Knowledge distillation from multi-modal to mono-modal segmentation networks

TL;DR

This paper introduces KD-Net, a knowledge distillation framework that transfers information from multi-modal to mono-modal segmentation networks, improving accuracy when only one modality is available.

Contribution

The paper presents a novel adaptation of generalized distillation to enable mono-modal networks to learn from multi-modal teachers, enhancing segmentation performance.

Findings

Student networks outperform baseline mono-modal models.

Effective knowledge transfer from multi-modal to mono-modal networks.

Improved segmentation accuracy demonstrated on BraTS 2018 dataset.

Abstract

The joint use of multiple imaging modalities for medical image segmentation has been widely studied in recent years. The fusion of information from different modalities has demonstrated to improve the segmentation accuracy, with respect to mono-modal segmentations, in several applications. However, acquiring multiple modalities is usually not possible in a clinical setting due to a limited number of physicians and scanners, and to limit costs and scan time. Most of the time, only one modality is acquired. In this paper, we propose KD-Net, a framework to transfer knowledge from a trained multi-modal network (teacher) to a mono-modal one (student). The proposed method is an adaptation of the generalized distillation framework where the student network is trained on a subset (1 modality) of the teacher's inputs (n modalities). We illustrate the effectiveness of the proposed framework in brain tumor segmentation with the BraTS 2018 dataset. Using different architectures, we show that the student network effectively learns from the teacher and always outperforms the baseline mono-modal network in terms of segmentation accuracy.