Inspection image quality evaluation method, device and equipment and storage medium

Abstract

The invention discloses an inspection image quality evaluation method. The method comprises the following specific steps: A, acquiring an inspection image to be analyzed; B, preprocessing the inspection image according to the actual scene to enable the inspection image to accord with subsequent processing; C, performing quality detection on the inspection image by adopting an image algorithm basedon a neural network; obtaining a total score of the inspection image quality and various index values influencing the total score; and D, classifying the samples into a sample library according to aquality detection result to realize standardized management of inspection image data, so that quality evaluation can be performed on the inspection image or a certain target or a certain type of target on the image, the inspection image with unqualified quality is filtered out, and meanwhile, unqualified indexes and quantitative parameters are given. The problem that objective comparison is difficult to carry out between different data sets and between different auxiliary diagnosis algorithms is solved, so that the stability and consistency of doctors and an auxiliary diagnosis and treatment system are improved, the reliability and accuracy of diagnosis results are ensured, and misdiagnosis and missed diagnosis are reduced.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a method, device, equipment and storage medium for evaluating the quality of inspection images. Background technique [0002] After thousands of years of practice and development, the four diagnostic methods of traditional Chinese medicine have become a systematic and complete diagnostic method; as one of the four diagnostic methods, the inspection of the tongue, represented by tongue diagnosis, requires doctors to observe and diagnose according to visual information under a standard light source At the same time, under the guidance of the theory of traditional Chinese medicine, combined with the three diagnostic information of auscultation, interrogation, and palpation, the inspection is carried out to eliminate the false and preserve the true, and to judge the basic health status and disease symptoms, so as to guide the prescription of medication and judge the curative ...

AI Technical Summary

Problems solved by technology

[0003] For a long time, the experience and theory of inspection diagnosis and treatment have been mainly based on the observation and analysis of individual doctors, while the results of doctor inspection and diagnosis are affected by subjective and objective factors such as individual sensory ability, personal experience, fatigue level, personal emotion, etc. The distribution of resources is uneven, and traditional diagnosis and treatment methods are prone to missed and misdiagnosed cases

[0004] In recent years, with the improvement of computer computing power and the stepwise growth of data volume, the deep learning technology based on neural network has developed rapidly and has been widely used in the medical field; A lot of research work has been done on the computer-aided diagnosis system, but not all the new methods are better than the old ones, and some methods are not effective after a period of clinical application, and are even eliminated; auxiliary diagnosis methods with uncertain effects not only increase the medical The cost may also cause harm to patients due to misdiagnosis and missed diagnosis. The fundamental reason is that all inspection-assisted diagnosis methods are based on inspection images, and inspection image acquisition will be affected by subjective and objective factors, resulting in uneven image quality. It is difficult for auxiliary diagnosis methods or doctors to distinguish whether the image quality is qualified or not, resulting in misdiagnosis and missed diagnosis

[0005] At the same time, the neural network-based deep learning-assisted diagnosis algorithm is based on a large amount of data and is driven by data. However, the algorithms adopted by different research institutions are based on the data sets they collect, and the amount of data between different data sets Data quality varies, making objective comparisons between methods difficult

Images