AMASS

AMASS segments mass spectrometry imaging (MSI) datasets to identify spatially distinct molecular signatures by prioritizing molecular signal discrimination over absolute intensity.

Key Features:

• Segmentation Based on Molecular Signatures: AMASS segments regions using the discriminating power of molecular signals rather than intensity, improving identification of lower-abundance molecules and smaller segments.
• Semi-supervised Segmentation and User Supervision: AMASS implements a semi-supervised segmentation workflow that allows user input to guide and refine the segmentation process.
• Internal Consistency Validation: AMASS applies an internal consistency measure to validate segmentation results independently of pre-existing anatomical labels.
• Dynamic Libraries and Executables: The software comprises two dynamic libraries and a suite of executables for processing MSI datasets.

Scientific Applications:

• Leech Embryo Imaging: Applied to leech embryo MSI datasets, AMASS produced domains aligning with anatomical structures such as the heart, central nervous system (CNS) ganglia, and nephridia, each exhibiting unique molecular signatures.
• Rat Brain Imaging: Applied to rat brain MSI slices, AMASS identified known brain features and revealed patterns of molecular co-expression between regions.

Methodology:

AMASS uses a semi-supervised segmentation algorithm that leverages intrinsic properties of molecular signals to delineate spatial domains in MSI, focuses on signal discrimination rather than intensity to enhance detection of lower-abundance molecules and smaller segments, and employs an internal consistency measure for validation while allowing user guidance of the segmentation process.