Image registration method So far, in the field of image processing research at home and abroad, quite a lot of image registration research work has been reported, and many image registration methods have been produced. In general, various methods are oriented to a certain range of application fields and have their own characteristics. Around the methods and methods of image registration, a variety of image registration algorithms and applications have been established, such as gray-scale information method, transform domain method, feature-based method, etc., each of which derives a variety of specific methods , Conduct multi-directional research on registration technology and matching accuracy [1-3].With the development of medical imaging technology, various imaging equipment generates a large number of different formats of medical imaging images, which have reached a fine level of expression in time and space resolution. Using voxel points to express images is one of the fine-grained expression methods. The pixel is the basic unit of image representation, and the identification and calculation of each pixel has an important role in the final result and application [3-5].

Object of the design

The precise registration of medical images is achieved by designing a matching algorithm on the granularity of voxel point expression. The granularity of medical image expression is getting finer, and the resolution of time and space is getting higher and higher, allowing in-depth exploration of precision medicine in medical research. Image registration is the basis of medical image processing in medical research and clinical applications. Therefore, designing a matching algorithm through voxel point expression to achieve image registration is also a way to solve medical image registration, which is the subsequent medical image processing[2-4]. Especially the subsequent operations in the field of functional and structural images lay the foundation.

Method of the design

The problem of image registration is transformed into the matching of text or numerical data in voxel points. The problem of image registration is how to compare images. Whether it is feature or correlation, it is much more complicated to measure and implement. The matching of text or value is more accurate and simple than the image itself [6-8]. Therefore, through the conversion of stored data, Realize the fast matching of images.

Using the combination of Matlab's advantages in image and Java language in data processing, the rapid comparison of voxel points is automatically realized through programming. Using the advantages of the software program itself, under the premise of ensuring the algorithms and functions, the comparison between the basic data is automatically realized. The comparison of the basic data has an absolute advantage in time compared with the image registration in image processing[3-6].

Process of the design

First, Image text or numerical conversion: Through the professional software or general editing software that obtains the image, the obtained image data is stored and converted to obtain the text or numerical image data stored in voxel points[8].

Second, matching algorithm design: Set parameters and matching algorithms according to the value of the voxel points after image conversion, as well as the definition, meaning and requirements of the image itself[9-11].

Third, Matlab and Java program design of voxel point speed matching image: design the registration program according to the exact matching algorithm, realize the matching function of text or numerical data through the language platform, and establish a software program or platform for voxel point speed matching image to realize voxel Point identification and accurate calculation of volume[12].

Forth, verification of accurate registration of images: Perform accurate registration of voxel points on the converted text or numerical image data, and verify the accurate registration of voxel points through other software, platforms or research results, thereby verifying the registration Algorithm and program software or platform[11].

In image conversion, due to different image acquisition methods and equipment, different storage formats, different representations, and different application environments, conversion to basic data format storage sometimes requires different software platforms for conversion or settings[12].

In matching algorithm, since the voxel points are different in different images, different matching algorithms need to be used to achieve registration platforms with different granularity. The converted image value needs to correspond to the meaning of the image itself, how to proceed from the image parameters Algorithm design needs to be designed in combination with the environment and conditions of the image[13,14].

For programming in different languages, how to deal with the same problem on two platforms, while also considering the time and space complexity of the program, to realize the software platform for voxel point speed matching image application[15].

In this study, the images expressed by voxel points are designed for accurate registration research. First, convert the acquired image into text or numerical data with voxel points as the unit, and then establish an accurate registration algorithm based on voxel point matching. Finally, the area that needs to be registered or calculated volume is based on the voxel point as the unit. The quasi-algorithm performs image registration and calculation, and realizes the identification and volume calculation of the image area. The calculation and realization of the project are based on the Matlab platform and the Java language platform as the experimental environment.

Medical image data generally occupies a large space, the processing process is complicated, and the data processing time is long. This project takes voxel points as the research object, converts the image data, and combines Java language program processing to achieve rapid image registration and accurate calculation of the activation volume, which provides strong support for medical image registration processing and research. At the same time, the time efficiency of image processing is improved through the Java program.

Acknowledgements

This research was supported by the National Students’ project for innovation and entrepreneurship training program under Grant Number S201910439039.The authors are grateful to the anonymous referees for their valuable comments and suggestions.

The authors declare that they have no conflict of interest

No funding sources

The study was approved by the Shandong First Medical University and Shandong Academy of Medical Sciences, Taian, China, 271016.

1. Hill, Derek L. G., et al. "Medical Image Registration." Physics in Medicine & Biology, 31.4 (2008), pp. 1-45.

2. Aladl, U. E., and T. Peters. "Medical Image Registration." Multi Modality State-of-the-Art Medical Image Segmentation and Registration Methodologies, Phys Med Biol, 2011.

3. Jenkinson, M., and S. Smith. "Medical Image Registration." Physics in Medicine & Biology, 46.3 (2001), pp. R1.

4. Peng, G., et al. "Medical Image Registration via Modified Differential Search Algorithm." Journal of Frontiers of Computer Science and Technology, 22.1 (2019), pp. 1-5.

5. Bashiri, Fereshteh S., et al. "Multi-Modal Medical Image Registration with Full or Partial Data: A Manifold Learning Approach." Journal of Imaging, 22.1 (2019), pp. 1-5.

6. Alam, F., et al. "An Automatic Medical Image Registration Approach Based on Common Sub-regions of Interest." Journal of Medical Imaging and Health Informatics, 22.1 (2019), pp. 1-5.

7. Boveiri, H. R., et al. "Medical Image Registration Using Deep Neural Networks: A Comprehensive Review." Computers & Electrical Engineering, 87 (2020), p. 106767.

8. Hu, J., et al. "Towards Accurate and Robust Multi-Modal Medical Image Registration Using Contrastive Metric Learning." IEEE Access, 7 (2019), pp. 132816-132827.

9. Gui, P., et al. "Cross-cumulative Residual Entropy-Based Medical Image Registration via Hybrid Differential Search Algorithm." International Journal of Imaging Systems and Technology, 29.6 (2019), pp. 1-5.

10. Zhao, S., et al. "Recursive Cascaded Networks for Unsupervised Medical Image Registration." 2019 IEEE/CVF International Conference on Computer Vision (ICCV), IEEE, 2020.

11. Wang, J., and M. Zhang. "DeepFLASH: An Efficient Network for Learning-Based Medical Image Registration." 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 2020.

12. Zhu, Y., et al. "New Loss Functions for Medical Image Registration Based on VoxelMorph." Image Processing, 22.1 (2020), pp. 1-5.

13. Haskins, G., et al. "Deep Learning in Medical Image Registration: A Survey." Machine Vision and Applications, 31.1-2 (2020), pp. 1-5.

14. Zhu, W., et al. "NeurReg: Neural Registration and Its Application to Image Segmentation." 2020 IEEE Winter Conference on Applications of Computer Vision (WACV), IEEE, 2020.

15. Fu, T., et al. "Groupwise Registration with Global-Local Graph Shrinkage in Atlas Construction." Medical Image Analysis, 64 (2020), p. 101711.