Fungal keratitis image identification method based on AMBP improved algorithm

Abstract

The invention discloses a fungal keratitis image identification method based on an AMBP improved algorithm. The method comprises the following steps: performing preprocessing and binaryzation on a mycelial image and a normal cornea nerve image by use of an RX abnormity detection algorithm, and accordingly, carrying out expansion corrosion processing to reinforce mycelium and nerve characteristic information in the images; improving an AMBP algorithm, calculating a mean value of other pixels apart from a center pixel in an analysis window, and taking the mean value as a new parameter; respectively solving an average value of the pixels in the analysis window and a median pixel, calculating variances between the two and other pixels in the analysis window, and by taking the sizes of differences as newly added discrimination conditions, extracting image texture characteristics; and training a classifier by use of the extracted characteristics, identifying images obtained by a confocal microscope by use of the classifier, and identifying nerves and mycelia in the images.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to an image recognition method for fungal keratitis based on an AMBP improved algorithm. Background technique [0002] Fungal keratitis is an infectious corneal disease with a high rate of blindness caused by pathogenic fungi. In recent years, due to the extensive use of antibiotics, immunosuppressants and glucocorticoids, its incidence has increased significantly. Due to the lack of fast and effective early diagnosis methods, it often leads to misdiagnosis of the disease and delay in early treatment, which can easily cause serious consequences such as corneal perforation, hypopyon, endophthalmitis, etc., so the harm is extremely serious. Early diagnosis and treatment are particularly important. [0003] At present, the routine examination methods for fungal keratitis include corneal scraping endoscopic method, corneal biopsy, corneal scraping culture, in vivo ...

AI Technical Summary

Problems solved by technology

However, the confocal microscope method also has some shortcomings: low light level reflection of corneal tissue and eye movement often lead to low signal-to-noise ratio, strong light causes eye discomfort, and thick corneal lesions often cause blurred pictures and their examination. Expensive, etc., and the confocal microscope method is only a kind of imaging diagnosis, which is highly subjective

The acquisition of these massive medical images brings two problems: on the one hand, doctors need to spend a lot of time reading the examination images of patients, which is a very heavy work

But in its feature extraction step, the traditional texture analysis algorithm does not achieve many good results in fungal keratitis images

Images