Nucleus classification in histology images using message passing network

Identification of nuclear components in the histology landscape is an important step towards developing computational pathology tools for the profiling of tumor micro-environment. Most existing methods for the identification of such components are limited in scope due to heterogeneous nature of the nuclei. Graph-based methods offer a natural way to formulate the nucleus classification problem to incorporate both appearance and geometric locations of the nuclei. The main challenge is to define models that can handle such an unstructured domain. Current approaches focus on learning better features and then employ well-known classifiers for identifying distinct nuclear phenotypes. In contrast, we propose a message passing network that is a fully learnable framework build on classical network flow formulation. Based on physical interaction of the nuclei, a nearest neighbor graph is constructed such that the nodes represent the nuclei centroids. For each edge and node, appearance and geometric features are computed which are then used for the construction of messages utilized for diffusing contextual information to the neighboring nodes. Such an algorithm can infer global information over an entire network and predict biologically meaningful nuclear communities. We show that learning such communities improves the performance of nucleus classification task in histology images. The proposed algorithm can be used as a component in existing state-of-the-art methods resulting in improved nucleus classification performance across four different publicly available datasets.     

Segmentation only uses sparse annotations: Unified weakly and semi-supervised learning in medical images

Since segmentation labeling is usually time-consuming and annotating medical images requires professional expertise, it is laborious to obtain a large-scale, high-quality annotated segmentation dataset. We propose a novel weakly- and semi-supervised framework named SOUSA (Segmentation Only Uses Sparse Annotations), aiming at learning from a small set of sparse annotated data and a large amount of unlabeled data. The proposed framework contains a teacher model and a student model. The student model is weakly supervised by scribbles and a Geodesic distance map derived from scribbles. Meanwhile, a large amount of unlabeled data with various perturbations are fed to student and teacher models. The consistency of their output predictions is imposed by Mean Square Error (MSE) loss and a carefully designed Multi-angle Projection Reconstruction (MPR) loss. Extensive experiments are conducted to demonstrate the robustness and generalization ability of our proposed method. Results show that our method outperforms weakly- and semi-supervised state-of-the-art methods on multiple datasets. Furthermore, our method achieves a competitive performance with some fully supervised methods with dense annotation when the size of the dataset is limited.     

Longitudinal diffusion tensor imaging of patients with degenerative cervical myelopathy following decompression surgery

Previous studies have reported the utility of diffusion tensor imaging (DTI) as an imaging biomarker for the severity of myelopathy and subsequent surgical outcome in patients with degenerative cervical myelopathy (DCM). We hypothesized that DTI may reflect neurological recovery following surgery. The purpose of this study was to evaluate the ability of DTI to assess the post-operative alteration of neural status in patients with DCM as well as to predict post-operative recovery. We enrolled 15 patients with DCM who underwent decompression surgery. The Japanese Orthopaedic Association (JOA) score was evaluated before and 1 year after surgery. The participants were examined using DTI on a 3.0 T magnetic resonance scanner before, and 1 year after surgery. Fractional anisotropy (FA) and mean diffusivity (MD) were assessed for both time points. The correlations between the pre- and post-operative FA and MD values and the pre- and post-operative JOA scores were analyzed. Although the JOA score improved significantly after surgery from 8.9 to 12.3, there was no significant change between the pre- and post-operative FA and MD values. The post-operative outcomes after 1 year moderately correlated with the pre-operative FA values (Spearman’s ρ = 0.55, p = 0.03 and Spearman’s ρ = 0.56, p = 0.03 for change and recovery rate of the JOA score, respectively). However, there was no correlation between the post-operative FA and post-operative JOA scores nor between MD and clinical outcomes. DTI cannot be utilized as a biomarker for post-operative alterations of neural status of the spinal cord; however, pre-operative DTI may be useful as a predictor of surgical outcomes.     

Fall in the Proportion of Atherothrombotic Strokes During the Last Decade

ObjectiveTo determine the proportion of subtypes of ischemic strokes, vascular risk factors and treatment prior to stroke between 1997 and 2018 in a single institution in Argentina.MethodsDemographics, risk factors, medications and TOAST subtypes were assessed and compared in ischemic stroke patients admitted during two periods of time, 1997-2007 (P1) and 2008-2018 (P2).ResultsThere were 2747 patients (64% men, aged 67 ±15 years), 920 subjects in P1 and 1827 in P2. Age and gender distribution did not change over time. Proportion of large artery atherothrombotic strokes decreased from 29% in P1 to 14% in P2 (p <0.0001) and small vessel strokes from 15% to 11% (p <0.05). Cardioembolic and undetermined strokes increased from 17 to 25% (p <0.0001) and from 30% to 41% (p <0.0001), respectively. There were no changes in stroke of other etiologies (9% in both periods). Detection of atrial fibrillation increased from 14% to 19% (p<0.001). Use of medications prior to stroke increased for aspirin from 27% to 45% (p <0.0001), for antihypertensive drugs from 26% to 62% (p <0.0001), for statins from 14% to 42% (p<0.0001) and for anticoagulants from 4% to 9% (p<0.0001).ConclusionsThe proportion of strokes associated to large and small vessel atherosclerosis is declining in our population with an increase in the proportion of cardioembolic and undetermined strokes. Better management of risk factors and higher prevalence and/or better screening for atrial fibrillation could explain, at least in part, these findings.     

Multi-site clustering and nested feature extraction for identifying autism spectrum disorder with resting-state fMRI

Brain functional connectivity (FC) derived from resting-state functional magnetic resonance imaging (rs-fMRI) has been widely employed to study neuropsychiatric disorders such as autism spectrum disorder (ASD). Existing studies usually suffer from (1) significant data heterogeneity caused by different scanners or studied populations in multiple sites, (2) curse of dimensionality caused by millions of voxels in each fMRI scan and a very limited number (tens or hundreds) of training samples, and (3) poor interpretability, which hinders the identification of reproducible disease biomarkers. To this end, we propose a Multi-site Clustering and Nested Feature Extraction (MC-NFE) method for fMRI-based ASD detection. Specifically, we first divide multi-site training data into ASD and healthy control (HC) groups. To model inter-site heterogeneity within each category, we use a similarity-driven multiview linear reconstruction model to learn latent representations and perform subject clustering within each group. We then design a nested singular value decomposition (SVD) method to mitigate inter-site heterogeneity and extract FC features by learning both local cluster-shared features across sites within each category and global category-shared features across ASD and HC groups, followed by a linear support vector machine (SVM) for ASD detection. Experimental results on 609 subjects with rs-fMRI from the ABIDE database with 21 imaging sites suggest that the proposed MC-NFE outperforms several state-of-the-art methods in ASD detection. The most discriminative FCs identified by the MC-NFE are mainly located in default mode network, salience network, and cerebellum region, which could be used as potential biomarkers for fMRI-based ASD analysis.     

Disease prediction with edge-variational graph convolutional networks

The need for computational models that can incorporate imaging data with non-imaging data while investigating inter-subject associations arises in the task of population-based disease analysis. Although off-the-shelf deep convolutional neural networks have empowered representation learning from imaging data, incorporating data of different modalities complementarily in a unified model to improve the disease diagnostic quality is still challenging. In this work, we propose a generalizable graph-convolutional framework for population-based disease prediction on multi-modal medical data. Unlike previous methods constructing a static affinity population graph in a hand-crafting manner, the proposed framework can automatically learn to build a population graph with variational edges, which we show can be optimized jointly with spectral graph convolutional networks. In addition, to estimate the predictive uncertainty related to the constructed graph, we propose Monte–Carlo edge dropout uncertainty estimation. Experimental results on four multi-modal datasets demonstrate that the proposed method can substantially improve the predictive accuracy for Autism Spectrum Disorder, Alzheimer’s disease, and ocular diseases. A sufficient ablation study with in-depth discussion is conducted to evaluate the effectiveness of each component and the choice of algorithmic details of the proposed method. The results indicate the potential and extendability of the proposed framework in leveraging multi-modal data for population-based disease prediction.     

Percutaneous CT-Guided Cryoablation of the Celiac Plexus: A Retrospective Cohort Comparison with Ethanol

PurposeTo retrospectively analyze and compare the incidence of diarrhea in patients who underwent cryoablation of the celiac plexus for intractable abdominal pain versus ethanol therapy over a 5-year period.Materials and MethodsFrom June 2014 to August 2019, 83 patients were identified who underwent neurolysis of the celiac plexus for management of intractable abdominal pain by using either cryoablation (n = 39 [59% female; age range, 36–79 years old [average, 60 ± 11 years old]) or alcohol (n = 44 [48% female; age range, 29–76 years old [average, 60 ± 12 years old]). Pain scores and reports of procedure-related complications or side effects, with special attention to diarrhea and/or other gastrointestinal symptoms, were collected from follow-up visits at 1 week, 1 month, and 3 months post-intervention and were compared between groups.ResultsThe mean time of follow-up was 17.7 days. Four patients who underwent cryoablation developed gastrointestinal symptoms consisting of 2 cases of nausea and vomiting and 2 cases of diarrhea (5.1%). Twelve patients who underwent ethanol ablation developed gastrointestinal symptoms, including 1 case of nausea, 3 cases of vomiting, and 9 cases of diarrhea (20.5%). There was a significantly higher incidence of both diarrhea (chi-squared likelihood ratio, P = .03) and overall gastrointestinal symptoms (chi-squared likelihood ratio, P = .04) in the ethanol group than in the cryoablation group.ConclusionsCryoablation of the celiac plexus may provide a new treatment option for intractable abdominal pain, and it appears to have a lower incidence of diarrhea and fewer gastrointestinal side effects than ablation using ethanol.