High-reflection bright spot segmentation and quantification method in spectral domain optical coherence tomography

The invention discloses a high-reflection bright spot segmentation and quantification method in spectral domain optical coherence tomography. The method is composed of two parallel processes of ROI generation and HRFs estimation. The method comprises the steps: regarding the region of interest as an inner boundary membrane (ILM) and inner and outer segment (ISOS) layers, wherein HRFs are mainly located in the OCT image; obtaining a reconstructed image by using morphological reconstruction (MR), and constructing a histogram for data distribution and clustering. extracting the extreme value region from the connected region obtained from the component tree to estimate the HRFs; and finally, combining the generated investment return rate with the estimated HRFs process to obtain segmented HRFs. According to the method, the high-reflection bright spots (HRFs) of the OCT image can be clearly segmented and quantified, so the purposes of detection, visibility, segmentation and quantification are achieved. 本发明公开了一种光谱域