Model architecture

Hi,

We have been evaluating different models to detect/clasify if disease exists using MRI scan data. We tried with resesegnet, net(vgg16) for segmentation and densenet121 for classification. Though with small set of around 40 data points , we could see accuracy of nearly 50-60%. We would be getting more data once we have tie-up with institutions. Want to know what models architecture can we evaluate for CT scan of Brain for both segmentation as well for classification purpose?