[Research progress of deep learning applications in mass spectrometry imaging data analysis].

Area of Science:

• Analytical Chemistry
• Computational Biology
• Medical Imaging

Background:

• Mass spectrometry imaging (MSI) generates complex, high-dimensional data.
• Increasing data volume and complexity pose significant postprocessing challenges, including noise and registration errors.
• Deep learning (DL) offers advanced capabilities for automated data analysis and feature extraction.

Purpose of the Study:

• To review the current state of deep learning applications in mass spectrometry imaging data analysis.
• To highlight the potential of DL in overcoming MSI data challenges.
• To discuss future trends in combining AI and MSI technologies.

Main Methods:

• Review of existing literature on DL applications in MSI.
• Focus on four key stages: data preprocessing, image reconstruction, cluster analysis, and multimodal fusion.
• Illustration of DL-MSI in tumor diagnosis and subtype classification.

Main Results:

• DL effectively addresses noise, background interference, and registration deviations in MSI data.
• DL enables automated feature extraction and in-depth analysis through techniques like transfer learning.
• DL combined with MSI shows promise in medical applications, particularly in oncology.

Conclusions:

• Deep learning presents a powerful approach to enhance mass spectrometry imaging data analysis.
• DL integration is crucial for advancing MSI applications, especially in complex biological and medical studies.
• Future research should focus on further integrating AI and MSI for improved insights and applications.