Three-dimensional reconstruction of brain CT image

Abstract

The invention discloses a three-dimensional reconstruction method of a brain tumor CT image. The method comprises the steps of segmentation of a two-dimensional CT image of a brain and three-dimensional reconstruction. The segmentation method of the two-dimensional CT image provided by the invention eliminates noise interference, ensures effective information, removes over-segmented parts, speedsup segmentation, and improves segmentation accuracy. Meanwhile, a three-dimensional reconstruction algorithm greatly improves the execution efficiency of the algorithm, and has certain value for braintumor diagnosis, research and treatment in clinical practice.

Description

technical field [0001] The invention relates to an image segmentation processing method and a three-dimensional reconstruction method of brain CT images, which can be used for multiple image segmentation including micro-nano particle distribution, cell or defect detection. Background technique [0002] In recent years, brain tumor image segmentation technology has become more and more important in the clinical treatment of brain tumors. At present, the more popular segmentation methods are based on fuzzy clustering method, based on region growing method, based on deformable model method, based on morphology watershed method, Based on anomaly detection method and so on. Among them, watershed and region growth segmentation are accurate, fast and efficient, so they are most commonly used in practical operations. However, because the watershed is too sensitive to noise, although the improved watershed has obvious anti-noise effect, it still produces over-segmentation. [0003]...

AI Technical Summary

Problems solved by technology

However, because the watershed is too sensitive to noise, the improved watershed, although the anti-noise effect is obvious, will still produce over-segmentation, etc.

[0003] Due to the diversity of tissues in current medical images, it is difficult to segment accurately by a single segmentation method, but the comprehensive segmentation method combined with watershed and region growing has achieved a good segmentation effect, and the over-segmentation phenomenon has been improved to a certain extent, but it is still not achieved. Desirable effect, and sensitivity to noise is also a pressing issue

In order to solve these problems; for low-density, iso-density or mixed-density lesions, the gray scale change between the lesion area and the normal tissue in the CT image is not obvious, but the noise is obvious, which is not conducive to segmentation

The advantage of watershed is that it can generate closed and continuous edges, so watershed can be used to extract edges, but watershed is too sensitive to noise and prone to over-segmentation problems. For this, we can combine improved region growth and morphological reconstruction

Images