Inflammation “the fuel for cancer” - adding precision to patient stratification and therapy

Machine learning Approaches for Mass Spectrometry Imaging data to profile Prostate Cancer and Ovarian CancerThe aim of the project is to establish machine learning algorithms to analyze the large amount of imaging data from a prostate and an ovarian cancer cohort. We aim to analyze the spatial metabolomics data and provide understanding on inflammatory processes related to prostate- and ovarian cancer development.The PhD candidate has now finished all the required PhD courses and is now finishing the last part of the data processing of the large cohort of prostate cancer mass spectroscopy imaging (MSI) samples. The spatial metabolomics MSI data from the ovarian cancer tissue has been acquired in the MR Core Facility and the data is ready to be computationally analyzed. The project has been delayed by manual work on histopathology annotations digitally for each MSI sample where automatic integration from serial section has been challenging due to tissue heterogeneity. The pre-processing step has required massive work including re-alignment of about 3 million MSI spectra. The candidate is now performing the first machine learning analyses on parts of the cohort to establish the algorithms to be used for the larger cohort and the ovarian cohort. Further, he performed analysis on our previous small MSI dataset to establish a machine learning model to classify different tissue types (cancer, stroma and glands). This was presented at the Spatial Omics conference in Ghent, Belgium (14-16th June) and this is preliminary work that will be implemented within the computational pipeline.