Robustifying Deep Networks for Medical Image Segmentation.

The purpose of this study is to investigate the robustness of a commonly used convolutional neural network for image segmentation with respect to nearly unnoticeable adversarial perturbations, and suggest new methods to make these networks more robust to such perturbations. In this retrospective study, the accuracy of brain tumor segmentation was studied in subjects with low- and high-grade gliomas. Two representative UNets were implemented to segment four different MR series (T1-weighted, post-contrast T1-weighted, T2-weighted, and T2-weighted FLAIR) into four pixelwise labels (Gd-enhancing tumor, peritumoral edema, necrotic and non-enhancing tumor, and background). We developed attack strategies based on the fast gradient sign method (FGSM), iterative FGSM (i-FGSM), and targeted iterative FGSM (ti-FGSM) to produce effective but imperceptible attacks. Additionally, we explored the effectiveness of distillation and adversarial training via data augmentation to counteract these adversarial attacks. Robustness was measured by comparing the Dice coefficients for the attacks using Wilcoxon signed-rank tests. The experimental results show that attacks based on FGSM, i-FGSM, and ti-FGSM were effective in reducing the quality of image segmentation by up to 65% in the Dice coefficient. For attack defenses, distillation performed significantly better than adversarial training approaches. However, all defense approaches performed worse compared to unperturbed test images. Therefore, segmentation networks can be adversely affected by targeted attacks that introduce visually minor (and potentially undetectable) modifications to existing images. With an increasing interest in applying deep learning techniques to medical imaging data, it is important to quantify the ramifications of adversarial inputs (either intentional or unintentional).

Machine learning to detect signatures of disease in liquid biopsies - a user's guide.

New technologies that measure sparse molecular biomarkers from easily accessible bodily fluids (e.g. blood, urine, and saliva) are revolutionizing disease diagnostics and precision medicine. Microchip devices can measure more disease biomarkers with better sensitivity and specificity each year, but clinical interpretation of these biomarkers remains a challenge. Single biomarkers in 'liquid biopsy' often cannot accurately predict the state of a disease due to heterogeneity in phenotype and disease expression across individuals. To address this challenge, investigators are combining multiplexed measurements of different biomarkers that together define robust signatures for specific disease states. Machine learning is a useful tool to automatically discover and detect these signatures, especially as new technologies output increasing quantities of molecular data. In this paper, we review the state of the field of machine learning applied to molecular diagnostics and provide practical guidance to use this tool effectively and to avoid common pitfalls.

Evaluation of Changes in the Motor Network Following BCI Therapy Based on Graph Theory Analysis.

Despite the established effectiveness of the brain-computer interface (BCI) therapy during stroke rehabilitation (Song et al., 2014a, 2015; Young et al., 2014a,b,c, 2015; Remsik et al., 2016), little is understood about the connections between motor network reorganization and functional motor improvements. The aim of this study was to investigate changes in the network reorganization of the motor cortex during BCI therapy. Graph theoretical approaches are used on resting-state functional magnetic resonance imaging (fMRI) data acquired from stroke patients to evaluate these changes. Correlations between changes in graph measurements and behavioral measurements were also examined. Right hemisphere chronic stroke patients (average time from stroke onset = 38.23 months, standard deviation (SD) = 46.27 months, n = 13, 6 males, 10 right-handed) with upper-extremity motor deficits received interventional rehabilitation therapy using a closed-loop neurofeedback BCI device. Eyes-closed resting-state fMRI (rs-fMRI) scans, along with T-1 weighted anatomical scans on 3.0T MRI scanners were collected from these patients at four test points. Immediate therapeutic effects were investigated by comparing pre and post-therapy results. Results displayed that th average clustering coefficient of the motor network increased significantly from pre to post-therapy. Furthermore, increased regional centrality of ipsilesional primary motor area (p = 0.02) and decreases in regional centrality of contralesional thalamus (p = 0.05), basal ganglia (p = 0.05 in betweenness centrality analysis and p = 0.03 for degree centrality), and dentate nucleus (p = 0.03) were observed (uncorrected). These findings suggest an overall trend toward significance in terms of involvement of these regions. Increased centrality of primary motor area may indicate increased efficiency within its interactive network as an effect of BCI therapy. Notably, changes in centrality of the bilateral cerebellum regions have strong correlations with both clinical variables [the Action Research Arm Test (ARAT), and the Nine-Hole Peg Test (9-HPT)].