prepIMS: a robust data preprocessing workflow for ion mobility mass spectrometry imaging

Background: Ion mobility-mass spectrometry imaging (IM-MSI) enables the spatial localization of biomolecules through two specific molecular identifiers, mass-to-charge ratio (m/z) and collision cross-section (CCS) values. IM-MSI enhances the detection of low-abundance ions and facilitates the separation of isobaric or isomeric ion species, making it particularly valuable for analyzing complex biological tissues that require higher metabolome coverage, increased analysis throughput and precise isomer separation. Preprocessing IM-MSI data is a critical step for extracting molecular features at the pixel level to reconstruct molecular images corresponding to specific m/z and CCS values, serving as a foundation for reliable downstream analysis and interpretation. However, existing preprocessing methods often struggle to effectively detect peaks from the noise-affected signals, especially for ion mobility signals with low signal-to-noise ratios. Results: We present prepIMS, a versatile and robust preprocessing workflow that incorporates a density cluster-based peak detection strategy. Experiments on the simulation dataset demonstrate that the density cluster-based peak detection in prepIMS outperforms the conventional local maxima-based strategy under noise-affected conditions, with improvements of 0.75 % in accuracy, 33.67 % in sensitivity, 17.10 % in Matthews correlation coefficient and 17.99 % in F1-score. We further validated the effectiveness of prepIMS using two whole-body mouse pup tissue datasets acquired with MALDI-TIMS, where it successfully detects and differentiates isomeric and isobaric ions within the same mass peak, underscoring its efficacy in practical applications. Furthermore, segmentation analysis of the preprocessed 4D IM-MSI data reveals that prepIMS effectively distinguishes the differences in spatial distribution from ions with subtle variations in ion mobility. Significance: prepIMS provides a powerful solution for IM-MSI data preprocessing. By effectively detecting meaningful peaks from noise-affected ion mobility signals, prepIMS facilitates the separation of isomeric and isobaric ions, and supports downstream spatial segmentation. The proposed prepIMS is expected to enhance the reliability and depth of IM-MSI data analysis, advancing data interpretation and facilitating comprehensive molecular profiling in intricate biological systems.