MED-TEX: Transferring and Explaining Knowledge with Less Data from Pretrained Medical Imaging Models

TL;DR

MED-TEX introduces a framework that enables knowledge transfer from large pretrained medical imaging models to smaller models with less data, while also providing interpretability of the model's decisions.

Contribution

The paper presents a novel joint framework combining knowledge distillation and interpretability for medical image classification, trained from an information-theoretic perspective.

Findings

Outperforms state-of-the-art methods in knowledge distillation.

Effectively highlights important input regions for model predictions.

Reduces data requirements for training medical imaging models.

Abstract

Deep learning methods usually require a large amount of training data and lack interpretability. In this paper, we propose a novel knowledge distillation and model interpretation framework for medical image classification that jointly solves the above two issues. Specifically, to address the data-hungry issue, a small student model is learned with less data by distilling knowledge from a cumbersome pretrained teacher model. To interpret the teacher model and assist the learning of the student, an explainer module is introduced to highlight the regions of an input that are important for the predictions of the teacher model. Furthermore, the joint framework is trained by a principled way derived from the information-theoretic perspective. Our framework outperforms on the knowledge distillation and model interpretation tasks compared to state-of-the-art methods on a fundus dataset.