Method for improving prostate tumor MRI (Magnetic Resonance Imaging) image identification rate based on CAD (Computer-Aided Diagnosis) system

Abstract

The invention belongs to the field of prostate disease medical equipment and particularly relates to a method for improving a prostate tumor MRI (Magnetic Resonance Imaging) image identification rate based on a CAD (Computer-Aided Diagnosis) system. The method comprises the following steps: (1) collecting an MRI image of a prostate patient; (2) extracting MRI prostate tumor ROI region features; (3) carrying out feature-grade fusion on the ROI region features; and (4) identifying the fused features by classification through taking a neural network as a classification device. By virtue of the method, the capabilities of identifying the prostate benign and malignant tumors are at least improved by 10%, and the method has active meanings on the CAD of the MRI prostate tumors.

Description

technical field [0001] The invention belongs to the field of medical devices for prostate diseases, in particular to a method for improving the recognition rate of prostate tumor MRI images based on a CAD system. Background technique [0002] Prostate tumors mainly include prostate cancer and benign prostatic hyperplasia, both of which occur in the prostate. Prostate cancer is a common malignant tumor. Under normal circumstances, benign prostatic hyperplasia will not turn into prostate cancer. Prostate cancer is usually diagnosed clinically by abnormal digital rectal examination (DRE) and elevated serum prostate-specific antigen (PSA). Once prostate cancer is discovered, its treatment mainly depends on tumor histological classification and clinical staging. The TNM staging and Gleason staging of prostate cancer clearly define the extent of the lesion, and the correct staging is very important for determining whether surgery is possible, choosing treatment methods, and judgin...

AI Technical Summary

Problems solved by technology

From a technical point of view, the CAD method based on medical images is a target recognition technology, but the existing target recognition methods are only divided according to a single sample feature vector, and the high-dimensional pattern characteristics exhibited by a large number of similar samples are not fully considered.

Therefore, the reliability of the recognition results obtained by extracting only a few or more than a dozen dimensional features from the MRI prostate tumor ROI area is not high.

Although there are relevant literatures that have discussed the CAD of prostate tumors to a certain extent, compared with other human organs (such as breast, brain, etc.), there are fewer reports on prostate CAD, and these conclusions are only based on small samples and limited extraction features. It is obtained under certain conditions, but there is a common disadvantage of high false positive rate, and there is a lack of more convincing large-sample prospective studies

Images