Registration is open - make plans to attend
Get pricing and details
>
Conference 12033 > Paper 12033-19
Paper 12033-19

Deep hybrid convolutional wavelet networks: application to predicting response to chemoradiation in rectal cancers via MRI

In person: 22 February 2022 • 10:10 AM - 10:30 AM PST

Abstract

We present a novel tool to integrate wavelet networks into a convolutional neural network (CNN), termed a deep hybrid convolutional wavelet network (DHCWN). The proposed model comprises the wavelons that use the shift and scale parameters of a mother wavelet as its building units. Whereas the activation functions in a typical CNN are fixed and monotonic (e.g. ReLU), the activation functions of DHCWN are wavelets that are flexible and more stable during optimization. DHCWN was evaluated using a multi-institutional cohort of 95 pre-treatment rectal cancer MRI scans to predict pathologic response to neoadjuvant chemoradiation. Compared to CNN and a multilayer wavelet perceptron, DHCWN yielded significantly better performance in predicting treatment response in training and hold-out validation sets, with 90.67% and 91.17% accuracy, respectively. DHCWN thus offers a significantly more extensible and effective solution for characterizing predictive signatures via routine imaging data.

Presenter

Amir Reza Sadri
Case Western Reserve Univ. (United States)
Presenter/Author
Amir Reza Sadri
Case Western Reserve Univ. (United States)
Author
Thomas DeSilvio
Case Western Reserve Univ. (United States)
Author
Cleveland Clinic (United States)
Author
Univ. Hospitals Cleveland Medical Ctr. (United States)
Author
Univ. Hospitals Cleveland Medical Ctr. (United States)
Author
Cleveland Clinic (United States)
Author
Cleveland Clinic (United States)
Author
Univ. Hospitals Cleveland Medical Ctr. (United States)
Author
Case Western Reserve Univ. (United States)