A Method of Medical Image Fusion Based on Regional Contrast

Abstract

The invention discloses a medical image fusion method based on regional contrast, which belongs to the field of medical image processing. The method includes the following steps: a. Obtain registered multi-modal medical images, formulate a blocking strategy according to the size of the multi-modal medical images, and perform blocking operations on the images according to the principles of top to bottom and left to right; b. , Calculate the contrast of each pixel under different blocks and the histogram matrix of the block image, and use the histogram matrix as a coefficient to generate a regional contrast matrix for each pixel; c. Apply the typical correlation analysis method to calculate each pixel area For the correlation coefficient of the contrast matrix, solve for the coefficient matrix when the contrast matrix of the pixel area is most relevant; d. Use the obtained coefficient matrix as the weight to generate a fused image. Compared with the existing technology, the medical image fusion method based on regional contrast of the present invention has the characteristics of good fusion effect and high execution efficiency, and has good promotion and application value.

Description

technical field [0001] The invention relates to the field of medical image processing, in particular to a medical image fusion method based on region comparison. Background technique [0002] With the development of imaging technology, medical images have become an important means of modern medical diagnosis and treatment. At present, medical images commonly used in clinical practice are mainly divided into anatomical images and functional images. Anatomical images (such as CT, MR, etc.) can clearly provide the anatomical structure and morphological characteristics of organs, and have high resolution; functional images (such as PET, fMRI, etc.) can accurately provide organ function and metabolism information, but their The resolution is lower. The image of a single modality can only provide some characteristics of the patient, but cannot reflect all the pathological information. Therefore, in clinical practice, it is often necessary to organically integrate complementary ...

AI Technical Summary

Problems solved by technology

The spatial domain method fuses the pixel values ​​themselves by weighted average or morphological processing, which is simple in calculation and poor in fusion effect

The transform domain method first performs pyramid, wavelet or multi-scale geometric transformation on the medical image, extracts the subband coefficients of the transform domain, formulates different fusion rules for the subband coefficients, and finally generates the fused image through inverse transformation. The fusion effect is good, but the calculation High complexity and poor real-time performance

The intelligent fusion method iteratively calculates fusion coefficients through algorithms such as neural network, statistical learning, and fuzzy sets to generate fusion images. The fusion effect often depends on the design of the learning algorithm, and there are also problems such as high computational complexity and poor real-time performance.

Generally speaking, the existing pixel-level fusion methods are based on the structure, shape, color, time-frequency information and other characteristics of multi-mode medical images to formulate fusion strategies, lacking the analysis of factors such as pixel spatial position relationship and contrast; at the same time, Existing methods generally adopt a weighted average or a large absolute value as the fusion rule, which tends to cause the fusion result to fall into the global optimum, resulting in the loss or blurring of local details, which is difficult to meet the needs of practical applications.

Images