Knowledge distillation from multi-modal to mono-modal segmentation networks
This work addresses the challenge of medical image segmentation in resource-constrained clinical environments where only single imaging modalities are available, representing an incremental improvement by adapting existing distillation methods.
The authors tackled the problem of limited multi-modal data in clinical settings by proposing KD-Net, a knowledge distillation framework that transfers knowledge from a trained multi-modal teacher network to a mono-modal student network, resulting in improved segmentation accuracy over baseline mono-modal networks as demonstrated on the BraTS 2018 dataset.
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.