Brain tumor segmentation improves handling missing MRI scans
Progressive multi-scale tumor-state memory for incomplete multimodal brain tumor segmentation
Computer Vision and Pattern Recognition
Summary
Brain tumor images usually require several types of MRI scans to spot different tumor areas, but sometimes some scans are missing. The authors propose a new computer method that can accurately identify tumors even when some MRI types are not available. Their method keeps track of tumor information progressively through the brain scan at multiple scales to better combine available data. Tests show it works better than previous methods across many scenarios and datasets. This helps doctors analyze brain tumors even with incomplete imaging.
What this means in practice
- •For medical imaging teams: Improve tumor segmentation tools that must function when some MRI sequences are missing, enhancing robustness of diagnostic workflows.
- •For healthcare software developers: Integrate flexible brain tumor segmentation models into clinical diagnostic software that adapt to varying sets of MRI inputs.
Authors
Milad Rostamian, Mohammadreza Mohammadi, Mohsen Soryani
Abstract
Brain tumor segmentation uses complementary magnetic resonance imaging (MRI) sequences, but some sequences may be unavailable in practice. Their absence changes the information available for segmenting different tumor regions. Fusion methods can support different input subsets. However, fusion at several scales does not by itself specify how tumor evidence is retained and updated to guide fusion at later stages. We propose a progressive multi-scale tumor-state memory network (PMTM) for segmentation from any non-empty subset of T1ce, T1, FLAIR, and T2. PMTM maintains a compact state for the current subject across the encoder stages. At each stage, we use assignment maps to pool spatial features into ordered evidence vectors for each available modality. We average these vectors across the available modalities, and a learned gate combines this evidence with the previous state. The up- dated state guides residual fusion at that stage and is used in the state update at the next stage. A modality-token mixer also models interactions among the available modalities at each stage. We apply semantic supervision to the assignment maps at all four stages. We evaluate PMTM on BraTS2018, BraTS2020, and BraTS2023 GLI across all 15 non-empty modality subsets. Averaged over the three tumor regions and the 15 subsets, PMTM achieves Dice scores of 75.31%, 77.54%, and 80.47% on BraTS2018, BraTS2020, and BraTS2023 GLI, respectively. These scores are higher than those of RobustSeg, RFNet, and mmFormer in our experiments on each dataset. Component ablations and internal state analysis are performed on BraTS2020.