A priori segmentation mediated local depth-resolved optical attenuation coefficient mapping of RPE complex for the investigation of age-related macular degeneration

In the study of age-related macular degeneration (AMD), optical coherence tomography OCT has proven invaluable in identifying biomarkers that assess disease progression. However, Optical coherence tomography (OCT) signals are easily affected by light intensity, which has led to criticism in clinical applications. Optical attenuation coefficient (OAC) provides local tissue properties without the influences of the OCT system and accumulated light attenuation, which is useful for quantifying pathological changes by enhancing the differentiation between diseased and normal tissues. Previous methods of OAC for retinal imaging have not considered the influence of the backscattering fraction, which varies across different tissues. In this paper, we proposed a priori segmentation mediated local depth-resolved (LDR) algorithm to improve the accuracy and stability of retinal pigment epithelium (RPE) layer OAC calculations. The LDR algorithm utilizes a new method for estimating residual light intensity, which replaces the conventional approaches of accumulating all residual signals or using numerical fitting methods. This improvement enables the LDR algorithm to perform OAC conversion at any depth range in OCT images (including RPE), regardless of whether the light energy is completely attenuated. When there are drusen between the RPE and BM that are difficult to segment precisely, we use a backscattering compensation method to make the RPE and drusen have the same backscattering coefficient, and then use the LDR algorithm to calculate the OAC of the RPE layer. The algorithm was evaluated using phantoms, healthy subjects, and patients with early atrophic AMD, demonstrating its reliability and practicality.