M3AE: Multimodal Representation Learning for Brain Tumor Segmentation with Missing Modalities

Abstract

Multimodal magnetic resonance imaging (MRI) provides complementary information for sub-region analysis of brain tumors. Plenty of methods have been proposed for automatic brain tumor segmentation using four common MRI modalities and achieved remarkable performance. In practice, however, it is common to have one or more modalities missing due to image corruption, artifacts, acquisition protocols, allergy to contrast agents, or simply cost. In this work, we propose a novel two-stage framework for brain tumor segmentation with missing modalities. In the first stage, a multimodal masked autoencoder (M 3 AE) is proposed, where both random modalities (i.e., modality dropout) and random patches of the remaining modalities are masked for a reconstruction task, for self-supervised learning of robust multimodal representations against missing modalities. To this end, we name our framework M 3 AE. Meanwhile, we employ model inversion to optimize a representative full-modal image at marginal extra cost, which will be used to substitute for the missing modalities and boost performance during inference. Then in the second stage, a memory-efficient self distillation is proposed to distill knowledge between heterogenous missing-modal situations while fine-tuning the model for supervised segmentation. Our M 3 AE belongs to the 'catchall' genre where a single model can be applied to all possible subsets of modalities, thus is economic for both training and deployment. Extensive experiments on BraTS 2018 and 2020 datasets demonstrate its superior performance to existing state-of-the-art methods with missing modalities, as well as the efficacy of its components. Our code is available at: https://github.com/ccarliu/m3ae.

RedCore: Relative Advantage Aware Cross-Modal Representation Learning for Missing Modalities with Imbalanced Missing Rates

“…Multimodal learning leverages heterogeneous and comprehensive signals, such as acoustic, visual, lexical information to perform typical machine learning tasks, for instance, clustering, regression, classification, and retrieval (Sun, Dong, and Liu 2021;Zhang et al 2022;Han et al 2022). Relative to its unimodal counterpart, multimodal learning has demonstrated great success in numerous applications, including but not limited to medical analysis (Liu et al 2023), action recognition (Woo et al 2023), affective computing (Sun et al 2023). Nevertheless, multimodal learning is inevitably faced with the modality missing issue due to malfunctioning sensors, high data acquisition costs, privacy concerns, etc.…”

Survey of Brain Tumor Image Segmentation Using Artificial Intelligence Techniques

“…Tumor core is highlighted by T1-weighted (T1), T2-weighted (T2), T1-weighted (T1), contrast enhanced T1weighted (T1c), and T2 fluid attenuation inversion recovery (FLAIR). Peritumoral edema is highlighted by the final two [6] typically, each MR-imaging modality slice's aberrant areas are manually segmented by the neurologist. Nevertheless, the BTs' laborious and subjective manual MRI image segmentation.…”