Feasibility study of evaluating autosegmentation accuracy with geometric indices based on dosimetric parameters of esophageal cancer radiotherapy(PDF)
《中国医学物理学杂志》[ISSN:1005-202X/CN:44-1351/R]
- Issue:
- 2021年第2期
- Page:
- 148-152
- Research Field:
- 医学放射物理
- Publishing date:
Info
- Title:
- Feasibility study of evaluating autosegmentation accuracy with geometric indices based on dosimetric parameters of esophageal cancer radiotherapy
- Author(s):
- XIAN Lixun1; 2; LI Guangjun1; XIAO Qing1; CHEN Li1; HU Zhenyao1; BAI Sen1; 2
- 1.Department of Radiotherapy, West China Hospital of Sichuan University, Chengdu 610041, China 2. School of Physical Science and Technology, Sichuan University, Chengdu 610065, China
- Keywords:
- Keywords: esophageal cancer Atlas-based autosegmentation organs-at-risk geometric index dosimetric parameter
- PACS:
- R318；R811.1
- DOI:
- DOI:10.3969/j.issn.1005-202X.2021.02.004
- Abstract:
- Abstract: Objective To explore the differences among geometric indices HD, HDmean, HD95%, DSC and Jaccard, and to study the clinical feasibility of evaluating autosegmentation accuracy with geometric indices based on the dosimetric parameters in esophageal cancer radiotherapy. Methods Fifteen patients with esophageal cancer who received radiotherapy in Department of Radiotherapy, West China Hospital of Sichuan University were selected as the research objects, and their lungs, spinal cord and heart were outlined using Atlas-based autosegmentation software. The contours modified and reviewed by senior doctors were regarded as the references. The dosimetric parameters of the corresponding automatically-segmented structures were obtained on the original dose distribution map generated by the reference contours, and the linear regression R?was used to quantify the correlation between geometric indices and dosimetric parameters. Results The fluctuation range of distance-related geometric indices (0-44.7 mm) was larger than that of volume-related indices (0.7-1.0). The correlations between geometric indices and dosimetric parameters were different (45% correlated, P<0.05) and inconsistent. Conclusion The dosimetric parameter is deeply affected by the locations and differences of contours, which highlights its importance in clinical evaluation and shows that it is not reliable to use geometric indices only for clinical evaluation. The combination of geometric indices and dosimetric parameters is recommended for clinical evaluation.
Last Update: 2021-02-02