Medical image preprocessing method and system

Abstract

The embodiment of the invention discloses a medical image preprocessing method and system, and the method comprises the steps: constructing a medical image data set which comprises a normal image data set with normal image quality and an abnormal image data set with abnormal image quality; performing gradient calculation on the medical image data set, and respectively generating first gradient distribution corresponding to the normal image data set and second gradient distribution corresponding to the abnormal image data set; determining a gradient threshold according to the first gradient distribution and the second gradient distribution; acquiring a to-be-processed medical image, calculating a third gradient of the to-be-processed medical image, and acquiring an image category of the to-be-processed medical image according to a relationship between the third gradient and a gradient threshold; and executing corresponding operation according to the image category. According to the embodiment of the invention, before the medical picture data is input into the deep learning model for learning, the pictures with whitening or blackening quality are screened out, the noise of the pictures with poor quality is filtered out, and the training accuracy of the subsequent deep learning model is improved.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a medical image preprocessing method and system. Background technique [0002] Since the quality of DR images is affected by equipment parameters, the operating level of the operating technician, the patient's physical condition, and the patient's posture during shooting, there are significant differences in the brightness and contrast of the image imaging. When the deep model is trained, due to the excessive quality deviation of some input images, it is difficult to train the deep learning model, which affects the accuracy of model training and affects the use of the model. [0003] At present, for the detection of image quality, deep learning is usually used to construct data sets of images of different quality for classification training. The training cost of this method is high, and it has the characteristics of high false positives, that is, there is a misj...

AI Technical Summary

Problems solved by technology

[0005] In view of the above technical problems, the embodiment of the present invention provides a medical image preprocessing method and system, which can solve the problem of processing images through deep learning in the prior art, which is prone to misjudgment of normal images and cannot meet the requirements of low-quality images. Technical issues required for high detection rate and low false detection rate for high-quality images

Images