Get Adobe Flash player
Imaging Optical Coherence Tomography – Automatic Retina Classification Through Support Vector Machines R u i B e r n a r d e s , 1,2 P e d r o S e r r a n h o , 3,4 T o r c a t o S a n t o s , 1,2 V a l t e r G o n ç a l v e s 2 a n d J o s é C u n h a - V a z 1,2 1. Association for Innovation and Biomedical Research on Light and Image, Coimbra, Portugal; 2. Institute of Biomedical Research on Light and Image, Faculty of Medicine, University of Coimbra, Coimbra, Portugal; 3. Mathematics Section, Department of Sciences and Technology, Open University, Lisbon, Portugal Abstract Optical coherence tomography (OCT) is becoming one of the most important imaging modalities in ophthalmology due to its non-invasiveness and by allowing the visualisation the human retina structure in detail. It was recently proposed that OCT data embeds functional information from the human retina. Specifically, it was proposed that blood–retinal barrier status information is present within OCT data from the human retina. Besides this ability, the authors present data supporting the idea of having the OCT data encoding the ageing of the retina in addition to the disease (diabetes) condition from the healthy status. The methodology followed makes use of a supervised classification procedure, the support vector machine (SVM) classifier – based solely on the statistics of the distribution of OCT data from the human retina (i.e. OCT data between the inner limiting membrane and the retinal pigment epithelium). Results achieved suggest that information on both the healthy status of the blood–retinal barrier and on the ageing process co-exist encoded within the optical properties of the human retina. Keywords Optical coherence tomography, support vector machines, supervised classification, retina, diabetic retinopathy, ageing Disclosure: The publication of this article was supported in part by the Fundação para a Ciência e a Tecnologia (FCT) under the research project PTDC/SAU-BEB/103151/2008 and program COMPETE (FCOMP-01-0124-FEDER-010930). Acknowledgements: The authors would like to thank Dr Melissa Horne and Carl Zeiss Meditec (Dublin, CA, US) for their support on getting access to OCT data and AIBILI Clinical Trial Centre technicians for their support in managing data, working with patients and performing scans. Received: 12 October 2011 Accepted: 16 January 2012 Citation: European Ophthalmic Review, 2012;(6)4:200–3 Correspondence: Rui Bernardes, Association for Innovation and Biomedical Research on Light and Image (AIBILI), Azinhaga de Santa Comba, Celas, 3000-548 Coimbra, Portugal. E: rcb@aibili.pt Both the ageing process and diabetes promote changes on the human retina, although not always visible through the regular eye fundus examination. The authors’ research group has been focused on imaging changes within the human retina of diabetic patients aiming for better characterisation and on detection at the very early stages even when these cannot be detected in the eye fundus. Current trends in medical imaging point to the increasing use of non-invasive techniques. In this sense, the authors started focusing our efforts in assessing the possibility of gathering information from the eye fundus through non-invasive techniques. Nevertheless, these are required to provide the same or even higher levels of information than the one currently provided. A particularly interesting non-invasive technique in use in the field of ophthalmology is the optical coherence tomography (OCT). This imaging technique is spreading quickly and, in consequence, is becoming available in multiple eye care facilities. The authors research group has been interested in diabetic retinopathy with a special focus on the breakdown of the blood–retinal barrier (BRB) in consequence of diabetes. 1–3 The number of diabetic patients 200 is increasing worldwide 4 and this multifactorial disease has a large social and economic impact in the active working population. 5 Efforts made towards the assessment of the BRB function led the authors research group to the development of the retinal leakage analyser (RLA). 6,7 This imaging modality, based on the confocal imaging of the eye fundus after sodium fluorescein (NaFl) administration, led to a better characterisation of the changes occurring in the diabetic retina. 8 Additionally, this knowledge made possible the identification of different phenotypes of diabetic retinopathy progression, 9 showing over time the significantly different rates of progression in the diabetic retina and the variability between diabetic patients. Both instruments used in Lobo et al. 6 and Bernardes et al. 7 were modified for specific purposes. The former instrument was a Carl Zeiss confocal scanning laser ophthalmoscope (CSLO) prototype (Carl Zeiss, Jena, Germany) modified for the task at hand. The latter was a CSLO (Heidelberg Retina Angiograph [HRA], Heidelberg Engineering, Heidelberg, Germany) modified to increase confocality and to use a different positioning of the confocal planes within the ocular fundus to scan through the retina and vitreous. © TOUCH MEDICAL MEDIA 2012