Choroidal vasculature characteristics based choroid segmentation for enhanced depth imaging optical coherence tomography images. Issue 4 (11th March 2016)

• Record Type: Journal Article
• Authors: Chen, Qiang; Niu, Sijie; Yuan, Songtao; Fan, Wen; Liu, Qinghuai
• Is Part Of: Medical physics. Volume 43:Issue 4(2016)
• Page Start: 1649
• Page End: 1661
• Publication Date: 2016-03-11
• Subjects: biomedical measurement -- biomedical optical imaging -- blood vessels -- diseases -- eye -- image segmentation -- medical image processing -- thickness measurement; Visual imaging -- Segmentation -- Diseases; Measuring for diagnostic purposes; Identification of persons -- Biological material, e.g. blood, urine; Haemocytometers -- Digital computing or data processing equipment or methods, specially adapted for specific applications -- Image data processing or generation, in general; choroidal vasculature characteristics -- choroid segmentation -- EDI‐OCT image -- maximum intensity image; Medical image segmentation -- Optical coherence tomography -- Coherence imaging -- Eyes -- Medical image smoothing -- Negative resistance -- Visibility -- Optical properties -- Image enhancement; Medical physics -- Periodicals; Medical physics; Geneeskunde; Natuurkunde; Toepassingen; Biophysics; Periodicals; Periodicals; Electronic journals; 610.153
• DOI: 10.1118/1.4943382
• Languages: English
• ISSNs: 0094-2405

Abstract

Abstract : Purpose: In clinical research, it is important to measure choroidal thickness when eyes are affected by various diseases. The main purpose is to automatically segment choroid for enhanced depth imaging optical coherence tomography (EDI‐OCT) images with five B‐scans averaging. Methods: The authors present an automated choroid segmentation method based on choroidal vasculature characteristics for EDI‐OCT images with five B‐scans averaging. By considering the large vascular of the Haller's layer neighbor with the choroid‐sclera junction (CSJ), the authors measured the intensity ascending distance and a maximum intensity image in the axial direction from a smoothed and normalized EDI‐OCT image. Then, based on generated choroidal vessel image, the authors constructed the CSJ cost and constrain the CSJ search neighborhood. Finally, graph search with smooth constraints was utilized to obtain the CSJ boundary. Results: Experimental results with 49 images from 10 eyes in 8 normal persons and 270 images from 57 eyes in 44 patients with several stages of diabetic retinopathy and age‐related macular degeneration demonstrate that the proposed method can accurately segment the choroid of EDI‐OCT images with five B‐scans averaging. The mean choroid thickness difference and overlap ratio between the authors' proposed method and manual segmentation drawn by experts were −11.43 μ m and 86.29%, respectively. Conclusions: Good performance was achieved for normal and pathologic eyes, which proves that the authors' method is effective for the automated choroid segmentation of the EDI‐OCT images with five B‐scans averaging.