Beside genome sequencing, quantitative proteomics is one of the most useful tools in modern biological and medical research. Mass Spectrometry (MS) allows the complete molecular characterization of proteomic samples which can be treated as images. Convolutional Neural Networks (CNN) are widely used for visual recognition tasks. This project aimed at processing MS spectra images for peptide feature detection using CNN’s ability for solving localization and semantic segmentation problems.
U-Net, with five levels of resolution, was used during this project.
For more details, take a look at the PDFs.
[1] Ronneberger, Olaf, Philipp Fischer, and Thomas Brox. "U-net: Convolutional networks for biomedical image segmentation." International Conference on Medical image computing and computer-assisted intervention. Springer, Cham, 2015.