Rigid registration of multimodal medical images based on mask optimization(PDF)
《中国医学物理学杂志》[ISSN:1005-202X/CN:44-1351/R]
- Issue:
- 2018年第9期
- Page:
- 1022-1029
- Research Field:
- 医学影像物理
- Publishing date:
Info
- Title:
- Rigid registration of multimodal medical images based on mask optimization
- Author(s):
- PI Yifei1; WU Qian2; PEI Xi1; XU Xie1
- 1. School of Physical Sciences, University of Science and Technology of China, Hefei 230026, China; 2. School of Medical
Humanities, Anhui Medical University, Hefei 230032, China
- Keywords:
- medical image processing; rigid registration; mask; multimodality
- PACS:
- R319；TP311.52
- DOI:
- 10.3969/j.issn.1005-202X.2018.09.006
- Abstract:
- Abstract：Objective To develop a rigid registration program based on ITK and OpenCV and make it compatible with DeepPlan
treatment planning system for achieving rapid and precise multimodality image registration. Methods Mathematical morphology
analysis was used to erasure insignificant and tiny structures and partial scanning couch; Otsu filter was applied to further highlight
the regions of interest; and Canny operators were utilized to extract edge information from regions rich in information. Then the
mask required in image registration was obtained by pixel filling algorithm, and the pre-processing process was accelerated with
OpenMP parallel strategy. Finally, the mask acted on the fixed image and moving image during registration. Results The proposed
method was validated on clinical images datasets with different modalities and body parts, and the results revealed that the most
interferential background image and scanning couch were removed by rigid registration of multimodal medical image based on
mask optimization, saving more than more than half of the time while without degrading the image registration quality. The
registration of two groups of about 100 images could be finished within 1 minute. The relative codes were also successfully
integrated in DeepPlan treatment planning system in the form of dynamic-link library. Conclusion Rigid registration of multimodal
medical image based on mask optimization improves the efficiency of image registration, without affecting registration accuracy,
and the proposed algorithm is highly stable. Therefore, the proposed method has a good prospect in clinical application.
Last Update: 2018-09-28