14 February 2018 Automated detection of esophageal dysplasia in in vivo optical coherence tomography images of the human esophagus
Author Affiliations +
Abstract
Catheter-based Optical Coherence Tomography (OCT) devices allow real-time and comprehensive imaging of the human esophagus. Hence, they provide the potential to overcome some of the limitations of endoscopy and biopsy, allowing earlier diagnosis and better prognosis for esophageal adenocarcinoma patients. However, the large number of images produced during every scan makes manual evaluation of the data exceedingly difficult. In this study, we propose a fully automated tissue characterization algorithm, capable of discriminating normal tissue from Barrett’s Esophagus (BE) and dysplasia through entire three-dimensional (3D) data sets, acquired in vivo. The method is based on both the estimation of the scatterer size of the esophageal epithelial cells, using the bandwidth of the correlation of the derivative (COD) method, as well as intensity-based characteristics. The COD method can effectively estimate the scatterer size of the esophageal epithelium cells in good agreement with the literature. As expected, both the mean scatterer size and its standard deviation increase with increasing severity of disease (i.e. from normal to BE to dysplasia). The differences in the distribution of scatterer size for each tissue type are statistically significant, with a p value of < 0.0001. However, the scatterer size by itself cannot be used to accurately classify the various tissues. With the addition of intensity-based statistics the correct classification rates for all three tissue types range from 83 to 100% depending on the lesion size.
© (2018) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Michalis Kassinopoulos, Jing Dong, Guillermo J. Tearney, Costas Pitris, "Automated detection of esophageal dysplasia in in vivo optical coherence tomography images of the human esophagus", Proc. SPIE 10483, Optical Coherence Tomography and Coherence Domain Optical Methods in Biomedicine XXII, 104830R (14 February 2018); doi: 10.1117/12.2289612; https://doi.org/10.1117/12.2289612
PROCEEDINGS
6 PAGES


SHARE
Back to Top