outpatient clinic of the department of internal medicine at the AMC, underwent full ophthalmic examination ... CUNHA VAZ J 1, 2, 3 . (1) AIBILI - Association ...
FREE PAPERS Retina : Retinal Imaging 2161 Quantification of Drusen Deposits using Image Processing Techniques MORA A 1, 2, FONSECA J 1, 2, VIEIRA P 2 (1) Intelligent Robotics Center, Uninova, Caparica (2) Faculty of Sciences and Technologies, Universidade Nova de Lisboa, Lisbon Purpose The automatic analysis of Drusens quantitative evolution can be a useful tool for ophthalmologists that are currently analyzing it visually in a not reproducible process. In this paper we propose a methodology for automatic quantification of Drusens in retina images using image processing techniques. Methods Often, these images have a non-uniform illumination that produces different contrast areas. In this work we propose an algorithm that compensates the image nonuniform illumination. In a second step we propose a innovative approach for Drusens detection and quantification. It consists in the creation of a tri-dimensional model of each Drusen spot, allowing areas and estimated volumes to be computed. The algorithm starts by localizing spots and determining their influence areas. Finally, using the image intensity as tri-dimensional information, the spots detected on the previous step are modeled using tri-dimensional functions. Results The results of applying the methodology to retina images compared with images marked by ophthalmologists are presented. The comparing images were marked through a semi-automatic procedure in an application specifically developed for this purpose that is now being used on clinical practice.
2162 Early neuroglial loss in DM type 1 patients, an OCT study VERBRAAK FD 1, 2, BIALLOSTERSKI C 1, VAN VELTHOVEN ME 1, SCHLINGEMANN RO 1, MICHELS BPJ 3, DE VRIES HH 3 (1) Academic Medical Center, Dept of Ophthalmology, Amsterdam (2) Academic Medical Center, Laser Center, Amsterdam (3) Academic Medical Center, Dept of Internat Medicine, Amsterdam Purpose To confirm the hypothesis of early neuroglial loss in diabetic retinopathy (DR) we compared retinal thickness (RT), measured with Optical Coherence Tomography (OCT) in patients with type 1 diabetes (DM), with no or minimal DR, to healthy controls. Methods Fifty-three patients with type 1 DM with no or minimal DR, seen at the outpatient clinic of the department of internal medicine at the AMC, underwent full ophthalmic examination, fundus photography and OCT. Mean RT was calculated for the central fovea, the fovea, the pericentral and the peripheral area of the macula and compared to normal controls. Results Mean RT in the pericentral area was lower in patients with minimal DR (267µm ± 20µm; n=23, 43%) compared to healthy subjects (281µm ± 13µm; P value = 0.005, 95% CI = -23.10, -4.46; n=28). The mean pericentral RT in patients without DR (276µm ± 14µm; n=30, 57%) was less than pericentral RT in normal controls, but more than in patients with minimal DR, but did not differ significantly from either one. None of the other regions showed a significant change.
Conclusion We present a new methodology for quantitative analysis of Drusens on retina images that is reproducible and independent of visualization conditions. This software will allow quantitative studies about Drusens evolution along treatments to be elaborated.
Conclusion In this study a significantly decreased pericentral RT was measured in patients with minimal DR compared to healthy controls. This could be explained by a loss of intraretinal neural tissue in the earliest stage of DR.
2163 Establishment of a normative database by registration of RTA maps into the RT-Atlas
2164 Increased OCT Mapping Resolution by Merging Line and Circle Scan-Types
BAPTISTA P 1, SANTOS T 1, NUNES S 1, BERNARDES R 1, CUNHAVAZ J 1, 2, 3 (1) AIBILI - Association for Innovation and Biomedical Research on Light and Image, Coimbra (2) Center of Ophthalmology, Coimbra University Hospital, Coimbra (3) Faculty of Medicine, University of Coimbra, Coimbra
SANTOS T 1, BAPTISTA P 1, BERNARDES R 1, CUNHAVAZ J 1, 2, 3 (1) AIBILI - Association for Innovation and Biomedical Research on Light and Image, Coimbra (2) Center of Ophthalmology, Coimbra University Hospital, Coimbra (3) Faculty of Medicine, University of Coimbra, Coimbra
Purpose To establish an improved retinal thickness normative database. Methods Retinal thickness measurements by RTA II (Retinal Thickness Analyzer, Talia, Israel) were obtained from 27 eyes from 20 healthy volunteers aged from 40 to 64 years (mean±SD: 53.3±5.9 yrs). Each retinal thickness map was registered into the retinal thickness atlas (RT-Atlas) by considering translations and rotation, thus becoming into alignment with the RT-Atlas space, where the horizontal line connects the center of the fovea to the center of the optic disc and the fovea is at the origin of the coordinates system. While the RT-Atlas is established using principal components analysis, this normative database is established at the RT-Atlas space coordinates by thin-plate spline interpolation to compute the average, standard deviation (SD) and the 95% confidence interval (CI). Results Values for the established retinal thickness normative database range from 147.1 μm (95% CI = [139.0-155.2]) to 219.1 μm (95% CI = [211.8-226.4]), while the standard deviation range from 13.9 μm to 24.1 μm (SD/mean from 6.8% to 14.8%). Conclusion The established normative database in association to the registration of thickness maps into the RT-Atlas space, allows to better identify retinal thickness changes from the control healthy population and to better follow retinal changes over time.
Purpose To increase retinal thickness mapping resolution of the Optical Coherence Tomograph (OCT - Humphrey-Zeiss, USA). Methods A retinal thickness Atlas (RT-Atlas) of the human macula was established based on a series of 33 retinal thickness maps from the Retinal Thickness Analyzer (RTA II, Talia, Israel), from 20 healthy volunteers. A principal component analysis was performed to extract the major retinal thickness characteristics to build the RT-Atlas. Two scan-types were performed using the Stratus OCT: a radial scan set and a circular scan set. The radial scan set (six-radial scans - fast scan mode) are initially registered to the RT-Atlas by an iterative optimization procedure to compute the global position and orientation for the entire set, followed by individually finding the best fit of each linescan in the neighbourhood of the computed global parameters. A new retinal thickness map is then built based on the OCT data (now registered to the RT-Atlas, thus having an origin centered on the fovea and the horizontal axis pointing to the center of the optic disc). The set of circular scans (fast scan mode) is now registered into this new map following a similar procedure, i.e., global and individual fits, thus achieving the registration of both radial and circular scans. Results This method allows for the registration of radial and circular OCT scans to achieve an increased retinal thickness mapping resolution, and was applied to both healthy and diseased retinas. Conclusion This new map presents an increased resolution compared to the original OCT mapping and proved to be useful for detecting small retinal thickness changes.
92
EVER 2006 - Abstract book
