Breast Image Processing Based on Non-subsampled Contourlet and Visual Saliency Model

Abstract

The invention discloses a breast imaging processing method based on non-subsampled contourlet transform and a visual salient model, mainly aiming at recognizing a suspicious region in view of features of breast masses. The whole system comprises an image enhancement module and an image recognition module. The image enhancement module adopts non-subsampled contourlet to enhance an original image; the image recognition module extracts featured values and generates featured images in view of the enhanced image; the featured images are processed to obtain a general salient image, and candidate suspicious masses are obtained; and according to the extracted salient image features and the enhanced image features, a machine learning method is applied to filter false positive masses so as to finally obtain a precise suspicious region.

Description

technical field [0001] The invention belongs to the technical field of image processing, relates to a medical image processing method, and can be used for identifying regions of interest in medical images. Background technique [0002] In recent decades, medical imaging has become one of the fastest-growing fields in medical technology. As a result, clinicians have a more direct and clear observation of internal lesions in the human body, and the diagnosis rate is also higher. Computer Aided Diagnosis (CAD for short) technology is called the "second pair of eyes" of doctors. It studies how to effectively process these medical image information through image processing technology to assist doctors in diagnosis and even surgical planning. , has significant social benefits and broad application prospects. Since the development of medical image processing technology as the key to computer-aided diagnosis, the crossover of various disciplines has become an inevitable trend of de...

AI Technical Summary

Problems solved by technology

The identification methods of suspicious mass areas in general breast images include simple threshold method, filter-based method, pre-segmentation to obtain the region of interest, and then extract the features of the region of interest and combine with a simple classifier for classification and identification. There are large individual differences between people, and the characteristics of tumors vary widely. It is impossible to have complete features that can represent the difference between tumors and normal areas. Therefore, no matter how good the performance of the classifier is, it is impossible to obtain ideal results.

Therefore, when the above method is used for identification, the detection rate is generally low, and the false positive rate of the mass is high.

Images