3D Ultrasound Volumes Research Articles

Recent Advances in Artificial Intelligence-Assisted Ultrasound Scanning

Ultrasound (US) is a flexible imaging modality used globally as a first-line medical exam procedure in many different clinical cases. It benefits from the continued evolution of ultrasonic technologies and a well-established US-based digital health system. Nevertheless, its diagnostic performance still presents challenges due to the inherent characteristics of US imaging, such as manual operation and significant operator dependence. Artificial intelligence (AI) has proven to recognize complicated scan patterns and provide quantitative assessments for imaging data. Therefore, AI technology has the potential to help physicians get more accurate and repeatable outcomes in the US. In this article, we review the recent advances in AI-assisted US scanning. We have identified the main areas where AI is being used to facilitate US scanning, such as standard plane recognition and organ identification, the extraction of standard clinical planes from 3D US volumes, and the scanning guidance of US acquisitions performed by humans or robots. In general, the lack of standardization and reference datasets in this field makes it difficult to perform comparative studies among the different proposed methods. More open-access repositories of large US datasets with detailed information about the acquisition are needed to facilitate the development of this very active research field, which is expected to have a very positive impact on US imaging.

First trimester 3D ultrasound placental volume at 11th to 13th weeks gestational age for predicting preeclampsia

Background: After the 20th week of pregnancy, preeclampsia is diagnosed when a woman develops new-onset hypertension (systolic blood pressure 140 mm Hg or higher, or diastolic blood pressure 90 mm Hg or higher) and proteinuria (300 mg or more in a 24-hour urine collection, or 1+ on dipstick testing). The aim of this work was to determine the sensitivity of placental volume (PV) in the first trimester at 11th week to 13th week+6 days in predicting preeclampsia. Methods: This prospective observational study was carried out on 250 women with any gravidity, singleton pregnancy between 11th to 13th week of gestation based on the last menstrual period or first-trimester ultrasound. All patients were subjected to laboratory investigations (CBC, Renal function tests, Random blood sugar and tests to detect proteinuria.) and imaging. Results: The sensitivity and specificity of PV in differentiating between preeclampsia and normotensive were 60.0%; 60.0% respectively. There was high significant difference in between preeclmpesia and normotensive cases regarding to PV. Conclusions: The PV was lower in cases with preeclampsia as compared to nonpreeclamptic women with highly significant statistical difference and that means 3D PV measurement at 11th to 13th week of pregnancy could increase the predictive accuracy for pre-eclampsia.

Ultrasound Volume Reconstruction From Freehand Scans Without Tracking.

Transrectal ultrasound is commonly used for guiding prostate cancer biopsy, where 3D ultrasound volume reconstruction is often desired. Current methods for 3D reconstruction from freehand ultrasound scans require external tracking devices to provide spatial information of an ultrasound transducer. This paper presents a novel deep learning approach for sensorless ultrasound volume reconstruction, which efficiently exploits content correspondence between ultrasound frames to reconstruct 3D volumes without external tracking. The underlying deep learning model, deep contextual-contrastive network (DC 2-Net), utilizes self-attention to focus on the speckle-rich areas to estimate spatial movement and then minimizes a margin ranking loss for contrastive feature learning. A case-wise correlation loss over the entire input video helps further smooth the estimated trajectory. We train and validate DC 2-Net on two independent datasets, one containing 619 transrectal scans and the other having 100 transperineal scans. Our proposed approach attained superior performance compared with other methods, with a drift rate of 9.64 % and a prostate Dice of 0.89. The promising results demonstrate the capability of deep neural networks for universal ultrasound volume reconstruction from freehand 2D ultrasound scans without tracking information.

Adaptive tetrahedral interpolation for reconstruction of uneven freehand 3D ultrasound

Objective. Freehand 3D ultrasound volume reconstruction has received considerable attention in medical research because it can freely perform spatial imaging at a low cost. However, the uneven distribution of the original ultrasound images in space reduces the reconstruction effect of the traditional method. Approach. An adaptive tetrahedral interpolation algorithm is proposed to reconstruct 3D ultrasound volume data. The algorithm adaptively divides the unevenly distributed images into numerous tetrahedrons and interpolates the voxel value in each tetrahedron to reconstruct 3D ultrasound volume data. Main results. Extensive experiments on simulated and clinical data confirm that the proposed method can achieve more accurate reconstruction than six benchmark methods. Specifically, the averaged interpolation error at the gray level can be reduced by 0.22–0.82, and the peak signal-to-noise ratio and the mean structure similarity can be improved by 0.32–1.83 dB and 0.01–0.05, respectively. Significance. With the parallel implementation of the algorithm, one 3D ultrasound volume data with size 279 × 279 × 276 can be reconstructed from 100 slices 2D ultrasound images with size 200 × 200 at 1.04 s. Such a quick and accurate approach has practical value in medical research.

Reliability and Reproducibility of Analyzing 3D Transperineal Ultrasound Volumes Obtained in the First Phase of Labor - A Pilot Study.

The aim of this study was to investigate the reliability and reproducibility of transperineal ultrasound (TPUS) in the initial phase of labor. As TPUS is a common method, it could supplement vaginal palpation and even replace it in certain situations. In addition, we used a 4-dimensional method for the assessment of cervical effacement. 54 women in labor were included and underwent TPUS. The resulting images from the acquired 4D volumes were evaluated after the examination for the first time and a second time after 21 days. The measured values were cervical length, dilatation and effacement, the angle of progression (AoP), and head-perineum distance. 54 patients were examined. TPUS images were unable to be evaluated in 12 patients because of cervical dilatation of more than 5 cm or poor image quality. Thus, 42 measurements were included. The concordance correlation coefficients according to Lin are satisfactory overall, with one exception for cervical effacement. The accuracy component of cervical length (CCCLin: 0.93; accuracy: 1.00), dilatation (CCCLin: 0.93; accuracy: 1.00), and AoP (CCCLin: 0.87; accuracy: 1.00) is excellent and still high for the head-perineum distance (CCCLin: 0.89; accuracy: 0.96) and cervical effacement (CCCLin: 0.77; accuracy: 0.97). TPUS is a valuable noninvasive tool with good diagnostic accuracy for the AoP, cervical length, and dilatation. Our study provides support for the use of TPUS to complement a vaginal examination. It should not replace a digital examination but should serve as a suitable alternative method for monitoring labor progression in the future.

Assessment of thoracic spinal curvatures in static postures using spatially tracked 3D ultrasound volumes: a proof-of-concept study

The assessment of spinal posture is a difficult endeavour given the lack of identifiable bony landmarks for placement of skin markers. Moreover, potentially significant soft tissue artefacts along the spine further affect the accuracy of marker-based approaches. The objective of this proof-of-concept study was to develop an experimental framework to assess spinal postures by using three-dimensional (3D) ultrasound (US) imaging. A phantom spine model immersed in water was scanned using 3D US in a neutral and two curved postures mimicking a forward flexion in the sagittal plane while the US probe was localised by three electromagnetic tracking sensors attached to the probe head. The obtained anatomical ‘coarse’ registrations were further refined using an automatic registration algorithm and validated by an experienced sonographer. Spinal landmarks were selected in the US images and validated against magnetic resonance imaging data of the same phantom through image registration. Their position was then related to the location of the tracking sensors identified in the acquired US volumes, enabling the localisation of landmarks in the global coordinate system of the tracking device. Results of this study show that localised 3D US enables US-based anatomical reconstructions comparable to clinical standards and the identification of spinal landmarks in different postures of the spine. The accuracy in sensor identification was 0.49 mm on average while the intra- and inter-observer reliability in sensor identification was strongly correlated with a maximum deviation of 0.8 mm. Mapping of landmarks had a small relative distance error of 0.21 mm (SD = ± 0.16) on average. This study implies that localised 3D US holds the potential for the assessment of full spinal posture by accurately and non-invasively localising vertebrae in space.

3D Echocardiogram Reconstruction Employing a Flip Directional Texture Pyramid

Three dimensional (3D) echocardiogram enables cardiologists to visualize suspicious cardiac structures in detail. In recent years, this three-dimensional echocardiogram carries important clinical value in virtual surgical simulation. However, this 3D echocardiogram involves a trade-off difficulty between accuracy and efficient computation in clinical diagnosis. This paper presents a novel Flip Directional 3D Volume Reconstruction (FD-3DVR) method for the reconstruction of echocardiogram images. The proposed method consists of two main steps: multiplanar volumetric imaging and 3D volume reconstruction. In the creation of multiplanar volumetric imaging, two-dimensional (2D) image pixels are mapped into voxels of the volumetric grid. As the obtained slices are discontinuous, there are some missing voxels in the volume data. To restore the structural and textural information of 3D ultrasound volume, the proposed method creates a volume pyramid in parallel with the flip directional texture pyramid. Initially, the nearest neighbors of missing voxels in the multiplanar volumetric imaging are identified by 3D ANN (Approximate Nearest Neighbor) patch matching method. Furthermore, a flip directional texture pyramid is proposed and aggregated with distance in patch matching to find out the most similar neighbors. In the reconstruction step, structural and textural information obtained from different flip angle directions can reconstruct 3D volume well with the desired accuracy. Compared with existing 3D reconstruction methods, the proposed Flip Directional 3D Volume Reconstruction (FD-3DVR) method provides superior performance for the mean peak signal-to-noise ratio (40.538 for the proposed method I and 39.626 for the proposed method II). Experimental results performed on the cardiac datasets demonstrate the efficiency of the proposed method for the reconstruction of echocardiogram images.

Automatic Plane of Minimal Hiatal Dimensions Extraction From 3D Female Pelvic Floor Ultrasound.

There is an increasing interest in the applications of 3D ultrasound imaging of the pelvic floor to improve the diagnosis, treatment, and surgical planning of female pelvic floor dysfunction (PFD). Pelvic floor biometrics are obtained on an oblique image plane known as the plane of minimal hiatal dimensions (PMHD). Identifying this plane requires the detection of two anatomical landmarks, the pubic symphysis and anorectal angle. The manual detection of the anatomical landmarks and the PMHD in 3D pelvic ultrasound requires expert knowledge of the pelvic floor anatomy, and is challenging, time-consuming, and subject to human error. These challenges have hindered the adoption of such quantitative analysis in the clinic. This work presents an automatic approach to identify the anatomical landmarks and extract the PMHD from 3D pelvic ultrasound volumes. To demonstrate clinical utility and a complete automated clinical task, an automatic segmentation of the levator-ani muscle on the extracted PMHD images was also performed. Experiments using 73 test images of patients during a pelvic muscle resting state showed that this algorithm has the capability to accurately identify the PMHD with an average Dice of 0.89 and an average mean boundary distance of 2.25mm. Further evaluation of the PMHD detection algorithm using 35 images of patients performing pelvic muscle contraction resulted in an average Dice of 0.88 and an average mean boundary distance of 2.75mm. This work had the potential to pave the way towards the adoption of ultrasound in the clinic and development of personalized treatment for PFD.

Robotic system for accurate percutaneous puncture guided by 3D-2D ultrasound.

Ultrasound (US)-guided robotic systems can reduce the reliance on the experience and skills of surgeons and enable automatic and accurate percutaneous puncture. Two-dimensional (2D) and three-dimensional (3D) US guidance have various advantages and disadvantages. The planned puncture path in the US data directly affects the puncture quality and tissue injury risk. It is difficult to define the optimal path in 2D US images and achieve accurate and safe puncture under the guidance of 3D US volume. This study aims to propose a robotic system guided by 3D-2D US to realize accurate and safe percutaneous puncture. We proposed a 3D-2D US-guided percutaneous puncture robotic system by integrating a 3D US scanning robotic system and a 2D US-guided puncture robotic system. The optimal spatial puncture path that targets the lesion and avoids other important tissues was determined in the 3D US volume reconstructed through robotic US scanning. Thereafter, the puncture robot was placed at the puncture site determined according to the planned path. The optimal path was mapped to the 2D US image taken at the puncture site. Finally, the 2D US image and puncture path were used to guide the robot in performing an accurate and safe percutaneous puncture. The proposed robotic system based on the guidance of 3D-2D US exhibits the advantages of both 3D US and 2D US to improve the accuracy of percutaneous puncture and reduce the risk of tissue injury. The experimental results of phantom puncture demonstrate that the mean puncture accuracy of the system is 1.09 ± 0.35mm, and the puncture success rate with single needle insertion is 100%. A percutaneous puncture robotic system based on 3D-2D US guidance was proposed and tested successfully. The experimental results demonstrate the feasibility of the proposed system for achieving accurate and safe robotic percutaneous puncture.

A deep learning method for kidney segmentation in 2D ultrasound images.

Ultrasound (US) is a medical imaging modality widely used for diagnosis, monitoring, and guidance of surgical procedures. However, the accurate interpretation of US images is a challenging task. Recently, portable 2D US devices enhanced with Artificial intelligence (AI) methods to identify, in real-time, specific organs are widely spreading worldwide. Nevertheless, the number of available methods that effectively work in such devices is still limited. In this work, we evaluate the performance of the U-NET architecture to segment the kidney in 2D US images. To accomplish this task, we studied the possibility of using multiple sliced images extracted from 3D US volumes to achieve a large, variable, and multi-view dataset of 2D images. The proposed methodology was tested with a dataset of 66 3D US volumes, divided in 51 for training, 5 for validation, and 10 for testing. From the volumes, 3792 2D sliced images were extracted. Two experiments were conducted, namely: (i) using the entire database (WWKD); and (ii) using images where the kidney area is > 500 mm2 (500KD). As a proof-of-concept, the potential of our strategy was tested in real 2D images (acquired with 2D probes). An average error of 2.88 ± 2.63 mm in the testing dataset was registered. Moreover, satisfactory results were obtained in our initial proof-of-concept using pure 2D images. In short, the proposed method proved, in this preliminary study, its potential interest for clinical practice. Further studies are required to evaluate the real performance of the proposed methodology. Clinical Relevance- In this work a deep learning methodology to segment the kidney in 2D US images is presented. It may be a relevant feature to be included in the recent portable US ecosystems easing the interpretation of image and consequently the clinical analysis.

Deep learning-based plane pose regression in obstetric ultrasound

PurposeIn obstetric ultrasound (US) scanning, the learner’s ability to mentally build a three-dimensional (3D) map of the fetus from a two-dimensional (2D) US image represents a major challenge in skill acquisition. We aim to build a US plane localisation system for 3D visualisation, training, and guidance without integrating additional sensors.MethodsWe propose a regression convolutional neural network (CNN) using image features to estimate the six-dimensional pose of arbitrarily oriented US planes relative to the fetal brain centre. The network was trained on synthetic images acquired from phantom 3D US volumes and fine-tuned on real scans. Training data was generated by slicing US volumes into imaging planes in Unity at random coordinates and more densely around the standard transventricular (TV) plane.ResultsWith phantom data, the median errors are 0.90 mm/1.17^circ and 0.44 mm/1.21^circ for random planes and planes close to the TV one, respectively. With real data, using a different fetus with the same gestational age (GA), these errors are 11.84 mm/25.17^circ . The average inference time is 2.97 ms per plane.ConclusionThe proposed network reliably localises US planes within the fetal brain in phantom data and successfully generalises pose regression for an unseen fetal brain from a similar GA as in training. Future development will expand the prediction to volumes of the whole fetus and assess its potential for vision-based, freehand US-assisted navigation when acquiring standard fetal planes.

Impact of Pushing Timing on Occult Injury of Levator Ani: a Multicenter Randomized Controlled Trial

The impact of the timing of pushing during the second stage of labor on pelvic floor disorders is unknown. We sought to determine whether immediate or delayed pushing is associated with an increased proportion of injury to the levator ani after a first delivery among nulliparous women. The Optimizing Management of the Second Stage (OMSS) study was a multicenter randomized trial. Nulliparous women with term pregnancies and neuraxial analgesia were randomly assigned at complete cervical dilation to either immediate pushing or delay for 1 hour. Consenting participants underwent longitudinal pelvic floor assessments (PFA): 1) during postpartum stay (baseline); 2) at 6 weeks (PP1); and 3) at 6 months (PP2) postpartum. Eligible participants underwent transperineal 3D ultrasounds at each visit. Following the completion of all visits by all subjects, saved 3D ultrasound volumes were assessed in a masked fashion. The outcome was complete (but “occult”) levator ani muscle (LAM) injury on the right or left, defined as a widening of the attachment of the levator ani (beyond a set threshold in three contiguous slices) to its origin (Figure) utilizing the Levator Urethra Gap (LUG) measurement. The impact of an alternative threshold LUG distance was explored. Measurements and proportions were compared between the two groups. Of the 2414 subjects randomized, 941 (39.0%) participated in the PFA; 452 in the immediate pushing and 489 in the delayed pushing groups. Ultrasounds were performed on 67%, 83%, and 77% of the PFA participants at the baseline, PP1 and PP2 visits respectively. Demographic and labor characteristics (previously reported) were comparable between the 2 groups; 94% were non-Hispanic and 50% were Black participants. At all 3 timepoints, there were no differences in LAM injury (Table). These findings did not change when the threshold (sensitivity) of LAM injury was adjusted. Among nulliparous women at term with neuraxial analgesia, neither immediate nor delayed pushing in the second stage resulted in significant differences in occult LAM injury.View Large Image Figure ViewerDownload Hi-res image Download (PPT)

Centroid-based Distance Loss Function for Lamina Segmentation in 3D Ultrasound Spine Volumes.

Ultrasound imaging of the spine to diagnose the severity of scoliosis is a recent development in the field, offering 3D information that does not require a complicated procedure of reconstruction, unlike with radiography. Determining the severity of scoliosis on ultrasound volumes requires labelling vertebral features called laminae. To increase accuracy and reduce time spent on this task, this paper reported a novel custom centroid-based distance loss function for lamina segmentation in 3D ultrasound volumes, using convolutional neural networks (CNN). A comparison between the custom and two standard loss functions was performed by fitting a CNN with each loss function. The results showed that the custom loss network performed the best in terms of minimization of the distances between the centroids in the ground truth and the centroids in the predicted segmentation. On average, the custom network improved on the total distance between predicted and true centroids by 33 voxels (22%) when compared with the second best performing network, which used the Dice loss. In general, this novel custom loss function allowed the network to detect two more laminae on average in the lumbar region of the spine that the other networks tended to miss.

Unsupervised Deep Learning based Longitudinal Follicular Growth Tracking during IVF Cycle using 3D Transvaginal Ultrasound in Assisted Reproduction.

Longitudinal follicle tracking is needed in clinical practice for diagnosis and management in assisted reproduction. Follicles are tracked over the in-vitro fertilization (IVF) cycle, and this analysis is usually performed manually by a medical practitioner. It is a challenging manual analysis and is prone to error as it is largely operator dependent. In this paper we propose a two-stage framework to address the clinical need for follicular growth tracking. The first stage comprises of an unsupervised deep learning network SFR-Net to automate registration of each and every follicle across the IVF cycle. SFR-Net is composed of the standard 3DUNet [1] and Multi-Scale Residual Blocks (MSRB) [2] in order to register follicles of varying sizes. In the second stage we use the registration result to track individual follicles across the IVF cycle. The 3D Transvaginal Ultrasound (3D TVUS) volumes were acquired from 26 subjects every 2-3 days, resulting in a total of 96 volume pairs for the registration and tracking task. On the test dataset we have achieved an average DICE score of 85.84% for the follicle registration task, and we are successfully able to track follicles above 4 mm. Ours is the novel attempt towards automated tracking of follicular growth [3].Clinical Relevance- Accurate tracking of follicle count and growth is of paramount importance to increase the effectiveness of IVF procedure. Correct predictions can help doctors provide better counselling to the patients and individualize treatment for ovarian stimulation. Favorable outcome of this assisted reproductive technique depends on the estimates of the quality and quantity of the follicular pool. Therefore, automated longitudinal tracking of follicular growth is highly demanded in Assisted Reproduction clinical practice. [4].

Deformation-Aware Robotic 3D Ultrasound

Tissue deformation in ultrasound (US) imaging leads to geometrical errors when measuring tissues due to the pressure exerted by probes. Such deformation has an even larger effect on 3D US volumes as the correct compounding is limited by the inconsistent location and geometry. This work proposes a patient-specified stiffness-based method to correct the tissue deformations in robotic 3D US acquisitions. To obtain the patient-specified model, robotic palpation is performed at sampling positions on the tissue. The contact force, US images and the probe poses of the palpation procedure are recorded. The contact force and the probe poses are used to estimate the nonlinear tissue stiffness. The images are fed to an optical flow algorithm to compute the pixel displacement. Then the pixel-wise tissue deformation under different forces is characterized by a coupled quadratic regression. To correct the deformation at unseen positions on the trajectory for building 3D volumes, an interpolation is performed based on the stiffness values computed at the sampling positions. With the stiffness and recorded force, the tissue displacement could be corrected. The method was validated on two blood vessel phantoms with different stiffness. The results demonstrate that the method can effectively correct the force-induced deformation and finally generate 3D tissue geometries

Reference standards for facial measurements in early third trimester South African fetuses, and the effect of maternal and fetal characteristics

abstract Fetal alcohol spectrum disorder (FASD) is a major problem worldwide and dysmorphic facial features may be a prenatal biomarker for FASD. Deviations from normal facial development cannot be explored before establishing the normal variation in a specific population, since ethnic differences may exist. Objectives: Main objective: to establish reference standards for 23 facial measurements on 3D ultrasound volumes obtained between days 196 and 224 of gestation in healthy unexposed South African fetuses from an area with historically high alcohol consumption prevalence and a population group with no existing normative values. Secondary objective: to assess the confounding effect of maternal and fetal characteristics. Design: This study involves 97 women (including 43 smokers) who had been enrolled in the Safe Passage Study (SPS), a large prospective multinational cohort study assessing the effects of prenatal alcohol exposure. They had adequate 3 D ultrasound volumes of the fetal face acquired at 28+0–31+6 weeks in singleton pregnancies without comorbidities, congenital abnormalities or exposure to alcohol, marijuana, or methamphetamines from 4 weeks before conception. Participants, materials, setting, methods: The participants were recruited from two residential areas of low socioeconomic status in Cape Town. Meticulous information was collected on maternal and pregnancy characteristics, including alcohol use at different time points. Gestational age (GA) was based on ultrasound biometry before 24 weeks, and 3D ultrasound volumes were acquired trans-abdominally from a sagittal and axial plane of the fetal face. Volumes were independently assessed offline by two observers and the image with the best landmark definition was used for 23 facial measurements, representing features previously described in children with FASD. The relation to the exact GA was assessed by regression analysis, the expected mean value and standard error of the estimate (SEE) was determined to transform all raw measurements into z-scores, and the effect of possible confounders on z-scores was assessed by ANOVA. Results: Ten variables changed significantly with advancing GA (extraocular diameter, anteroposterior, medio-lateral and supero-inferior ocular diameter, ocular volume, interlens distance, prenasal thickness, nasal bone length, nose length and nose protrusion) and thirteen did not (interocular distance; interocular: extraocular diameter ratio, prenasal thickness: nasal bone length ratio, pronasal-subnasal distance, subnasal-mouth distance, philtrum length, upper vermillion thickness, nose-philtrum angle, maxillary angle, facial height, facial protrusion, frontomaxillary facial angle and maxilla-nasion-mandible angle). Reference values (expected mean and SEE) for the 23 measurements were established for each day. The z-scores of all facial measurements were not independently affected by maternal age, parity, gravidity, smoking or body mass index, but infant sex and birthweight z-score significantly influenced several z-scores (infant sex for extraocular, medio-lateral, and supero-inferior ocular diameter, ocular volume, prenasal thickness and nose protrusion; birthweight z-score for extraocular diameter, interocular and interlens distance, nose protrusion and maxillary angle). Limitations: GA was not always confirmed by first trimester ultrasound and some measurements could not be obtained in all cases due to suboptimal image quality. The cohort included few heavy smokers so an effect of heavy or continued smoking cannot be ruled out, and the effect of ethnicity was not assessed. Conclusions: These are the first local reference standards for fetal facial measurements and, to our knowledge, the first reference standards for the supero-inferior ocular diameter, face protrusion, upper vermillion thickness, maxillary angle, and nose-philtrum angle. They were broadly in keeping with published references, with small discrepancies explained by minor differences in technique. Even in this narrow GA window, the distribution of many variables changed over time and normal variation was significantly influenced by fetal sex and birthweight z-score. The possible confounding effect of these factors needs to be considered when assessing the impact of harmful exposures like alcohol on facial development.

Hole-filling based on content loss indexed 3D partial convolution network for freehand ultrasound reconstruction

Background and objectiveDuring the 3D reconstruction of ultrasound volume from 2D B-scan ultrasound images, holes are usually found in the reconstructed 3D volumes due to the fast scans. This condition will affect the positioning and judgment of the doctor to the lesion. Hence, in this study, we propose to fill the holes by using a novel content loss indexed 3D partial convolution network for 3D freehand ultrasound volume reconstruction. The network can synthesize novel ultrasound volume structures and reconstruct ultrasound volume with missing regions with variable sizes and at arbitrary locations. MethodsFirst, the 3D partial convolution is introduced into the convolutional layer, which is masked and renormalized to be conditioned on only valid voxels. Then, the mask in the next layer is automatically updated as a part of the forward pass. To better preserve texture and structure details of the reconstruction results, we couple the adversarial loss of the least squares generative adversarial network (LSGAN) with the innovative content loss, which consists of the context loss, the feature-matching loss and the total variation loss. Thereafter, we introduce a novel spectral-normalized LSGAN by adding spectral normalization (SN) to the generator and discriminator of the LSGAN. The proposed method is simple in formulation, and is stable in training. ResultsExperiments on public and in-vivo ultrasound datasets and comparisons with popular algorithms demonstrate that the proposed approach can generate high-quality hole-filling results with preserved perceptual image details. ConclusionsConsidering the high quality of the hole-filling results, the proposed method can effectively fill the missing regions in the reconstructed 3D ultrasound volume from 2D ultrasound image sequences.

Intra-operative resection margin model of tongue carcinoma using 3D reconstructed ultrasound

BackgroundAlthough an oncological surgeon aims to remove tongue cancer with a margin of no less than 5 mm, 85% of margins are reported as close or positive (<5 mm). Currently, the only feedback on the status of margins is provided by frozen sections and standard post-operative histopathological assessment. ObjectivesTo introduce a novel method for assessing intra-operative margins that includes intra-operative visual feedback by means of 3D ultrasound and to assess its feasibility. MethodsElectromagnetically tracked ultrasound imaging was used to scan tongue cancer (n = 8) directly after excision. Each specimen was fully submerged in saline during acquisition and scanned without contact between tissue and transducer. This imaging technique enables rapid generation of 3D margin models. A radiologist manually segmented the specimen and tumour intra-operatively in the 3D ultrasound volumes. The margins were computed in 3D by calculating the closest Euclidean distances between the segmented specimen and tumour boundary. ResultsUsing a colourmap, our proposed 3D ultrasound generated a 3D model representing the margins of the entire specimen. This model was easy to interpret and provided the surgeon with feedback on the margins intra-operatively. ConclusionsElectromagnetically tracked ultrasound imaging enables assessment of the margins of tongue cancer of the entire specimen. Future research will have to focus on improving the image reconstruction quality, the development of fully automated segmentation of tumours in 3D ultrasound and the correlation between the margins assessed by 3D ultrasound and histopathology.

Good and bad boundaries in ultrasound compounding: preserving anatomic boundaries while suppressing artifacts

PurposeUltrasound compounding is to combine sonographic information captured from different angles and produce a single image. It is important for multi-view reconstruction, but as of yet there is no consensus on best practices for compounding. Current popular methods inevitably suppress or altogether leave out bright or dark regions that are useful and potentially introduce new artifacts. In this work, we establish a new algorithm to compound the overlapping pixels from different viewpoints in ultrasound.MethodsInspired by image fusion algorithms and ultrasound confidence, we uniquely leverage Laplacian and Gaussian pyramids to preserve the maximum boundary contrast without overemphasizing noise, speckles, and other artifacts in the compounded image, while taking the direction of the ultrasound probe into account. Besides, we designed an algorithm that detects the useful boundaries in ultrasound images to further improve the boundary contrast.ResultsWe evaluate our algorithm by comparing it with previous algorithms both qualitatively and quantitatively, and we show that our approach not only preserves both light and dark details, but also somewhat suppresses noise and artifacts, rather than amplifying them. We also show that our algorithm can improve the performance of downstream tasks like segmentation.ConclusionOur proposed method that is based on confidence, contrast, and both Gaussian and Laplacian pyramids appears to be better at preserving contrast at anatomic boundaries while suppressing artifacts than any of the other approaches we tested. This algorithm may have future utility with downstream tasks such as 3D ultrasound volume reconstruction and segmentation.

Development of a Standardized Semantic Feature-Based Reporting Proforma for Intraoperative Ultrasound Findings in Brain Tumors and Application in High-Grade Gliomas - A Preliminary Study.

Purpose A semantic feature-based reporting proforma for intraoperative ultrasound findings in brain tumors was devised to standardize reporting. It was applied as a pilot study on a cohort of histologically confirmed high-grade supratentorial gliomas (Grade 3 and 4) for internal validation. Materials and Methods This intraoperative semantic ultrasound proforma was used to evaluate 3D ultrasound volumes using Radiant DICOM software by 3 surgeons. The ultrasound semantic features were correlated with histological features like tumor grade, IDH status, and MIB index. Results 68 patients were analyzed using the semantic proforma. Irregular crenated was the most common margin (63.2%) and lesions were heterogeneously hyperechoic (95.6%). Necrosis was commonly seen and noted as single (67.6%) or multiple (13.2%) in over 80% cases. A separate perilesional zone, which was predominantly hyperechoic in 41.8% and both hypo and hyperechoic in 12.7%, could be identified in 54.5% of cases. Grade 4 tumors were more likely to have an irregular crenated margin (71.2%) with a single large area of necrosis, while Grade 3 tumors were likely to have smooth (31.3%) or non-characterizable margins (31.2%) with no or multiple areas of necrosis. IDH-negative tumors were more likely to have a single large focus of necrosis. Among the GBMs (52 cases), MIB labelling index of>15% was associated with poorly delineated, uncharacterizable margins, when compared with MIB labelling index<15% (23.5 vs. 0%), (p=0.046). Conclusion A detailed semantic proforma was developed for brain tumors and was internally validated. A few ultrasound sematic features were identified correlating with histological features in high-grade gliomas. It will require further external validation for refinement and acceptability.