BRAX, Brazilian labeled chest x-ray dataset.

Chest radiographs allow for the meticulous examination of a patient's chest but demands specialized training for proper interpretation. Automated analysis of medical imaging has become increasingly accessible with the advent of machine learning (ML) algorithms. Large labeled datasets are key elements for training and validation of these ML solutions. In this paper we describe the Brazilian labeled chest x-ray dataset, BRAX: an automatically labeled dataset designed to assist researchers in the validation of ML models. The dataset contains 24,959 chest radiography studies from patients presenting to a large general Brazilian hospital. A total of 40,967 images are available in the BRAX dataset. All images have been verified by trained radiologists and de-identified to protect patient privacy. Fourteen labels were derived from free-text radiology reports written in Brazilian Portuguese using Natural Language Processing.

Contralesional Cathodal Transcranial Direct Current Stimulation Does Not Enhance Upper Limb Function in Subacute Stroke: A Pilot Randomized Clinical Trial.

Transcranial direct current stimulation (tDCS) has the potential to improve upper limb motor outcomes after stroke. According to the assumption of interhemispheric inhibition, excessive inhibition from the motor cortex of the unaffected hemisphere to the motor cortex of the affected hemisphere may worsen upper limb motor recovery after stroke. We evaluated the effects of active cathodal tDCS of the primary motor cortex of the unaffected hemisphere (ctDCSM1UH) compared to sham, in subjects within 72 hours to 6 weeks post ischemic stroke. Cathodal tDCS was intended to inhibit the motor cortex of the unaffected hemisphere and hence decrease the inhibition from the unaffected to the affected hemisphere and enhance motor recovery. We hypothesized that motor recovery would be greater in the active than in the sham group. In addition, greater motor recovery in the active group might be associated with bigger improvements in measures in activity and participation in the active than in the sham group. We also explored, for the first time, changes in cognition and sleep after ctDCSM1UH. Thirty subjects were randomized to six sessions of either active or sham ctDCSM1UH as add-on interventions to rehabilitation. The NIH Stroke Scale (NIHSS), Fugl-Meyer Assessment of Motor Recovery after Stroke (FMA), Barthel Index (BI), Stroke Impact Scale (SIS), and Montreal Cognitive Assessment (MoCA) were assessed before, after treatment, and three months later. In the intent-to-treat (ITT) analysis, there were significant GROUP∗TIME interactions reflecting stronger gains in the sham group for scores in NIHSS, FMA, BI, MoCA, and four SIS domains. At three months post intervention, the sham group improved significantly compared to posttreatment in FMA, NIHSS, BI, and three SIS domains while no significant changes occurred in the active group. Also at three months, NIHSS improved significantly in the sham group and worsened significantly in the active group. FMA scores at baseline were higher in the active than in the sham group. After adjustment of analysis according to baseline scores, the between-group differences in FMA changes were no longer statistically significant. Finally, none of the between-group differences in changes in outcomes after treatment were considered clinically relevant. In conclusion, active CtDCSM1UH did not have beneficial effects, compared to sham. These results were consistent with other studies that applied comparable tDCS intensities/current densities or treated subjects with severe upper limb motor impairments during the first weeks post stroke. Dose-finding studies early after stroke are necessary before planning larger clinical trials.

Diffusion Tensor Imaging Biomarkers to Predict Motor Outcomes in Stroke: A Narrative Review.

Stroke is a leading cause of disability worldwide. Motor impairments occur in most of the patients with stroke in the acute phase and contribute substantially to disability. Diffusion tensor imaging (DTI) biomarkers such as fractional anisotropy (FA) measured at an early phase after stroke have emerged as potential predictors of motor recovery. In this narrative review, we: (1) review key concepts of diffusion MRI (dMRI); (2) present an overview of state-of-art methodological aspects of data collection, analysis and reporting; and (3) critically review challenges of DTI in stroke as well as results of studies that investigated the correlation between DTI metrics within the corticospinal tract and motor outcomes at different stages after stroke. We reviewed studies published between January, 2008 and December, 2018, that reported correlations between DTI metrics collected within the first 24 h (hyperacute), 2-7 days (acute), and >7-90 days (early subacute) after stroke. Nineteen studies were included. Our review shows that there is no consensus about gold standards for DTI data collection or processing. We found great methodological differences across studies that evaluated DTI metrics within the corticospinal tract. Despite heterogeneity in stroke lesions and analysis approaches, the majority of studies reported significant correlations between DTI biomarkers and motor impairments. It remains to be determined whether DTI results could enhance the predictive value of motor disability models based on clinical and neurophysiological variables.