A computational workflow for microscopy-guided ion identification in clinical mass spectrometry imaging datasets

Matrix-assisted laser desorption/ionisation mass spectrometry imaging (MALDI-MSI) datasets were acquired from brain tissue samples obtained from living traumatic brain injury (TBI) patients. This is a proof-of-concept study which presents a computational workflow for identifying TBI-specific ions in MSI datasets through integrated pathological annotation and mass spectrometry imaging analysis. Pathological annotations of TBI regions were obtained from haematoxylin and eosin-stained microscopy images of the same sections. The microscopy images were then registered with the MSI datasets to enable precise delineation of TBI areas within the MSI data. Binary masks of the registered TBI regions were then used to perform cosine similarity searches, identifying ions potentially associated with TBI pathology. In order to relatively quantify the difference magnitude between TBI regions and surrounding tissue, bootstrap sampling was applied to non-TBI tissue areas, generating mean intensity values for comparison with average TBI region intensities. Additionally, a synthetic MSI dataset was generated to validate and optimize this analytical approach. Using this integrated computational workflow, TBI specific ions were identified in clinical MSI datasets.