Automated segmentation of macular layers in OCT images and quantitative evaluation of performances

• Published in: Pattern recognition Vol. 44; no. 8; pp. 1590 - 1603
• Main Authors: Ghorbel, Itebeddine, Rossant, Florence, Bloch, Isabelle, Tick, Sarah, Paques, Michel
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.08.2011; Elsevier
• Subjects: Algorithms; Applied sciences; Automated; Automated segmentation; Biological and medical sciences; Computer Science; Computer Vision and Pattern Recognition; Computerized, statistical medical data processing and models in biomedicine; Detection, estimation, filtering, equalization, prediction; Diseases; Exact sciences and technology; Image processing; Information, signal and communications theory; Investigative techniques of ocular function and vision; Investigative techniques, diagnostic techniques (general aspects); Medical Imaging; Medical management aid. Diagnosis aid; Medical sciences; Optical coherence tomography; Quantitative analysis; Quantitative evaluation; Retinal images; Retinal imaging; Segmentation; Signal and communications theory; Signal and Image Processing; Signal processing; Signal, noise; Telecommunications and information theory; Thickness measurements; Optical coherence tomography; Quantitative evaluation; Automated segmentation; Retinal imaging; Performance evaluation; High resolution; Disease; Segmentation; Image processing; Kalman filter; Photoreceptor; Retina; Random field; Imaging; Testing equipment; Active contour; Quantitative analysis; Image evaluation; Probabilistic approach; Ophthalmoscopy; Markov model; Algorithm; In vivo; Algorithm performance; Quality control; Medical imagery; Diagnostic aid; Edge detection; Optical coherence tomography, Retinal imaging, Automated segmentation, Quantitative evaluation
• ISSN: 0031-3203; 1873-5142
• DOI: 10.1016/j.patcog.2011.01.012

Optical coherence tomography (OCT) allows high-resolution and noninvasive imaging of the structure of the retina in humans. This technique revolutionized the diagnosis of retinal diseases in routine clinical practice. Nevertheless, quantitative analysis of OCT scans is yet limited to retinal thickness measurements. We propose a novel automated method for the segmentation of eight retinal layers in these images. Our approach is based on global segmentation algorithms, such as active contours and Markov random fields. Moreover, a Kalman filter is designed in order to model the approximate parallelism between the photoreceptor segments and detect them. The performance of the algorithm was tested on a set of retinal images acquired in-vivo from healthy subjects. Results have been compared with manual segmentations performed by five different experts, and intra and inter-physician variability has been evaluated as well. These comparisons have been carried out directly via the computation of the root mean squared error between the segmented interfaces, region-oriented analysis, and retrospectively on the thickness measures derived from the segmentations. This study was performed on a large database including more than seven hundred images acquired from more than one hundred healthy subjects. ► We propose a novel method for segmenting eight layers in OCT retinal images. ► It is based on global methods to overcome the limits of using only local information. ► We model the parallelism between the layers and introduce it in a Kalman filter. ► We validate our algorithm on a large database, from different acquisition devices. ► The accuracy is very good, within the range of intra and inter-physician variability.