Motiro: a unified automatic framework for statistical analysis of probe-based confocal laser endomicroscopy videos of colorectal mucosa

Probe-based Confocal Laser Endomicrocopy (pCLE) enables imaging the colorectal mucosa for screening and surveillance of cancer. Analyzing acquired videos relies on subjectivity of the endomicroscopists. Quantitative criteria are needed to enhance the diagnostics obtained using pCLE. We present Motiro, an automatic framework to extract features of the colorectal mucosa imaged by pCLE. Morphometric features of the crypts of the healthy colorectal mucosa are analysed and their variability quantified using the Shannon entropy. Hellinger distance compares the statistics of a morphometric parameter in multiple mucosas (or mucosas' regions). Quantification of variability of the healthy mucosa is a prerequisite for pCLE-based early diagnostics of colorectal cancer.