ParallelDots Research & Development

AI Matches the Accuracy of Trained Radiologists for Identifying Brain Hemorrhage in a Head-to-head Test

RADNet, our proposed architecture to detect Brain hemorrhage emulates radiologists trying to detect hemorrhages in brain by sliding/up down among CT Slices, by treating Hemorrhage detection as a sequence modeling problem where the elements of sequences are 2D CT Slices. A Dense Convnet with attention is used to deduuce things at a slice level and an LSTM is then used to classify a sequence of slices. When evaluated against radiologists, RadNet had a performance comparable to radiologists and a F1 score better than them.