William Kovacs

Holistic Segmentation of Intermuscular Adipose Tissues on Thigh MRI.

Muscular dystrophies (MD) cause muscles to gradually degenerate into fat. In order to effectively study and track disease progression, it is important to quantify both muscle and fat volumes, especially the intermuscular adipose tissue (IMAT). Existing methods were mostly based on unsupervised pixel clustering and morphological models. We propose a method integrating two holistic neural networks (one for edges and one for regions) and a dual active contour model to accurately locate the fascia lata and segment multiple tissue types on thigh MRIs. The proposed method is robust to image artifacts and weak boundaries, and thus it performs well for severe MD cases. Our method was tested on 104 data sets and achieved Dice coefficients 0.940 and 0.943 for muscle and IMAT in challenging severe cases, respectively.

Identification of muscle and subcutaneous and intermuscular adipose tissue on thigh MRI of muscular dystrophy

Muscular dystrophies can affect the muscle distribution within the legs. In order to effectively study and track disease progression, it is important to quantify both muscle and fat volumes, and distinguish between subcutaneous (SAT) and intermuscular adipose tissue (IMAT). While several techniques have been previously described that perform such classification, they rely heavily on the muscle location, and so may not be suitable for differentiating SAT and IMAT in severe cases of dystrophy. We propose a method that utilizes muscle location if available, but also identifies the fascia lata to serve as the boundary between SAT and IMAT. Our method achieved DICE coefficients of 0.93, 0.88, and 0.68 for muscle, SAT, and IMAT, respectively, in mild cases, of 0.94, 0.91, and 0.85 in moderate cases, and of 0.79, 0.90, and 0.91 in severe cases.

Respiratory magnetic resonance imaging biomarkers in Duchenne muscular dystrophy

To examine the diaphragm and chest wall dynamics with cine breathing magnetic resonance imaging (MRI) in ambulatory boys with Duchenne muscular dystrophy (DMD) without respiratory symptoms and controls.

Differentiation of fat, muscle, and edema in thigh MRIs using random forest classification

There are many diseases that affect the distribution of muscles, including Duchenne and fascioscapulohumeral dystrophy among other myopathies. In these disease cases, it is important to quantify both the muscle and fat volumes to track the disease progression. There has also been evidence that abnormal signal intensity on the MR images, which often is an indication of edema or inflammation can be a good predictor for muscle deterioration. We present a fully-automated method that examines magnetic resonance (MR) images of the thigh and identifies the fat, muscle, and edema using a random forest classifier. First the thigh regions are automatically segmented using the T1 sequence. Then, inhomogeneity artifacts were corrected using the N3 technique. The T1 and STIR (short tau inverse recovery) images are then aligned using landmark based registration with the bone marrow. The normalized T1 and STIR intensity values are used to train the random forest. Once trained, the random forest can accurately classify the aforementioned classes. This method was evaluated on MR images of 9 patients. The precision values are 0.91±0.06, 0.98±0.01 and 0.50±0.29 for muscle, fat, and edema, respectively. The recall values are 0.95±0.02, 0.96±0.03 and 0.43±0.09 for muscle, fat, and edema, respectively. This demonstrates the feasibility of utilizing information from multiple MR sequences for the accurate quantification of fat, muscle and edema.

Retrieval, visualization, and mining of large radiation dosage data

Radiation dose monitoring has become an essential service that hospitals must perform. Depending on the system in place, this can result in the collection of large quantities of data, ripe for analysis. These data should include a wide variety of variables for each study because assessment of the propriety of the patient’s dose is dependent on many factors, including patient age and size, as well as the body section that is being scanned. Moreover, the scanners themselves have many properties that affect patient dose, such as model, pitch and kVp. In this paper, we propose an engine that seamlessly integrated with a clinical PACS to retrieve radiation dosage data. We devised several schemes to analyze these data through visualization and mining techniques that examine it at different scopes. We demonstrate the utility of such visual methods at examining large, noisy, and multi-dimensional data, which is embodied in the collected radiation data.

Cumulative Radiation Exposures from CT Screening and Surveillance Strategies for von Hippel-Lindau–associated Solid Pancreatic Tumors

Purpose To assess the potential ionizing radiation exposure from CT scans for both screening and surveillance of patients with von Hippel-Lindau (VHL) syndrome. Materials and Methods For this retrospective study, abdomen-pelvic (AP) and chest-abdomen-pelvic (CAP) CT scans were performed with either a three-phase (n = 1242) or a dual-energy virtual noncontrast protocol (VNC; n = 149) in 747 patients with VHL syndrome in the National Institutes of Health Clinical Center between 2009 and 2015 (mean age, 47.6 years ± 14.6 [standard deviation]; age range, 12-83 years; 320 women [42.8%]). CT scanning parameters for patients with pancreatic neuroendocrine tumors (PNETs; 124 patients and 381 scans) were compared between a tumor diameter-based surveillance protocol and a VHL genotype and tumor diameter-based algorithm (a tailored algorithm) developed by three VHL clinicians. Organ and lifetime radiation doses were estimated by two radiologists and five radiation scientists. Cumulative radiation doses were compared between the PNET surveillance algorithms by analyses of variance, and a two-tailed P value less than .05 indicated statistical significance. Results Median cumulative colon doses for annual CAP and AP CT scans from age 15 to 40 years ranged from 0.34 Gy (5th-95th percentiles, 0.18-0.75; dual-energy VNC CT) to 0.89 Gy (5th-95th percentiles, 0.42-1.0; three-phase CT). For the current PNET surveillance protocol, the cumulative effective radiation dose from age 40 to 65 years was 682 mSv (tumors < 1.2 cm) and 2125 mSv (tumors > 3 cm). The tailored algorithm could halve these doses for patients with initial tumor diameter less than 1.2 cm (P < .001). Conclusion CT screening of patients with von Hippel-Lindau syndrome can lead to substantial radiation exposures, even with dual-energy virtual noncontrast CT. A genome and tumor diameter-based algorithm for pancreatic neuroendocrine tumor surveillance may potentially reduce lifetime radiation exposure.

Holistic segmentation of the lung in cine MRI

Abstract. Duchenne muscular dystrophy (DMD) is a childhood-onset neuromuscular disease that results in the degeneration of muscle, starting in the extremities, before progressing to more vital areas, such as the lungs. Respiratory failure and pneumonia due to respiratory muscle weakness lead to hospitalization and early mortality. However, tracking the disease in this region can be difficult, as current methods are based on breathing tests and are incapable of distinguishing between muscle involvements. Cine MRI scans give insight into respiratory muscle movements, but the images suffer due to low spatial resolution and poor signal-to-noise ratio. Thus, a robust lung segmentation method is required for accurate analysis of the lung and respiratory muscle movement. We deployed a deep learning approach that utilizes sequence-specific prior information to assist the segmentation of lung in cine MRI. More specifically, we adopt a holistically nested network to conduct image-to-image holistic training and prediction. One frame of the cine MRI is used in the training and applied to the remainder of the sequence (>200 frames). We applied this method to cine MRIs of the lung in the axial, sagittal, and coronal planes. Characteristic lung motion patterns during the breathing cycle were then derived from the segmentations and used for diagnosis. Our data set consisted of 31 young boys, age 9.6±2.3 years, 15 of whom suffered from DMD. The remaining 16 subjects were age-matched healthy volunteers. For validation, slices from inspiratory and expiratory cycles were manually segmented and compared with results obtained from our method. The Dice similarity coefficient for the deep learning-based method was 97.2±1.3 for the sagittal view, 96.1±3.8 for the axial view, and 96.6±1.7 for the coronal view. The holistic neural network approach was compared with an approach using Demon’s registration and showed superior performance. These results suggest that the deep learning-based method reliably and accurately segments the lung across the breathing cycle.