Advanced Computational Approaches for Multimodal Fusion in Mass Spectrometry Imaging

Mass spectrometry imaging (MSI) enables the spatially resolved mapping of molecular distributions within biological tissues, which has become an indispensable analytical technique across diverse biomedical applications. Nevertheless, the utility of MSI in complex biological specimens is often hampered by intrinsic limitations in spatial resolution and a lack of detailed structural or functional information. Consequently, MSI-based multimodal fusion, which integrates MSI data with complementary imaging modalities, has become a commonly used strategy for advancing the MSI technique. In this perspective, we critically review representative advances in computational approaches for MSI-based multimodal fusion, with particular focus on its two fundamental components: image registration and data integration. Then, we further discuss key challenges along with the potential solutions, including the assessment of fusion performance, design of effective optimization objectives, interpretability of fused multimodal datasets, and the development of user-friendly software. By summarizing current trends and discussing future perspectives, we aim to provide potential solutions that may facilitate the development of robust, reproducible, and biologically insightful computational frameworks for MSI-based multimodal fusion, with the hope of supporting researchers in gaining deeper mechanistic insights into complex biological systems.