MGML: A plug-and-play meta-guided multi-modal learning framework for incomplete multimodal brain tumor segmentation

Clinical Snapshot

Key Findings

• P Value: Not reported

• Effect Size: Average Dice scores: WT 87.55, TC 79.36, ET 62.67 (BraTS2020)

• Primary Outcome: Dice scores for brain tumor segmentation across missing modality combinations

• Nnt Or Sensitivity: Dice coefficient improvements over baseline methods across fifteen missing modality scenarios

• Confidence Interval: Not reported

Clinical Application

Requires significant computational infrastructure and technical expertise for implementation in clinical practice Could be relevant for Australian radiology departments dealing with incomplete MRI sequences, though TGA approval would be needed for clinical deployment. Integration with existing PACS systems would require careful validation Patients with brain tumors undergoing multimodal MRI where complete imaging sequences may not be available

Abstract

Leveraging multimodal information from Magnetic Resonance Imaging (MRI) plays a vital role in lesion segmentation, especially for brain tumors. However, in clinical practice, multimodal MRI data are often incomplete, making it challenging to fully utilize the available information. Therefore, maximizing the utilization of this incomplete multimodal information presents a crucial research challenge. We present a novel meta-guided multi-modal learning (MGML) framework that comprises two components: meta-parameterized adaptive modality fusion and consistency regularization module. The meta-parameterized adaptive modality fusion (Meta-AMF) enables the model to effectively integrate information from multiple modalities under varying input conditions. By generating adaptive soft-label supervision signals based on the available modalities, Meta-AMF explicitly promotes more coherent multimodal fusion. In addition, the consistency regularization module enhances segmentation performance and implicitly reinforces the robustness and generalization of the overall framework. Notably, our approach does not alter the original model architecture and can be conveniently integrated into the training pipeline for end-to-end model optimization. We conducted extensive experiments on the public BraTS2020 and BraTS2023 datasets. Compared to multiple state-of-the-art methods from previous years, our method achieved superior performance. On BraTS2020, for the average Dice scores across fifteen missing modality combinations, building upon the baseline, our method obtained scores of 87.55, 79.36, and 62.67 for the whole tumor (WT), the tumor core (TC), and the enhancing tumor (ET), respectively. We have made our source code publicly available at https://github.com/worldlikerr/MGML.

References

1. 1.Zou, Y., Liu, B., Zheng, C. J., Geng, Y. M., Li, S., Zuo, Q., Wang, S., Zhang, Y., & Hong, J. (2026). MGML: A plug-and-play meta-guided multi-modal learning framework for incomplete multimodal brain tumor segmentation. Neural Networks, 108525. https://doi.org/10.1016/j.neunet.2025.108525