Reconstruction and Feature Selection for Desorption Electrospray Ionization Mass Spectroscopy Imagery

Area of Science:

• Analytical Chemistry
• Medical Imaging
• Computational Biology

Background:

• Desorption electrospray ionization mass spectrometry (DESI-MS) offers sensitive molecular-level tissue imaging.
• Accurate tumor boundary delineation is crucial for intra-operative surgical decisions.
• Current imaging and analysis times can be critical limiting factors in surgical settings.

Purpose of the Study:

• To accelerate DESI-MS imaging for intra-operative tumor margin assessment.
• To reduce data acquisition and analysis time while maintaining diagnostic accuracy.
• To identify key biochemical molecules for pathological analysis.

Main Methods:

• Utilized compressive sensing for sparse sampling, reducing tissue scanning time by 50%.
• Implemented sparse feature selection to decrease data dimensionality from ~10^4 to <50.
• Validated the combined methods on brain and breast tumor datasets.

Main Results:

• Achieved significant reduction in both imaging and data analysis times.
• Successfully identified biochemically important molecules relevant for pathological analysis.
• Demonstrated the efficacy of the methods on clinical tumor samples.

Conclusions:

• The developed approach significantly enhances the speed of DESI-MS for intra-operative applications.
• Compressive sensing and sparse feature selection are effective in accelerating molecular imaging analysis.
• This technique holds promise for improving surgical decision-making by providing rapid, molecular-level tissue information.