Effect of Treatment Using 3-Dimentional Disease Generating Model on Optical Coherence Tomography Images

Abstract

Automatic extraction of retinal border lines for optical coherence tomography (OCT) images is important for assisting ophthalmologists in clinical decision making in terms of both diagnosis and treatment. 3D-OCT consists of 128 2D-OCT images representing cross-sectional structure. OCT has proven clinically useful for diagnosing a variety of retinal diseases. Some ophthalmologists desire the automatic measurement of a retinal thickness and its quantitative evaluation. Previously, the automatic measurement methods of the retinal thickness have been reported for retinal OCT images. However, the methods extracted the retinal border lines intermittently for OCT images, but they cannot measure the effect of treatment for an abnormal area automatically. And it’s difficult in evaluating the effectiveness of laser treatment or medication.

In this paper, we propose a new three dimensional (3D) disease generating model to display the effect of treatment. To extract the change of the abnormal area, we applied the automatic measurement method of a retinal thickness in OCT image, which is the dynamic contour model (“One Directional Active Net (ODAN)”) proposed by our research group to extract two retinal border lines, which are Inner Limiting Membrane (ILM) and Retinal Pigment Epithelium (RPE). The proposed generating model generates the border line of disease based on ILM, and employs the temporal subtraction technique to detect the effect of treatment. We confirmed the usefulness of the proposed model using the generated disease OCT images from normal retinal OCT images. We are considering that the proposed method is useful as the visual evaluation for the effect of treatment for retinal disease.