iSegMSI: An Interactive Strategy to Improve Spatial Segmentation of Mass Spectrometry Imaging Data

Area of Science:

• Biomedical Imaging
• Computational Biology
• Mass Spectrometry Imaging Analysis

Background:

• Spatial segmentation is crucial for mass spectrometry imaging (MSI) biochemical analysis.
• Unsupervised segmentation methods struggle with high-dimensional, low signal-to-noise MSI data.
• Precise prior knowledge, often difficult to obtain, can improve segmentation.

Purpose of the Study:

• To develop an improved method for spatial segmentation of MSI data.
• To demonstrate the utility of incorporating partial or coarse prior knowledge.
• To introduce the interactive segmentation strategy iSegMSI.

Main Methods:

• Proposed iSegMSI, an interactive strategy for MSI data segmentation.
• Incorporated prior information via scribble-regularization within an unsupervised model.
• Validated the method on whole-body mouse fetus and human thyroid cancer MSI datasets.

Main Results:

• iSegMSI effectively improved MSI segmentation results.
• The method allows subdivision or merging of regions based on user scribbles.
• Segmentation results showed high tolerance to imprecise scribbles.

Conclusions:

• iSegMSI is an effective interactive strategy for MSI data segmentation.
• Incorporating coarse prior knowledge significantly enhances segmentation accuracy.
• iSegMSI serves as a valuable preprocessing tool for MSI data analysis.