Multi-parameter mri prostate cancer cad method and system based on two classifiers

Abstract

The invention discloses a multi-parameter MRI prostate cancer CAD method and system based on two classifiers, and relates to the field of medical image processing. The method includes the following steps: automatic detection of candidate lesions: three MRI sequences of each case: T2WI, DWI, ADC, are preprocessed respectively to make them have the same resolution and size, and the pixels at the same position basically correspond to those in the human body. The same site; extract point features on the three MRI sequences of each case, and input these point features into the lesion detection classifier to obtain candidate lesions; computer-aided diagnosis of candidate lesions: calculate the candidate lesions among the three MRI sequences of each case The regional characteristics of the lesion are input into the lesion diagnosis classifier to obtain the corresponding diagnosis results. The invention can provide radiologists with a series of quantitative indexes and corresponding malignant probability values.

Description

technical field [0001] The invention relates to the field of medical image processing, in particular to a multi-parameter MRI prostate cancer CAD method and system based on two classifiers. Background technique [0002] PCa (Prostate Cancer, prostate cancer) is the second most common cancer in men in the world. In the United States, about 1 / 6 men will get prostate cancer, and 1 / 36 men will die from this disease. At present, the diagnostic methods for prostate cancer include: TRUS (Trans-rectal Ultrasound, ultrasound-guided transrectal) prostate biopsy and PSA (Prostate-specific Antigen, prostate-specific antigen) serum examination. A prostate biopsy can be very uncomfortable for the patient and serious infections can occur. Prostate-specific antigen is specific to prostate tissue, but not to prostate cancer. There is considerable overlap in serum PSA results between prostate cancer and benign prostatic hyperplasia. [0003] Different from the above two methods, multi-param...

AI Technical Summary

Problems solved by technology

[0006] At present, there are some methods to extract pixel features. These methods generally use the statistical features in a local window centered on a certain pixel as the feature of the point. The disadvantage of these methods is that the pixel features used cannot describe the whole picture of the entire lesion area. Not as well understood by physicians as regional traits

[0007] Other methods extract features of lesion candidate areas, such as size, shape, texture, and asymmetry. The disadvantage is that cancer lesions in the prostate area need to be obtained by manual segmentation by experienced doctors. In addition to texture features, more effective features are needed to describe the lesion area of ​​prostate cancer

Images