Ultrasound Probe Pose Classification for Task Recognition in Central Venous Catheterization.

Central Line Tutor is a system that facilitates real-time feedback during training for central venous catheterization. One limitation of Central Line Tutor is its reliance on expensive, cumbersome electromagnetic tracking to facilitate various training aids, including ultrasound task identification and segmentation of neck vasculature. The purpose of this study is to validate deep learning methods for vessel segmentation and ultrasound pose classification in order to mitigate the system's reliance on electromagnetic tracking. A large dataset of segmented and classified ultrasound images was generated from participant data captured using Central Line Tutor. A U-Net architecture was used to perform vessel segmentation, while a shallow Convolutional Neural Network (CNN) architecture was designed to classify the pose of the ultrasound probe. A second classifier architecture was also tested that used the U-Net output as the CNN input. The mean testing set Intersect over Union score for U-Net cross-validation was 0.746 ± 0.052. The mean test set classification accuracy for the CNN was 92.0% ± 3.0, while the U-Net + CNN achieved 92.7% ± 2.1%. This study highlights the potential for deep learning on ultrasound images to replace the current electromagnetic tracking-based methods for vessel segmentation and ultrasound pose classification, and represents an important step towards removing the electromagnetic tracker altogether. Removing the need for an external tracking system would significantly reduce the cost of Central Line Tutor and make it far more accessible to the medical trainees that would benefit from it most.

An End-to-End Solution for Automatic Contouring of Tumor Region in Intraoperative Images of Breast Lumpectomy.

Breast-conserving surgery, also known as lumpectomy, is an early stage breast cancer treatment that aims to spare as much healthy breast tissue as possible. A risk associated with lumpectomy is the presence of cancer positive margins post operation. Surgical navigation has been shown to reduce cancer positive margins but requires manual segmentation of the tumor intraoperatively. In this paper, we propose an end-to-end solution for automatic contouring of breast tumor from intraoperative ultrasound images using two convolutional neural network architectures, the U-Net and residual U-Net. The networks are trained on annotated intraoperative breast ultrasound images and evaluated on the quality of predicted segmentations. This work brings us one step closer to providing surgeons with an automated surgical navigation system that helps reduce cancer-positive margins during lumpectomy.

Prediction of myocardial tissue loss by quantitative densitometric myocardial blush parameters following ST-elevation myocardial infarction.

UNLABELLED: Tissue level myocardial perfusion is one of the most important prognostic factors after successful recanalisation of the occluded coronary artery in patients suffering acute ST elevation myocardial infarction (STEMI). The primary objective of the present study was to examine the relationship between videodensitometric myocardial perfusion parameters as assessed on coronary angiograms directly following successful recanalization therapy and magnetic resonance imaging (MRI)-derived myocardial tissue loss late after STEMI. The study comprised 29 STEMI patients. Videodensitometric parameter G(max)/T(max) was calculated to characterize myocardial perfusion, derived from the plateau of grey-level intensity (G(max)), divided by the time-to-peak intensity (Tmax). Myocardial loss index (MLI) was assessed by cardiac MRI following 376 ± 254 days after PCI. RESULTS: Significant correlations could be demonstrated between MLI and G(max) (r = 0.36, p = 0.05) and G(max)/T(max) (r = 0.40, p = 0.03) using vessel masking. Using receiver operating characteristic curve analysis, G(max)/T(max) < 2.17 predicted best MLI = 0.3, 0.4, 0.5 and 0.6 with good sensitivity and specificity data, while G(max)/T(max) < 3.25 proved to have a prognostic role in the prediction of MLI = 0.7. CONCLUSIONS: Selective myocardial tissue level perfusion quantitative measurement method is feasible and can serve as a good predictor of myocardial tissue loss following STEMI and revascularization therapy.

Videodensitometric time-density curve change after alcohol septal ablation of obstructive hypertrophic cardiomyopathy.

A recently developed computerized method for estimation of myocardial perfusion, based on the analysis of the time-density curves, is demonstrated to assess myocardial blush over a selected myocardial region of interest in a patient with obstructive hypertrophic cardiomyopathy before and after alcohol septal ablation.

Design and development of a mobile image overlay system for needle interventions.

Previously, a static and adjustable image overlay systems were proposed for aiding needle interventions. The system was either fixed to a scanner or mounted over a large articulated counterbalanced arm. Certain drawbacks associated with these systems limited the clinical translation. In order to minimize these limitations, we present the mobile image overlay system with the objective of reduced system weight, smaller dimension, and increased tracking accuracy. The design study includes optimal workspace definition, selection of display device, mirror, and laser source. The laser plane alignment, phantom design, image overlay plane calibration, and system accuracy validation methods are discussed. The virtual image is generated by a tablet device and projected into the patient by using a beamsplitter mirror. The viewbox weight (1.0 kg) was reduced by 8.2 times and image overlay plane tracking precision (0.21 mm, STD = 0.05) was improved by 5 times compared to previous system. The automatic self-calibration of the image overlay plane was achieved in two simple steps and can be done away from patient table. The fiducial registration error of the physical phantom to scanned image volume registration was 1.35 mm (STD = 0.11). The reduced system weight and increased accuracy of optical tracking should enable the system to be hand held by the physician and explore the image volume over the patient for needle interventions.

Ultrasound-guided facet joint injection training using Perk Tutor.

PURPOSE: Facet syndrome is a condition that may cause 15-45 % of chronic lower back pain. It is commonly diagnosed and treated using facet joint injections. This needle technique demands high accuracy, and ultrasound (US) is a potentially useful modality to guide the needle. US-guided injections, however, require physicians to interpret 2-D sonographic images while simultaneously manipulating an US probe and needle. Therefore, US-guidance for facet joint injections needs advanced training methodologies that will equip physicians with the requisite skills. METHODS: We used Perk Tutor-an augmented reality training system for US-guided needle insertions-in a configuration for percutaneous procedures of the lumbar spine. In a pilot study of 26 pre-medical undergraduate students, we evaluated the efficacy of Perk Tutor training compared to traditional training. RESULTS: The Perk Tutor Trained group, which had access to Perk Tutor during training, had a mean success rate of 61.5 %, while the Control group, which received traditional training, had a mean success rate of 38.5 % ([Formula: see text]). No significant differences in procedure times or needle path lengths were observed between the two groups. CONCLUSIONS: The results of this pilot study suggest that Perk Tutor provides an improved training environment for US-guided facet joint injections on a synthetic model.