SR-Unet: A Super-Resolution Algorithm for Ion Trap Mass Spectrometers Based on the Deep Neural Network

Area of Science:

• Analytical Chemistry
• Spectrometry
• Artificial Intelligence in Science

Background:

• Mass spectrometry is crucial for chemical analysis, with miniature versions enabling field applications.
• Quadrupole ion traps offer miniaturization potential but are typically low-resolution.
• Balancing mass sensitivity, resolution, and scanning speed is key for ion trap mass spectrometers.

Purpose of the Study:

• To develop a super-resolution algorithm (SR-Unet) for ion trap mass spectrometry.
• To improve the equivalent resolution and detection capabilities of miniature mass spectrometers without sacrificing speed or sensitivity.

Main Methods:

• Implementation of a super-resolution U-net algorithm (SR-Unet) to estimate ions from low-resolution spectra.
• Acquisition of mass spectra using a linear ion trap mass spectrometer (LTQ XL) in Turbo and Normal scan modes.
• Application of SR-Unet to data from miniature mass spectrometers for volatile organic compound (VOC) detection and peptide analysis.

Main Results:

• Achieved unit mass resolution at 125,000 Da/s, matching the resolution of 16,667 Da/s.
• Enhanced identification of VOC species in Photochemical Assessment Monitoring Stations (PAMS) from 31 to 50.
• Improved resolution for peptide detection, reducing FWHM of bradykinin divalent ions from 0.35 to 0.15 Da (equivalent resolution of 3540) at 375 Da/s.

Conclusions:

• The SR-Unet algorithm significantly enhances the resolution and analytical speed of ion trap mass spectrometers.
• Demonstrated successful migration and application of SR-Unet to miniature mass spectrometers for field analysis.
• The method offers a new approach to improve the mixture detection capabilities of portable mass spectrometry devices.