Colorectal cancer (CRC) is one of the top causes of malignancy in both men and women. Although screening has significantly reduced CRC mortality, colonoscopy suffers from inadequate inspection and sampling of the tissue, a limitation that could be addressed by Optical Coherence Tomography (OCT). However, thus far, most studies have concentrated on the qualitative evaluation of morphological features and, only recently, the automatic classification of OCT images is being explored. To improve the classification of human tissues, manual or automatic, the spectral information in the OCT interferogram can be exploited. It can provide additional information regarding disease-related absorption and/or scattering changes in the tissue. In this study, we propose the use of multi-spectral analysis of OCT images, i.e. the utilization of images created from different bands of the available spectrum, to classify human colon polyps as normal or abnormal. Multiple, narrow-band, images, at different center wavelengths, were combined to create a “spectral score” for each pixel of the image. This fusion of information allowed both easier visual evaluation of the images as well as automatic classification (80 % accuracy per patient with leave-one-patient-out cross-validation). The proposed approach must be expanded to include more polyps and explore more sophisticated multi-spectral deep learning methods to improve its accuracy. However, these preliminary results provide evidence that this method has the potential to improve the accuracy of OCT and, in the future, enable clinical applications for colon cancer diagnosis.
|