Back

Same same but different: a web-based deep learning application for the histopathologic distinction of cortical malformations

Kubach, J.; Muhlebner-Fahrngruber, A.; Soylemezoglu, F.; Miyata, H.; Niehusmann, P.; Honavar, M.; Rogerio, F.; Kim, S.-H.; Aronica, E.; Garbelli, R.; Vilz, S.; Popp, A.; Walcher, S.; Christoph, N.; Scholz, M.; Kuerten, S.; Schropp, V.; Roeder, S.; Eichhorn, P.; Eckstein, M.; Brehmer, A.; Kobow, K.; Coras, R.; Bluemcke, I.; Jabari, S.

2019-10-15 pathology
10.1101/804682 bioRxiv
Show abstract

We trained a convolutional neural network (CNN) to classify H.E. stained microscopic images of focal cortical dysplasia type IIb (FCD IIb) and cortical tuber of tuberous sclerosis complex (TSC). Both entities are distinct subtypes of human malformations of cortical development that share histopathological features consisting of neuronal dyslamination with dysmorphic neurons and balloon cells. The microscopic review of routine stainings of such surgical specimens remains challenging. A digital processing pipeline was developed for a series of 56 FCD IIb and TSC cases to obtain 4000 regions of interest and 200.000 sub-samples with different zoom and rotation angles to train a CNN. Our best performing network achieved 91% accuracy and 0.88 AUCROC (area under the receiver operating characteristic curve) on a hold-out test-set. Guided gradient-weighted class activation maps visualized morphological features used by the CNN to distinguish both entities. We then developed a web application, which combined the visualization of whole slide images (WSI) with the possibility for classification between FCD IIb and TSC on demand by our pretrained and build-in CNN classifier. This approach might help to introduce deep learning applications for the histopathologic diagnosis of rare and difficult-to-classify brain lesions.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.