Extracting Morphological and Sub-Resolution Features from Optical Coherence Tomography Images, a Review with Applications in Cancer Diagnosis

Abstract

Before they become invasive, early cancer cells exhibit specific and characteristic changes that are routinely used by a histopathologist for diagnosis. Currently, these early abnormalities are only detectable ex vivo by histopathology or, non-invasively and in vivo, by optical modalities that have not been clinically implemented due to their complexity and their limited penetration in tissues. Optical coherence tomography (OCT) is a noninvasive medical imaging technology with increasing clinical applications in areas such as ophthalmology, cardiology, gastroenterology, etc. In addition to imaging the tissue micro-structure, OCT can also provide additional information, describing the constituents and state of the cellular components of the tissue. Estimates of the nuclear size, sub-cellular morphological variations, dispersion and index of refraction can be extracted from the OCT images and can serve as diagnostically useful biomarkers. Moreover, the development of fully automated algorithms for tissue segmentation and feature extraction and the application of machine learning, can further enhance the clinical potential of OCT. When fully exploited, OCT has the potential to lead to accurate and sensitive, image-derived, biomarkers for disease diagnosis and treatment monitoring of cancer.