Human body anatomical structure similarity-based medical image compression algorithm

Abstract

The invention relates to a human body anatomical structure similarity-based medical image compression algorithm. According to the algorithm of the invention, a traditional strength-based segmentation algorithm and the knowledge of anatomy are combined together so as to segment a specific organ in an abdominal CT data set. The candidate regions of each organ are obtained based on the prior anatomical knowledge of a current data set; and the data of the organ are extracted accurately from the candidate regions through using a density-based method. The algorithm of the present invention can be applied to images of patients of different sizes through using the relative positions of organs in human bodies. Segmentation is implemented in a gradual way; and the candidate regions are roughly defined, a target region is refined by using the density-based segmentation method, and therefore, the method of the invention makes segmentation accuracy more satisfactory. The algorithm of the invention can be applied to segmenting a single organ in a medical image and can be applied to anatomical variability among different patients, thereby helping to reduce segmentation errors and ultimately improving subsequent compression operation.

Description

technical field [0001] The invention relates to the field of computer application technology, especially the field of digital image processing. Background technique [0002] Medical imaging technology plays an indispensable role in modern clinical medicine and is an important basis for doctors to diagnose diseases. However, with the improvement of the resolution of medical imaging equipment, the huge amount of data generated by the acquired images has brought enormous pressure to image storage and real-time transmission. Therefore, it is very necessary to find an effective compression algorithm. The difficulty of this technology lies in compressing the image while maintaining a good image quality. [0003] Most medical images are three-dimensional image sequences, and there is not only intra-slice correlation, but also strong correlation between slices. The more slices, the stronger the correlation. This characteristic of medical images determines that the compression of m...

AI Technical Summary

Problems solved by technology

[0010] Medical images have different characteristics from general images, such as weak edges due to the partial volume effect (PVE) phenomenon, and pixels on the boundary have the average value of all surrounding pixels, resulting in "blurring" of the boundary area, so general prediction methods do not It is suitable for direct application to the compression of medical images. Moreover, considering the properties of medical image data, namely, bilateral anatomical symmetry and structural anatomical similarity across different patients, it is very useful to propose a compression algorithm tailored for medical images. necessary

Images