Abdominal MRI Edge Detection Algorithm Based on Flow-XDoG Operator and Multi-Strategy Fusion

As the basis of medical image processing, edge detection technology can extract many valuable diagnostic information contained in the image, which helps to improve the efficiency and accuracy of doctor's diagnosis of diseases. However, abdominal magnetic resonance images have the characteristics of noise interference, uneven gray distribution, complex background texture, etc., and the boundaries between organs and tissues are often blurred. Traditional edge detection algorithms will generate edges when applied to such images. In order to address issues related to positioning errors and edge fractures in abdominal magnetic resonance images, a multi-strategy fusion edge detection algorithm and a flow-based extended difference of Gaussian (Flow-XDoG) operator are proposed. First, the image is denoised using bilateral filtering and fractional differential filtering while retaining the edge information of the original image. This helps suppress the influence of noise on the image. Next, the improved difference of Gaussian operator and edge tangential flow are combined, and soft threshold extension is performed to obtain the image contour line. Afterwards, the two-parameter neighborhood growth strategy and the weighted full variational edge discrimination strategy are utilized to filter out error edges generated during the extraction process. Finally, fuzzy membership is used to connect broken edges and form a closed contour. The experimental results show that the algorithm has a good edge detection effect, overcomes the disadvantage that the traditional method cannot detect the edge in the area where the gray level changes gently, and effectively solves the problem of wrong edge and broken edge through multi-strategy optimization, and the extracted contour is more complete and clear.