Classification was performed by a 3-layer neural network in MATLAB (20 units per hidden layer). The neural network trains on the output of Image Processing Part 1 (a feature vector with an equal number of healthy and unhealthy windows, and its corresponding label vector). The neural network can then be applied to any CT and make predictions on windows regarding whether they are healthy or unhealthy.

Binary Classification:

‘0’ = healthy

‘1’ = unhealthy