Deep Learning Based HPV Status Prediction for Oropharyngeal Cancer Patients

TL;DR

This study demonstrates that transfer learning using a video pre-trained 3D CNN can effectively predict HPV status in oropharyngeal cancer patients from CT images, achieving high accuracy on external data.

Contribution

It introduces a novel transfer learning approach with a video pre-trained model for HPV status prediction from CT scans, outperforming traditional CNNs.

Findings

Achieved AUC of 0.81 on external test set.

Video pre-trained model outperformed CNNs trained from scratch.

Transfer learning improved HPV classification accuracy.

Abstract

We investigated the ability of deep learning models for imaging based HPV status detection. To overcome the problem of small medical datasets we used a transfer learning approach. A 3D convolutional network pre-trained on sports video clips was fine tuned such that full 3D information in the CT images could be exploited. The video pre-trained model was able to differentiate HPV-positive from HPV-negative cases with an area under the receiver operating characteristic curve (AUC) of 0.81 for an external test set. In comparison to a 3D convolutional neural network (CNN) trained from scratch and a 2D architecture pre-trained on ImageNet the video pre-trained model performed best.