Multimodal medical image registration method

Abstract

The invention provides a multimodal medical image registration method. On the basis of virtual simulation, the method includes the steps: S11, respectively acquiring an MRI (magnetic resonance imaging) image and a real ultrasound image; S12, performing manual threshold segmentation on the MRI image to obtain corresponding histograms of tissue and organs; S13, determining an impedance value of each of the tissue and organs according to the histograms; S14, utilizing the impedance values for performing simulation computation on the ultrasound image to obtain a virtual simulated ultrasound image; S15, performing similarity measurement on the virtual simulated ultrasound image and the real ultrasound image to obtain a measured value; S16, optimizing parameters in the step S14 until the measured value reaches a set threshold range, and finishing image registration accordingly. The multimodal medical image registration method can convert multimodal image registration into image registration in the same mode, and quick, accurate and robust image registration is realized.

Description

technical field [0001] The invention relates to medical image technology, in particular to a multi-modal medical image registration method based on virtual simulation technology. Background technique [0002] Multimodal image fusion guidance technology assists doctors in surgical operations by registering and fusing images acquired by different imaging devices. Ultrasound imaging has the advantages of wide adaptability, low price, no radiation, fast and non-invasive, and has been widely used in the medical field; Improve the safety and success rate of surgery. MRI can provide high-contrast and high-resolution images of soft tissues with less radiation, and has gradually become the mainstream imaging device for preoperative image acquisition. MRI images are collected before the operation and three-dimensional reconstruction is performed, and ultrasound images are collected in real time during the operation. Through accurate and real-time registration algorithms, dual-mode n...

AI Technical Summary

Problems solved by technology

However, the existing multimodal fusion algorithm has not explored the physical imaging mechanism of different modalities, so multimodal image registration has always been a technical problem. In addition, the structural information of the ultrasound image itself is not obvious, and its accuracy is difficult. It is guaranteed that how to automatically and quickly realize the registration and fusion of MRI images and ultrasound images, there is still no good implementation method, and there are many challenging problems

Due to the different physical characteristics of images of two different imaging modalities and the quality problems of ultrasound images, most of the registration methods are manual, and it is difficult to achieve an accurate, fast and intelligent registration method

In the current MRI and ultrasound registration methods, the registration is basically performed by means of external landmarks. It is difficult to achieve non-rigid body transformation for the non-invasive landmark method, and the invasive implantation of external landmarks, and It directly brings pain to patients, especially for frail patients such as children and the elderly, it is generally impossible to achieve the implantation of landmarks

[0004] At present, the medical image registration methods mainly include: a registration based on image feature information, by extracting the common feature information on the image to be registered, such as: point, line, surface and other information, and then establishing a target optimization function, by optimizing The algorithm realizes the solution of the image transformation matrix, but the accuracy of this type of method largely depends on the extraction accuracy of the feature points, and the anti-noise performance is poor, and for medical images, there is very little effective feature information, such as ultrasound images, images themselves The signal-to-noise ratio is not high, and the feature information is less

The other is registration based on image grayscale information. On the one hand, it has a large amount of calculation, and more importantly, on the other hand, due to the incomparability between image data of different modalities and the inconsistent pixel size, it is difficult to achieve registration based on grayscale information. Degree information can effectively achieve the purpose of image registration

Images