Tissue biomolecular and microstructure profiles in optical colorectal cancer delineation

Colorectal cancer (CRC) is the 3rd most common and the 2nd most deadly type of cancer worldwide. Understanding the biochemical and microstructural aspects of carcinogenesis is a critical step towards developing new technologies for accurate CRC detection. To date, optical detection through analyzing tissue chromophore concentrations and scattering parameters has been mostly limited to chromophores in the visible region and analytical light diffusion models. In this study, tissue parameters were extracted by fitting diffuse reflectance spectra (DRS) within the range 350–1900 nm based on reflectance values from a look-up table built using Monte Carlo simulations of light propagation in tissues. This analysis was combined with machine learning models to estimate parameter thresholds leading to best differentiation between mucosa and tumor tissues based on almost 3000 DRS recorded from fresh ex vivo tissue samples from 47 subjects. DRS spectra were measured with a probe for superficial tissue and another for slightly deeper tissue layers. By using the classification and regression tree algorithm, the most important parameters for CRC detection were the total lipid content ( f lipid ), the reduced scattering amplitude ( α ′), and the Mie scattering power ( b Mie ). Successful classification with an area under the receiver operating characteristic curve higher than 90% was achieved. To the best of our knowledge, this is the first study to evaluate the potential tissue biomolecule concentrations and scattering properties in superficial and deeper tissue layers for CRC detection in the luminal wall. This may have important clinical applications for the rapid diagnosis of colorectal neoplasia.