Hybrid convolutional wavelet networks for predicting treatment response via radiological images of bowel disease

Abstract

In some embodiments, the present disclosure relates to a non-transitory computer-readable medium storing computer-executable instructions that, when executed, cause a processor to perform operations, including forming an imaging dataset having imaging data corresponding to one or more radiological images of a patient having a bowel disease; operating upon the imaging data with one or more convolutional neural network (CNN) segments configured to generate a plurality of CNN outputs, the one or more CNN segments respectively including a convolution layer configured to perform a convolution on the imaging data; and applying a wavelet network to the plurality of CNN outputs to generate a plurality of convolution wavelet network (CWN) outputs, the wavelet network being configured to decompose the plurality of CNN outputs according to a mother wavelet. A predictive signature associated with disease response or risk is constructed using the plurality of CWN outputs.