On-Demand medical image segmentation
Magnetic resonance images (MRIs) aren't utilized to their full potential because they are subjectively evaluated, often leaving out crucial quantitative information. This quantitative information is slow and expensive to obtain because first, the MRI needs to be segmented. Segmentation can be thought of as colouring in an MRI to identify the structures of interest (i.e. tumour, cartilage). Segmentation is slow because an MRI is made up of hundreds of individual images and manually colouring each image takes hours. Segmentations provide quantitative metrics and help plan for surgeries by viewing 3D models, or 3D printing structures.
Current algorithms are slow, with reported segmentation times of 30 minutes to 2 days. NeuralSeg uses deep learning, and segments with an accuracy equivalent to the best reported. NeuralSeg's primary advantage is that it segments MRIs in 2-minutes. We are harnessing this efficiency to offer on-demand image segmentation via a web-application.