Digital stereo image analyzer for generating automated 3-D measures of optic disc deformation in glaucoma.

The major limitations of precise evaluation of retinal structures in present clinical situations are the lack of standardization, the inherent subjectivity involved in the interpretation of retinal images, and intra- as well as interobserver variability. While evaluating optic disc deformation in glaucoma, these limitations could be overcome by using advanced digital image analysis techniques to generate precise metrics from stereo optic disc image pairs. A digital stereovision system for visualizing the topography of the optic nerve head from stereo optic disc images is presented. We have developed an algorithm, combining power cepstrum and zero-mean-normalized cross correlation techniques, which extracts depth information using coarse-to-fine disparity between corresponding windows in a stereo pair. The gray level encoded sparse disparity matrix is subjected to a cubic B-spline operation to generate smooth representations of the optic cup/disc surfaces and new three-dimensional (3-D) metrics from isodisparity contours. Despite the challenges involved in 3-D surface recovery, the robustness of our algorithm in finding disparities within the constraints used has been validated using stereo pairs with known disparities. In a preliminary longitudinal study of glaucoma patients, a strong correlation is found between the computer-generated quantitative cup/disc volume metrics and manual metrics commonly used in a clinic. The computer generated new metrics, however, eliminate the subjective variability and greatly reduce the time and cost involved in manual metric generation in follow-up studies of glaucoma.

On the Characteristics of the Autoassociative Memory with Nonzero-Diagonal Terms in the Memory Matrix.

A statistical method is applied to explore the unique characteristics of a certain class of neural network autoassociative memory with N neurons and first-order synaptic interconnections. The memory matrix is constructed to store M = αN vectors based on the outer-product learning algorithm. We theoretically prove that, by setting all the diagonal terms of the memory matrix to be M and letting the input error ratio ρ = 0, the probability of successful recall Pr steadily decreases as α increases, but as α increases past 1.0, Pr begins to increase slowly. When 0 < ρ ≤ 0.5, the network exhibits strong error-correction capability if α ≤ 0.15 and this capability is shown to rapidly decrease as α increases. The network essentially loses all its error-correction capability at α = 2, regardless of the value of ρ. When 0 < ρ ≤ 0.5, and under the constraint of Pr > 0.99, the tradeoff between the number of stable states and their attraction force is analyzed and the memory capacity is shown to be 0.15N at best.