP, distal phalanx; DV, dorsal vein; EDT, extensor digitorum tendon; FCUM, flexor carpi ulnaris muscle; FDSM, flexor digitorum superficialis muscle; FDST, flexor digitorum superficialis tendon; FT, fat; MC, metacarpal; MP, middle phalanx; ODM, opponens digiti minimi muscle; PIPJ, proximal interphalangeal joint; PP, proximal phalanx; R, radius; SK, skin; U, ulna.