Frequency Disentangled Learning for Segmentation of Midbrain Structures from Quantitative Susceptibility Mapping Data
This addresses the challenge of training deep learning models with scarce data for medical imaging tasks, particularly in neurology for parkinsonian syndromes, though it is incremental as it builds on known frequency properties of deep models.
The paper tackles the problem of limited annotated data for deep segmentation models in less common imaging modalities like Quantitative Susceptibility Mapping (QSM) by proposing a frequency-domain disentanglement training method, resulting in considerable performance improvements for midbrain structure segmentation and up to 7 points of Dice gain on public datasets.
One often lacks sufficient annotated samples for training deep segmentation models. This is in particular the case for less common imaging modalities such as Quantitative Susceptibility Mapping (QSM). It has been shown that deep models tend to fit the target function from low to high frequencies. One may hypothesize that such property can be leveraged for better training of deep learning models. In this paper, we exploit this property to propose a new training method based on frequency-domain disentanglement. It consists of two main steps: i) disentangling the image into high- and low-frequency parts and feature learning; ii) frequency-domain fusion to complete the task. The approach can be used with any backbone segmentation network. We apply the approach to the segmentation of the red and dentate nuclei from QSM data which is particularly relevant for the study of parkinsonian syndromes. We demonstrate that the proposed method provides considerable performance improvements for these tasks. We further applied it to three public datasets from the Medical Segmentation Decathlon (MSD) challenge. For two MSD tasks, it provided smaller but still substantial improvements (up to 7 points of Dice), especially under small training set situations.