Prior-knowledge Embedded U-Net based Fully Automatic Vessel Wall Volume Measurement of the Carotid Artery in 3D Ultrasound Image.

The vessel-wall-volume (VWV) measured based on three-dimensional (3D) carotid artery (CA) ultrasound (US) images can help to assess carotid atherosclerosis and manage patients at risk for stroke. Manual involvement for measurement work is subjective and requires well-trained operators, and fully automatic measurement tools are not yet available. Thereby, we proposed a fully automatic VWV measurement framework (Auto-VWV) using a CA prior-knowledge embedded U-Net (CAP-UNet) to measure the VWV from 3D CA US images without manual intervention. The Auto-VWV framework is designed to improve the repeated VWV measuring consistency, which resulted in the first fully automatic framework for VWV measurement. CAP-UNet is developed to improve segmentation accuracy on the whole CA, which composed of a U-Net type backbone and three additional prior-knowledge learning modules. Specifically, a continuity learning module is used to learn the spatial continuity of the arteries in a sequence of image slices. A voxel evolution learning module was designed to learn the evolution of the artery in adjacent slices, and a topology learning module was used to learn the unique topology of the carotid artery. In two 3D CA US datasets, CAP-UNet architecture achieved state-of-the-art performance compared to eight competing models. Furthermore, CAP-UNet-based Auto-VWV achieved better accuracy and consistency than Auto-VWV based on competing models in the simulated repeated measurement. Finally, using 10 pairs of real repeatedly scanned samples, Auto-VWV achieved better VWV measurement reproducibility than intra- and inter-operator manual measurements.

Three-dimensional trans-rectal and trans-abdominal ultrasound image fusion for the guidance of gynecologic brachytherapy procedures: a proof of concept study.

High dose-rate brachytherapy is a treatment technique for gynecologic cancers where intracavitary applicators are placed within the patient's pelvic cavity. To ensure accurate radiation delivery, localization of the applicator at the time of insertion is vital. This study proposes a novel method for acquiring, registering, and fusing three-dimensional (3D) trans-abdominal and 3D trans-rectal ultrasound (US) images for visualization of the pelvic anatomy and applicators during gynecologic brachytherapy. The workflow was validated using custom multi-modal pelvic phantoms and demonstrated during two patient procedures. Experiments were performed for three types of intracavitary applicators: ring-and-tandem, ring-and-tandem with interstitial needles, and tandem-and-ovoids. Fused 3D US images were registered to magnetic resonance (MR) and computed tomography (CT) images for validation. The target registration error (TRE) and fiducial localization error (FLE) were calculated to quantify the accuracy of our fusion technique. For both phantom and patient images, TRE and FLE across all modality registrations (3D US versus MR or CT) resulted in mean ± standard deviation of 4.01 ± 1.01 mm and 0.43 ± 0.24 mm, respectively. This work indicates proof of concept for conducting further clinical studies leveraging 3D US imaging as an accurate, accessible alternative to advanced modalities for localizing brachytherapy applicators.

Machine learning and magnetic resonance image texture analysis predicts accelerated lung function decline in ex-smokers with and without chronic obstructive pulmonary disease.

Purpose: Our objective was to train machine-learning algorithms on hyperpolarized He 3 magnetic resonance imaging (MRI) datasets to generate models of accelerated lung function decline in participants with and without chronic-obstructive-pulmonary-disease. We hypothesized that hyperpolarized gas MRI ventilation, machine-learning, and multivariate modeling could be combined to predict clinically-relevant changes in forced expiratory volume in 1 s ( FEV 1 ) across 3 years. Approach: Hyperpolarized He 3 MRI was acquired using a coronal Cartesian fast gradient recalled echo sequence with a partial echo and segmented using a k-means clustering algorithm. A maximum entropy mask was used to generate a region-of-interest for texture feature extraction using a custom-developed algorithm and the PyRadiomics platform. The principal component and Boruta analyses were used for feature selection. Ensemble-based and single machine-learning classifiers were evaluated using area-under-the-receiver-operator-curve and sensitivity-specificity analysis. Results: We evaluated 88 ex-smoker participants with 31 ± 7 months follow-up data, 57 of whom (22 females/35 males, 70 ± 9 years) had negligible changes in FEV 1 and 31 participants (7 females/24 males, 68 ± 9 years) with worsening FEV 1 ≥ 60 mL / year . In addition, 3/88 ex-smokers reported a change in smoking status. We generated machine-learning models to predict FEV 1 decline using demographics, spirometry, and texture features, with the later yielding the highest classification accuracy of 81%. The combined model (trained on all available measurements) achieved the overall best classification accuracy of 82%; however, it was not significantly different from the model trained on MRI texture features alone. Conclusion: For the first time, we have employed hyperpolarized He 3 MRI ventilation texture features and machine-learning to identify ex-smokers with accelerated decline in FEV 1 with 82% accuracy.

Semi-supervised learning framework with shape encoding for neonatal ventricular segmentation from 3D ultrasound.

BACKGROUND: Three-dimensional (3D) ultrasound (US) imaging has shown promise in non-invasive monitoring of changes in the lateral brain ventricles of neonates suffering from intraventricular hemorrhaging. Due to the poorly defined anatomical boundaries and low signal-to-noise ratio, fully supervised methods for segmentation of the lateral ventricles in 3D US images require a large dataset of annotated images by trained physicians, which is tedious, time-consuming, and expensive. Training fully supervised segmentation methods on a small dataset may lead to overfitting and hence reduce its generalizability. Semi-supervised learning (SSL) methods for 3D US segmentation may be able to address these challenges but most existing SSL methods have been developed for magnetic resonance or computed tomography (CT) images. PURPOSE: To develop a fast, lightweight, and accurate SSL method, specifically for 3D US images, that will use unlabeled data towards improving segmentation performance. METHODS: We propose an SSL framework that leverages the shape-encoding ability of an autoencoder network to enforce complex shape and size constraints on a 3D U-Net segmentation model. The autoencoder created pseudo-labels, based on the 3D U-Net predicted segmentations, that enforces shape constraints. An adversarial discriminator network then determined whether images came from the labeled or unlabeled data distributions. We used 887 3D US images, of which 87 had manually annotated labels and 800 images were unlabeled. Training/validation/testing sets of 25/12/50, 25/12/25 and 50/12/25 images were used for model experimentation. The Dice similarity coefficient (DSC), mean absolute surface distance (MAD), and absolute volumetric difference (VD) were used as metrics for comparing to other benchmarks. The baseline benchmark was the fully supervised vanilla 3D U-Net while dual task consistency, shape-aware semi-supervised network, correlation-aware mutual learning, and 3D U-Net Ensemble models were used as state-of-the-art benchmarks with DSC, MAD, and VD as comparison metrics. The Wilcoxon signed-rank test was used to test statistical significance between algorithms for DSC and VD with the threshold being p < 0.05 and corrected to p < 0.01 using the Bonferroni correction. The random-access memory (RAM) trace and number of trainable parameters were used to compare the computing efficiency between models. RESULTS: Relative to the baseline 3D U-Net model, our shape-encoding SSL method reported a mean DSC improvement of 6.5%, 7.7%, and 4.1% with a 95% confidence interval of 4.2%, 5.7%, and 2.1% using image data splits of 25/12/50, 25/12/25, and 50/12/25, respectively. Our method only used a 1GB increase in RAM compared to the baseline 3D U-Net and required less than half the RAM and trainable parameters compared to the 3D U-Net ensemble method. CONCLUSIONS: Based on our extensive literature survey, this is one of the first reported works to propose an SSL method designed for segmenting organs in 3D US images and specifically one that incorporates unlabeled data for segmenting neonatal cerebral lateral ventricles. When compared to the state-of-the-art SSL and fully supervised learning methods, our method yielded the highest DSC and lowest VD while being computationally efficient.

JoCoRank: Joint correlation learning with ranking similarity regularization for imbalanced fetal brain age regression.

Estimating fetal brain age based on sulci by magnetic resonance imaging (MRI) is clinically crucial in determining the normal development of fetal brains. Deep learning provides a possible way for fetal brain age estimation using MRI. Previous studies have mainly emphasized optimizing individual-wise correlation criteria, such as mean square error. However, they ignored the very important global and peer-wise criterion, which are essential for learning the structured relationships among regression samples. Moreover, the imbalanced label distribution introduces an adverse bias, which impairs the reliability and interpretation of correlation estimation and the model's fairness and generalizability. In this work, we propose a novel joint correlation learning with ranking similarity regularization (JoCoRank) algorithm for deep imbalanced regression of fetal brain age. Joint correlation learning concurrently captures individual, global, and peer-level valuable relationship information, and the customized optimization scheme for each criterion exhibits strong robustness against outliers and imbalanced regression. Ranking similarity regularization is designed to calibrate the biased feature representations by aligning the sorted list of neighbors in the label space with those in the feature space. A total of 1327 MRI images from 157 healthy fetuses between 22 and 34 weeks were collected at Wuhan Children's Hospital and utilized to evaluate the performance of JoCoRank in fetal brain age estimation. JoCoRank achieved promising results with an average mean absolute error of 0.693±0.064 weeks and R2 coefficient of 0.930±0.019. Our fetal brain age estimation algorithm would be useful for identifying abnormalities in fetal brain development and undertaking early intervention in clinical practice.

An image registration-based self-supervised Su-Net for carotid plaque ultrasound image segmentation.

BACKGROUND AND OBJECTIVES: Total Plaque Area (TPA) measurement is critical for early diagnosis and intervention of carotid atherosclerosis in individuals with high risk for stroke. The delineation of the carotid plaques is necessary for TPA measurement, and deep learning methods can automatically segment the plaque and measure TPA from carotid ultrasound images. A large number of labeled images is essential for training a good deep learning model, but it is very difficult to collect such large labeled datasets for carotid image segmentation in clinical practice. Self-supervised learning can provide a possible solution to improve the deep-learning models on small labeled training datasets by designing a pretext task to pre-train the models without using the segmentation masks. However, the existing self-supervised learning methods do not consider the feature presentations of object contours. METHODS: In this paper, we propose an image registration-based self-supervised learning method and a stacked U-Net (SSL-SU-Net) for carotid plaque ultrasound image segmentation, which can better exploit the semantic features of carotid plaque contours in self-supervised task training. RESULTS: Our network was trained on different numbers of labeled images (n = 10, 33, 50 and 100 subjects) and tested on 44 subjects from the SPARC dataset (n = 144, London, Canada). The network trained on the entire SPARC dataset was then directly applied to an independent dataset collected in Zhongnan hospital (n = 497, Wuhan, China). For the 44 subjects tested on the SPARC dataset, our method yielded a DSC of 80.25-89.18% and the produced TPA measurements, which were strongly correlated with manual segmentation (r = 0.965-0.995, ρ< 0.0001). For the Zhongnan dataset, the DSC was 90.3% and algorithm TPAs were strongly correlated with manual TPAs (r = 0.985, ρ< 0.0001). CONCLUSIONS: The results demonstrate that our proposed method yielded excellent performance and good generalization ability when trained on a small labeled dataset, facilitating the use of deep learning in carotid ultrasound image analysis and clinical practice. The code of our algorithm is available https://github.com/a610lab/Registration-SSL.

Three-dimensional ultrasound to investigate synovitis in first carpometacarpal osteoarthritis: A feasibility study.

BACKGROUND: Synovitis is one of the defining characteristics of osteoarthritis (OA) in the carpometacarpal (CMC1) joint of the thumb. Quantitative characterization of synovial volume is important for furthering our understanding of CMC1 OA disease progression, treatment response, and monitoring strategies. In previous studies, three-dimensional ultrasound (3-D US) has demonstrated the feasibility of being a point-of-care system for monitoring knee OA. However, 3-D US has not been tested on the smaller joints of the hand, which presents unique physiological and imaging challenges. PURPOSE: To develop and validate a novel application of 3-D US to monitor soft-tissue characteristics of OA in a CMC1 OA patient population compared to the current gold standard, magnetic resonance imaging (MRI). METHODS: A motorized submerged transducer moving assembly was designed for this device specifically for imaging the joints of the hands and wrist. The device used a linear 3-D scanning approach, where a 14L5 2-D transducer was translated over the region of interest. Two imaging phantoms were used to test the linear and volumetric measurement accuracy of the 3-D US device. To evaluate the accuracy of the reconstructed 3-D US geometry, a multilayer monofilament string-grid phantom (10 mm square grid) was scanned. To validate the volumetric measurement capabilities of the system, a simulated synovial tissue phantom with an embedded synovial effusion was fabricated and imaged. Ten CMC1 OA patients were imaged by our 3-D US and a 3.0 T MRI system to compare synovial volumes. The synovial volumes were manually segmented by two raters on the 2D slices of the 3D US reconstruction and MR images, to assess the accuracy and precision of the device for determining synovial tissue volumes. The Standard Error of Measurement and Minimal Detectable Change was used to assess the precision and sensitivity of the volume measurements. Paired sample t-tests were used to assess statistical significance. Additionally, rater reliability was assessed using Intra-Class Correlation (ICC) coefficients. RESULTS: The largest percent difference observed between the known physical volume of synovial extrusion in the phantom and the volume measured by our 3D US was 1.1% (p-value = 0.03). The mean volume difference between the 3-D US and the gold standard MRI was 1.78% (p-value = 0.48). The 3-D US synovial tissue volume measurements had a Standard Error Measurement (SEm ) of 11.21 mm3 and a Minimal Detectible Change (MDC) of 31.06 mm3 , while the MRI synovial tissue volume measurements had an SEM of 16.82 mm3 and an MDC of 46.63 mm3 . Excellent inter- and intra-rater reliability (ICCs = 0.94-0.99) observed across all imaging modalities and raters. CONCLUSION: Our results indicate the feasibility of applying 3-D US technology to provide accurate and precise CMC1 synovial tissue volume measurements, similar to MRI volume measurements. Lower MDC and SEm values for 3-D US volume measurements indicate that it is a precise measurement tool to assess synovial volume and that it is sensitive to variation between volume segmentations. The application of this imaging technique to monitor OA pathogenesis and treatment response over time at the patient's bedside should be thoroughly investigated in future studies.

Deep learning-based ultrasound auto-segmentation of the prostate with brachytherapy implanted needles.

BACKGROUND: Accurate segmentation of the clinical target volume (CTV) corresponding to the prostate with or without proximal seminal vesicles is required on transrectal ultrasound (TRUS) images during prostate brachytherapy procedures. Implanted needles cause artifacts that may make this task difficult and time-consuming. Thus, previous studies have focused on the simpler problem of segmentation in the absence of needles at the cost of reduced clinical utility. PURPOSE: To use a convolutional neural network (CNN) algorithm for segmentation of the prostatic CTV in TRUS images post-needle insertion obtained from prostate brachytherapy procedures to better meet the demands of the clinical procedure. METHODS: A dataset consisting of 144 3-dimensional (3D) TRUS images with implanted metal brachytherapy needles and associated manual CTV segmentations was used for training a 2-dimensional (2D) U-Net CNN using a Dice Similarity Coefficient (DSC) loss function. These were split by patient, with 119 used for training and 25 reserved for testing. The 3D TRUS training images were resliced at radial (around the axis normal to the coronal plane) and oblique angles through the center of the 3D image, as well as axial, coronal, and sagittal planes to obtain 3689 2D TRUS images and masks for training. The network generated boundary predictions on 300 2D TRUS images obtained from reslicing each of the 25 3D TRUS images used for testing into 12 radial slices (15° apart), which were then reconstructed into 3D surfaces. Performance metrics included DSC, recall, precision, unsigned and signed volume percentage differences (VPD/sVPD), mean surface distance (MSD), and Hausdorff distance (HD). In addition, we studied whether providing algorithm-predicted boundaries to the physicians and allowing modifications increased the agreement between physicians. This was performed by providing a subset of 3D TRUS images of five patients to five physicians who segmented the CTV using clinical software and repeated this at least 1 week apart. The five physicians were given the algorithm boundary predictions and allowed to modify them, and the resulting inter- and intra-physician variability was evaluated. RESULTS: Median DSC, recall, precision, VPD, sVPD, MSD, and HD of the 3D-reconstructed algorithm segmentations were 87.2 [84.1, 88.8]%, 89.0 [86.3, 92.4]%, 86.6 [78.5, 90.8]%, 10.3 [4.5, 18.4]%, 2.0 [-4.5, 18.4]%, 1.6 [1.2, 2.0] mm, and 6.0 [5.3, 8.0] mm, respectively. Segmentation time for a set of 12 2D radial images was 2.46 [2.44, 2.48] s. With and without U-Net starting points, the intra-physician median DSCs were 97.0 [96.3, 97.8]%, and 94.4 [92.5, 95.4]% (p < 0.0001), respectively, while the inter-physician median DSCs were 94.8 [93.3, 96.8]% and 90.2 [88.7, 92.1]%, respectively (p < 0.0001). The median segmentation time for physicians, with and without U-Net-generated CTV boundaries, were 257.5 [211.8, 300.0] s and 288.0 [232.0, 333.5] s, respectively (p = 0.1034). CONCLUSIONS: Our algorithm performed at a level similar to physicians in a fraction of the time. The use of algorithm-generated boundaries as a starting point and allowing modifications reduced physician variability, although it did not significantly reduce the time compared to manual segmentations.

Self-supervised enhanced thyroid nodule detection in ultrasound examination video sequences with multi-perspective evaluation.

Objective.Ultrasound is the most commonly used examination for the detection and identification of thyroid nodules. Since manual detection is time-consuming and subjective, attempts to introduce machine learning into this process are ongoing. However, the performance of these methods is limited by the low signal-to-noise ratio and tissue contrast of ultrasound images. To address these challenges, we extend thyroid nodule detection from image-based to video-based using the temporal context information in ultrasound videos.Approach.We propose a video-based deep learning model with adjacent frame perception (AFP) for accurate and real-time thyroid nodule detection. Compared to image-based methods, AFP can aggregate semantically similar contextual features in the video. Furthermore, considering the cost of medical image annotation for video-based models, a patch scale self-supervised model (PASS) is proposed. PASS is trained on unlabeled datasets to improve the performance of the AFP model without additional labelling costs.Main results.The PASS model is trained by 92 videos containing 23 773 frames, of which 60 annotated videos containing 16 694 frames were used to train and evaluate the AFP model. The evaluation is performed from the video, frame, nodule, and localization perspectives. In the evaluation of the localization perspective, we used the average precision metric with the intersection-over-union threshold set to 50% (AP@50), which is the area under the smoothed Precision-Recall curve. Our proposed AFP improved AP@50 from 0.256 to 0.390, while the PASS-enhanced AFP further improved the AP@50 to 0.425. AFP and PASS also improve the performance in the valuations of other perspectives based on the localization results.Significance.Our video-based model can mitigate the effects of low signal-to-noise ratio and tissue contrast in ultrasound images and enable the accurate detection of thyroid nodules in real-time. The evaluation from multiple perspectives of the ablation experiments demonstrates the effectiveness of our proposed AFP and PASS models.

Cost-effective, portable, patient-dedicated three-dimensional automated breast ultrasound for point-of-care breast cancer screening.

Breast cancer screening has substantially reduced mortality across screening populations. However, a clinical need persists for more accessible, cost-effective, and robust approaches for increased-risk and diverse patient populations, especially those with dense breasts where screening mammography is suboptimal. We developed and validated a cost-effective, portable, patient-dedicated three-dimensional (3D) automated breast ultrasound (ABUS) system for point-of-care breast cancer screening. The 3D ABUS system contains a wearable, rapid-prototype 3D-printed dam assembly, a compression assembly, and a computer-driven 3DUS scanner, adaptable to any commercially available US machine and transducer. Acquisition is operator-agnostic, involves a 40-second scan time, and provides multiplanar 3D visualization for whole-breast assessment. Geometric reconstruction accuracy was evaluated with a 3D grid phantom and tissue-mimicking breast phantoms, demonstrating linear measurement and volumetric reconstruction errors < 0.2 mm and < 3%, respectively. The system's capability was demonstrated in a healthy male volunteer and two healthy female volunteers, representing diverse patient geometries and breast sizes. The system enables comfortable ultrasonic coupling and tissue stabilization, with adjustable compression to improve image quality while alleviating discomfort. Moreover, the system effectively mitigates breathing and motion, since its assembly affixes directly onto the patient. While future studies are still required to evaluate the impact on current clinical practices and workflow, the 3D ABUS system shows potential for adoption as an alternative, cost-effective, dedicated point-of-care breast cancer screening approach for increased-risk populations and limited-resource settings.

Clinical implementation of 3D transvaginal ultrasound for intraoperative guidance of needle implant in template interstitial gynecologic high-dose-rate brachytherapy.

PURPOSE: To demonstrate novel clinical implementation of a 3D transvaginal ultrasound (3DTVUS) system for intraoperative needle insertion guidance in perineal template interstitial gynecologic high-dose-rate brachytherapy and assess its impact on implant quality. METHODS AND MATERIALS: Interstitial implants began with preimplant 3DTVUS to visualize the tumor and anatomy, with intermittent 3DTVUS to assess the implant and guide needle adjustment. Analysis includes visualization of the implant relative to anatomy, identification of cases where 3DTVUS is beneficial, dosimetry, and a survey distributed to 3DTVUS clinicians. RESULTS: Seven patients treated between November 2021 and October 2022 were included in this study. Twenty needles were inserted under 3DTVUS guidance. The tumor and vaginal wall were well-differentiated in four and all seven patients, respectively. Patients with tumours below the superior aspect of the vagina are suited for 3DTVUS. Four radiation oncologists responded to the survey. There was general agreement that 3DTVUS improves implant and anatomy visualization and is preferred over standard 2D ultrasound guidance techniques. CONCLUSIONS: Based on qualitative feedback from primary users and a small preliminary patient cohort, 3DTVUS imaging improves tumor and vaginal wall visualization during gynecologic perineal template interstitial needle implant and is a powerful tool for implant assessment in an intraoperative setting.

3D US-CT/MRI registration for percutaneous focal liver tumor ablations.

PURPOSE: US-guided percutaneous focal liver tumor ablations have been considered promising curative treatment techniques. To address cases with invisible or poorly visible tumors, registration of 3D US with CT or MRI is a critical step. By taking advantage of deep learning techniques to efficiently detect representative features in both modalities, we aim to develop a 3D US-CT/MRI registration approach for liver tumor ablations. METHODS: Facilitated by our nnUNet-based 3D US vessel segmentation approach, we propose a coarse-to-fine 3D US-CT/MRI image registration pipeline based on the liver vessel surface and centerlines. Then, phantom, healthy volunteer and patient studies are performed to demonstrate the effectiveness of our proposed registration approach. RESULTS: Our nnUNet-based vessel segmentation model achieved a Dice score of 0.69. In healthy volunteer study, 11 out of 12 3D US-MRI image pairs were successfully registered with an overall centerline distance of 4.03±2.68 mm. Two patient cases achieved target registration errors (TRE) of 4.16 mm and 5.22 mm. CONCLUSION: We proposed a coarse-to-fine 3D US-CT/MRI registration pipeline based on nnUNet vessel segmentation models. Experiments based on healthy volunteers and patient trials demonstrated the effectiveness of our registration workflow. Our code and example data are publicly available in this r epository.

Cross-sectional reliability and concurrent validity of a quantitative 2-dimensional ultrasound image analysis of effusion and synovial hypertrophy in knee osteoarthritis.

Objective: Effusion-synovitis is related to pain and progression in knee osteoarthritis (OA), but current gold standard ultrasound (US) measures are limited to semi-quantitative grading of joint distension or 1-dimensional thickness measures. A novel quantitative 2-dimensional image analysis methodology is applied to US images of effusion-synovitis; reliability and concurrent validity was assessed in patients with knee OA. Methods: Cross sectional analysis of US images collected from 51 patients with symptomatic knee OA were processed in ImageJ and segmented in 3DSlicer to produce a binary mask of the supra-patellar synovitis region of interest (ROI). Area measures (mm2) of total synovitis, effusion and hypertrophy components were exported. Intra-rater reliability and test-retest reliability (1-14 days washout) were estimated with intra-class correlation coefficients (ICCs). Concurrent validity was measured by Spearman correlations between quantitative measures and gold standard OMERACT and caliper measurements of synovitis. Results: Intra-rater reliability for hypertrophy area was estimated at 0.98, 0.99 for effusion area, and 0.99 for total synovitis area. The test-retest reliability for total synovitis area was 0.63 (SEM 87.8 â€‹mm2), 0.59 for hypertrophy area (SEM 21.0 â€‹mm2), and 0.64 for effusion area (SEM 73.8 â€‹mm2). Correlation between total synovitis area and OMERACT grade was 0.84, 0.81 between total synovitis area and effusion-synovitis calipers, and 0.81 between total effusion area and effusion calipers. Conclusion: This new research tool for image analysis demonstrated excellent intra-rater reliability, good concurrent validity, and moderate test-retest reliability. Quantitative 2D US measures of effusion-synovitis and its individual components may enhance the study and management of knee OA.

Ultrasound carotid plaque segmentation via image reconstruction-based self-supervised learning with limited training labels.

Carotid total plaque area (TPA) is an important contributing measurement to the evaluation of stroke risk. Deep learning provides an efficient method for ultrasound carotid plaque segmentation and TPA quantification. However, high performance of deep learning requires datasets with many labeled images for training, which is very labor-intensive. Thus, we propose an image reconstruction-based self-supervised learning algorithm (IR-SSL) for carotid plaque segmentation when few labeled images are available. IR-SSL consists of pre-trained and downstream segmentation tasks. The pre-trained task learns region-wise representations with local consistency by reconstructing plaque images from randomly partitioned and disordered images. The pre-trained model is then transferred to the segmentation network as the initial parameters in the downstream task. IR-SSL was implemented with two networks, UNet++ and U-Net, and evaluated on two independent datasets of 510 carotid ultrasound images from 144 subjects at SPARC (London, Canada) and 638 images from 479 subjects at Zhongnan hospital (Wuhan, China). Compared to the baseline networks, IR-SSL improved the segmentation performance when trained on few labeled images (n = 10, 30, 50 and 100 subjects). For 44 SPARC subjects, IR-SSL yielded Dice-similarity-coefficients (DSC) of 80.14-88.84%, and algorithm TPAs were strongly correlated (r=0.962-0.993, p < 0.001) with manual results. The models trained on the SPARC images but applied to the Zhongnan dataset without retraining achieved DSCs of 80.61-88.18% and strong correlation with manual segmentation (r=0.852-0.978, p < 0.001). These results suggest that IR-SSL could improve deep learning when trained on small labeled datasets, making it useful for monitoring carotid plaque progression/regression in clinical use and trials.

A power Doppler ultrasound method for improving intraoperative tip localization for visually obstructed needles in interstitial prostate brachytherapy.

PURPOSE: High-dose-rate (HDR) interstitial brachytherapy (BT) is a common treatment technique for localized intermediate to high-risk prostate cancer. Transrectal ultrasound (US) imaging is typically used for guiding needle insertion, including localization of the needle tip which is critical for treatment planning. However, image artifacts can limit needle tip visibility in standard brightness (B)-mode US, potentially leading to dose delivery that deviates from the planned dose. To improve intraoperative tip visualization in visually obstructed needles, we propose a power Doppler (PD) US method which utilizes a novel wireless mechanical oscillator, validated in phantom experiments and clinical HDR-BT cases as part of a feasibility clinical trial. METHODS: Our wireless oscillator contains a DC motor housed in a 3D printed case and is powered by rechargeable battery allowing the device to be operated by one person with no additional equipment required in the operating room. The oscillator end-piece features a cylindrical shape designed for BT applications to fit on top of the commonly used cylindrical needle mandrins. Phantom validation was completed using tissue-equivalent agar phantoms with the clinical US system and both plastic and metal needles. Our PD method was tested using a needle implant pattern matching a standard HDR-BT procedure as well as an implant pattern designed to maximize needle shadowing artifacts. Needle tip localization accuracy was assessed using the clinical method based on ideal reference needles as well as a comparison to computed tomography (CT) as a gold standard. Clinical validation was completed in five patients who underwent standard HDR-BT as part of a feasibility clinical trial. Needle tips positions were identified using B-mode US and PD US with perturbation from our wireless oscillator. RESULTS: Absolute mean ± standard deviation tip error for B-mode alone, PD alone, and B-mode combined with PD was respectively: 0.3 ± 0.3 mm, 0.6 ± 0.5 mm, and 0.4 ± 0.2 mm for the mock HDR-BT needle implant; 0.8 ± 1.7 mm, 0.4 ± 0.6 mm, and 0.3 ± 0.5 mm for the explicit shadowing implant with plastic needles; and 0.5 ± 0.2 mm, 0.5 ± 0.3 mm, and 0.6 ± 0.2 mm for the explicit shadowing implant with metal needles. The total mean absolute tip error for all five patients in the feasibility clinical trial was 0.9 ± 0.7 mm using B-mode US alone and 0.8 ± 0.5 mm when including PD US, with increased benefit observed for needles classified as visually obstructed. CONCLUSIONS: Our proposed PD needle tip localization method is easy to implement and requires no modifications or additions to the standard clinical equipment or workflow. We have demonstrated decreased tip localization error and variation for visually obstructed needles in both phantom and clinical cases, including providing the ability to visualize needles previously not visible using B-mode US alone. This method has the potential to improve needle visualization in challenging cases without burdening the clinical workflow, potentially improving treatment accuracy in HDR-BT and more broadly in any minimally invasive needle-based procedure.

Can ventricular 3D ultrasound of neonates with posthemorrhagic hydrocephalus inform on the need for a ventriculoperitoneal shunt?

OBJECTIVE: Some neonates born prematurely with intraventricular hemorrhage develop posthemorrhagic hydrocephalus and require lifelong treatment to divert the flow of CSF. Early prediction of the eventual need for a ventriculoperitoneal shunt (VPS) is difficult, and early discussions with families are based on statistics and the grade of hemorrhage. The authors hypothesize that change in ventricular volume during ventricular taps that is measured with repeated 3D ultrasound (3D US) imaging of the lateral ventricles could be used to assess the risk of the future requirement of a VPS. METHODS: A total of 92 neonates with intraventricular hemorrhage who were treated in the NICU were recruited between April 2012 and November 2019. Only patients who required ventricular taps (VTs) were included in this study, resulting in the analysis of 19 patients with a total of 61 VTs. Among them, 14 patients were treated with a VPS, and in 5 patients the hydrocephalus resolved spontaneously. Parameters studied were total ventricular volume measured with 3D US, ventricular volume change after VT, the ratio between volume reduction and tap amount, the difference between tap amount and volume reduction after tap, the average tap amount, the average number of days between taps, pre-tap head circumference, and reduction in head circumference after tap. RESULTS: Statistically significant differences were found in ventricular volume reduction after tap (p = 0.007), the ratio between volume reduction and tap amount (p = 0.03), the difference between tap amount and volume reduction after tap (p = 0.05), and the interval of days between taps (p = 0.0115). CONCLUSIONS: Measuring with 3D US before and after VT can be a useful tool for quantifying ventricular volume. The findings in this study showed that neonates who experience a large reduction of ventricular volume after VT are more likely to be treated with a shunt than are neonates who experience a small reduction.

Carotid Vessel-Wall-Volume Ultrasound Measurement via a UNet++ Ensemble Algorithm Trained on Small Data Sets.

Vessel wall volume (VWV) is a 3-D ultrasound measurement for the assessment of therapy in patients with carotid atherosclerosis. Deep learning can be used to segment the media-adventitia boundary (MAB) and lumen-intima boundary (LIB) and to quantify VWV automatically; however, it typically requires large training data sets with expert manual segmentation, which are difficult to obtain. In this study, a UNet++ ensemble approach was developed for automated VWV measurement, trained on five small data sets (n = 30 participants) and tested on 100 participants with clinically diagnosed coronary artery disease enrolled in a multicenter CAIN trial. The Dice similarity coefficient (DSC), average symmetric surface distance (ASSD), Pearson correlation coefficient (r), Bland-Altman plots and coefficient of variation (CoV) were used to evaluate algorithm segmentation accuracy, agreement and reproducibility. The UNet++ ensemble yielded DSCs of 91.07%-91.56% and 87.53%-89.44% and ASSDs of 0.10-0.11 mm and 0.33-0.39 mm for the MAB and LIB, respectively; the algorithm VWV measurements were correlated (r = 0.763-0.795, p < 0.001) with manual segmentations, and the CoV for VWV was 8.89%. In addition, the UNet++ ensemble trained on 30 participants achieved a performance similar to that of U-Net and Voxel-FCN trained on 150 participants. These results suggest that our approach could provide accurate and reproducible carotid VWV measurements using relatively small training data sets, supporting deep learning applications for monitoring atherosclerosis progression in research and clinical trials.

Validation of a surface-based deformable MRI-3D ultrasound image registration algorithm toward clinical implementation for interstitial prostate brachytherapy.

PURPOSE: The purpose of this study was to evaluate and clinically implement a deformable surface-based magnetic resonance imaging (MRI) to three-dimensional ultrasound (US) image registration algorithm for prostate brachytherapy (BT) with the aim to reduce operator dependence and facilitate dose escalation to an MRI-defined target. METHODS AND MATERIALS: Our surface-based deformable image registration (DIR) algorithm first translates and scales to align the US- and MR-defined prostate surfaces, followed by deformation of the MR-defined prostate surface to match the US-defined prostate surface. The algorithm performance was assessed in a phantom using three deformation levels, followed by validation in three retrospective high-dose-rate BT clinical cases. For comparison, manual rigid registration and cognitive fusion by physician were also employed. Registration accuracy was assessed using the Dice similarity coefficient (DSC) and target registration error (TRE) for embedded spherical landmarks. The algorithm was then implemented intraoperatively in a prospective clinical case. RESULTS: In the phantom, our DIR algorithm demonstrated a mean DSC and TRE of 0.74 ± 0.08 and 0.94 ± 0.49 mm, respectively, significantly improving the performance compared to manual rigid registration with 0.64 ± 0.16 and 1.88 ± 1.24 mm, respectively. Clinical results demonstrated reduced variability compared to the current standard of cognitive fusion by physicians. CONCLUSIONS: We successfully validated a DIR algorithm allowing for translation of MR-defined target and organ-at-risk contours into the intraoperative environment. Prospective clinical implementation demonstrated the intraoperative feasibility of our algorithm, facilitating targeted biopsies and dose escalation to the MR-defined lesion. This method provides the potential to standardize the registration procedure between physicians, reducing operator dependence.

Toward mechatronic MRI-guided focal laser ablation of the prostate: Robust registration for improved needle delivery.

BACKGROUND: Multiparametric MRI (mpMRI) is an effective tool for detecting and staging prostate cancer (PCa), guiding interventional therapy, and monitoring PCa treatment outcomes. MRI-guided focal laser ablation (FLA) therapy is an alternative, minimally invasive treatment method to conventional therapies, which has been demonstrated to control low-grade, localized PCa while preserving patient quality of life. The therapeutic success of FLA depends on the accurate placement of needles for adequate delivery of ablative energy to the target lesion. We previously developed an MR-compatible mechatronic system for prostate FLA needle guidance and validated its performance in open-air and clinical 3T in-bore experiments using virtual targets. PURPOSE: To develop a robust MRI-to-mechatronic system registration method and evaluate its in-bore MR-guided needle delivery accuracy in tissue-mimicking prostate phantoms. METHODS: The improved registration multifiducial assembly houses thirty-six aqueous gadolinium-filled spheres distributed over a 7.3 × 7.3 × 5.2 cm volume. MRI-guided needle guidance accuracy was quantified in agar-based tissue-mimicking prostate phantoms on trajectories (N = 44) to virtual targets covering the mechatronic system's range of motion. 3T gradient-echo recalled (GRE) MRI images were acquired after needle insertions to each target, and the air-filled needle tracks were segmented. Needle guidance error was measured as the shortest Euclidean distance between the target point and the segmented needle trajectory, and angular error was measured as the angle between the targeted trajectory and the segmented needle trajectory. These measurements were made using both the previously designed four-sphere registration fiducial assembly on trajectories (N = 7) and compared with the improved multifiducial assembly using a Mann-Whitney U test. RESULTS: The median needle guidance error of the system using the improved registration fiducial assembly at a depth of 10 cm was 1.02 mm with an interquartile range (IQR) of 0.42-2.94 mm. The upper limit of the one-sided 95% prediction interval of needle guidance error was 4.13 mm. The median (IQR) angular error was 0.0097 rad (0.0057-0.015 rad) with a one-sided 95% prediction interval upper limit of 0.022 rad. The median (IQR) positioning error using the previous four-sphere registration fiducial assembly was 1.87 mm (1.77-2.14 mm). This was found to be significantly different (p = 0.0012) from the median (IQR) positioning error of 0.28 mm (0.14-0.95 mm) using the new registration fiducial assembly on the same trajectories. No significant difference was detected between the medians of the angular errors (p = 0.26). CONCLUSION: This is the first study presenting an improved registration method and validation in tissue-mimicking phantoms of our remotely actuated MR-compatible mechatronic system for delivery of prostate FLA needles. Accounting for the effects of needle deflection, the system was demonstrated to be capable of needle delivery with an error of 4.13 mm or less in 95% of cases under ideal conditions, which is a statistically significant improvement over the previous method. The system will next be validated in a clinical setting.

Proof-of-Concept Study of a 3-D Ultrasound Scanner Used for Ankle Joint Assessment.

Joint arthropathies often require continuous monitoring of the joint condition, typically performed using magnetic resonance (MR) or ultrasound (US) imaging. US imaging is often the preferred screening or diagnostic tool as it is fast and inexpensive. However, conventional 2-D US has limited capability to compare imaging results between examinations because of its operator dependence and challenges related to repeat imaging in the same location and orientation. Comparison between several imaging sessions is crucial to assess the interval progression of joint conditions. We propose a novel 3-D US scanner for ankle joint assessment that can partially overcome these issues by enabling 3-D imaging. Here, we (i) present the design of the 3-D US ankle scanner system, (ii) validate the geometric reconstruction accuracy of the system, (iii) provide preliminary images of healthy volunteer ankles and (iv) compare 3-D US imaging results with MR imaging. The 3-D ankle scanner consists of a tub filled with water, a linear US probe attached to the wall of the tub and a motorized unit that rotates the US probe 360° around the center of the tub. As the probe rotates, a 3-D US image is formed of the ankle of the patient positioned in the middle of the tub. US probe height, angle and distance from the tub center can be adjusted. The reconstruction accuracy of the system was validated in each of the coordinate directions at different probe angles using two test phantoms. A phantom consisting of numerous Ø200-µm nylon threads with known spacing and a metal rod with machined grooves was used for validation in the horizontal and vertical directions, respectively. The volumetric reconstruction accuracy validation was performed by imaging an agar phantom with two embedded spheres of known volumes and comparing the segmented sphere volume and surface area with the expected. Three-dimensional US and MR images of both ankles of five healthy volunteers were acquired. Distal tibia and proximal talus were segmented in both imaging modalities and the surfaces of these segmentations were compared using the 95% Hausdorff and mean surface distances. The observed mean linear measurement error in all the coordinate directions and over several probe angles was 2.98%. The mean measured volumetric measurement error was 3.45%. The volunteer study revealed useful features for joint assessment present in the 3-D ankle scanner images, such as joint spacing, distal tibia and proximal talus. The mean 95% Hausdorff and mean surface distances between segmentations in 3-D US and MR images were 5.68 ± 0.83 and 2.01 ± 0.30 mm, respectively. In this proof-of-concept study, the 3-D US ankle scanner enabled visualization of the ankle joint features that are useful for joint assessment.