Glomerulonephritis pathology intelligent classifier training method, diagnostic instrument and storage medium

Abstract

The invention discloses a glomerulonephritis pathology intelligent classifier training method, a diagnostic instrument and a storage medium. The method is implemented on the basis of a convolutionalalneural network, and comprises the following steps: constructing a pathological intelligent classification convolutionalal neural network model of glomerulonephritis; obtaining training samples; pre-processing the training samples; inputting the pre-processed training samples into the glomerulonephritis pathological intelligent classification convolutionalal neural network model, and training theglomerulonephritis pathological intelligent classification convolutionalal neural network model, so that the glomerulonephritis pathological intelligent classification convolutionalal neural network model is trained into the glomerulonephritis pathological intelligent classifier. The glomerulonephritis pathology intelligent classifier trained by the invention can realize automatic detection of theglomerulonephritis pathological cell image and carry out specific glomerulonephritis classification, and provides a basis for the treatment of clinical glomerulonephritis.

Description

technical field [0001] The invention relates to the field of pathological analysis and diagnosis, in particular to a training method, a diagnostic instrument and a storage medium for an intelligent pathological classifier of glomerulonephritis. Background technique [0002] According to the first global kidney disease health report published by the World Congress of Nephrology (ISN) in 2017, the global prevalence of kidney disease accounts for 10% of the world's total population. The prevalence of nephropathy in my country is estimated to be about 10.08%, and the number of nephropathy patients in my country is as high as 133 million according to the population ratio of our country. There are many types of glomerulonephritis, and the etiology is complex. Only based on clinical manifestations and clinical examinations, it is often impossible to make a definite diagnosis and treatment. Renal biopsy pathological examination is usually used to display the lesion tissue through pa...

AI Technical Summary

Problems solved by technology

[0004] (1) The pathological examination of renal biopsy has particularity: in terms of method, the combination of light microscopy, immunopathology and electron microscopy is required. These requirements; technically, accurate classification of glomerulonephritis is recognized by the pathology community as one of the most difficult techniques in daily pathological work. In terms of diagnosis, it is closely integrated with the clinic, requiring pathologists to have certain clinical knowledge, and requires close cooperation and mutual penetration between clinic and pathology to make a correct diagnosis. However, most grassroots medical institutions lack kidney disease pathologists.

[0005] (2) The current pathological classification mainly relies on experience to test the pathological classification of test specimens. Such test results are relatively subjective, and pathologists with different experience will have different pathological classification results of nephritis. Doctors will also have different pathological classification results

[0006] (3) In the case that the manual inspection of nephritis specimens by pathologists is labor-intensive and prone to fatigue, the alternate inspection of the four inspection methods has brought great trouble to pathologists and also affected Improved speed and results for pathologists

Images