Authors

YIHUA LAN*,**, JIANFANG WANG* , XING WANG***, XIAO SONG*,**, YANG WANG****, NING CHENG*,**, XIAOLI WANG*****

Departments

* School of Computer and Information Technology, Nanyang Normal University, Nanyang 473061, China - **Center for Biomedical Imaging and Bioinformatics, School of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan, Hubei 430074, China - ***Network center, Nanyang Normal University, Nanyang 473061, China - ****Radiology department,Central Hospital of Nanyang, Nanyang 473061, China - *****School of Mathematics and Statistics, Nanyang Normal University, Nanyang 473061, China

Abstract

Introduction: For the step and shoot model, which is a traditional realization way in Intensity Modulated Radiation Therapy, the inverse planning optimization system needs a set of fluence map for each beam with low complexity to be modulated by the multi-leaf collimator. A lot of image denoising methods have been introduced to smooth the fluence map. Recently, some convex optimization models have been proposed for this work. These methods only consider depressing the map complexity, but neglect the latter leaf sequencing process. Therefore, they can not bring the map complexity to the best state for dose modulation.

Material and methods: The monitor unit based map convex smoothing model has depressed the total number of monitor unit of the fluence map to the lowest state, but they do not consider the uniformity of each beam dose modulation. Large or small monitor units will lead to the inaccuracy of ray modulation. Furthermore, large difference of the monitor units for each beam fields will lead to the beam angle or beam number adjustment. Based on the above consideration, we proposed a novel regularization smoothing model for fluence map optimization of inverse planning in Intensity Modulated Radiation Therapy. We not only consider depressing the total number of monitor units, but also balancing all of them in the model.

Results: We test this model in a prostate case, and compare with two kind of smoothing methods as well as the monitor unit based map convex smoothing model. The results show our regularization smoothing method will lead to a lower complexity and much more balanced fluence map for the latter leaf sequencing process.

Discussion: The obvious advantage of the proposed model is that the fluence map complexity is low, as well as the monitor unit distribution of each beam is more evenly.

Keywords

fluence map, inverse planning, total number of monitor units, regularization smoothing.