Abstract
Optical coherence tomography (OCT) has emerged as a promising imaging modality for biomedical research. In this paper, a commercial OCT system was used to differentiate two colon cell lines implanted in a chick embryo chorioallantoic membrane (CAM) model, using the attenuation coefficient of light. Malignant and non-malignant tumors were developed from RKO and NCM460 cell lines, respectively. OCT B-scan images were collected from 30 CAM models, 15 with RKO-cells and 15 with NCM460-cells. A support vector machine (SVM) classifier was trained and tested using features obtained from two different bands of interest in the depth-resolved attenuation coefficient images. Quantitative analysis revealed that the non-malignant tumors present a higher attenuation coefficient than the malignant tumors immediately after the air-tissue interface. On the other side, at a depth of 206 µm, malignant tumors exhibit higher attenuation coefficient values. Using the features median, interquartile range, and maximum of the A-scans attenuation coefficient values for two adjacent bands in-depth, the SVM nested cross-validation presents an accuracy, recall, precision, and AUC of 1.0, which gives evidence that non-malignant and malignant tumors can be distinguished with high accuracy. This work represents an important step toward the use of OCT imaging for real-time biopsy of colorectal tissues.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have