Investigation the motion data clustering of lung tumor on its position estimation at external surrogates’ radiotherapy
Among thorax tumors, lung tumors move mainly due to respiration. In order to enhance the precision of radiotherapy, one solution is estimating tumor motion from external motion of chest wall and abdomen regions. For this aim, consistent prediction models are constructed and then implemented for real...
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | fas |
| Published: |
Semnan University
2022-03-01
|
| Series: | مجله مدل سازی در مهندسی |
| Subjects: | |
| Online Access: | https://modelling.semnan.ac.ir/article_5728_b2d224b2da2e71d1563cda61310c09eb.pdf |
| Summary: | Among thorax tumors, lung tumors move mainly due to respiration. In order to enhance the precision of radiotherapy, one solution is estimating tumor motion from external motion of chest wall and abdomen regions. For this aim, consistent prediction models are constructed and then implemented for real time tumor motion tracking. In these models, clustering of database extracted from tumor motion and chest wall motion has non-negligible effect which has been taken into account in this work. In this investigation, motion database of fifteen patients with lung cancer who were treated by means of Cyberknife Synchrony system at Georgetown University hospital, has been used. Two subtractive and fuzzy C-means as common available clustering strategies have been employed in order to investigate their quantitative effects, in a comparative fashion. Final analyzed results show that the average targeting error of prediction models (difference between tumor position estimated by model and actual position of tumor) over all patients are 6.5 and 7.5 mm implementing subtractive and fuzzy C-means clustering, respectively. Moreover, using fuzzy C-means algorithm, tumor tracking is done with more stability. Since, breathing phenomena has high degree of variations, motion data clustering has an important role on the accuracy of prediction model performance by determining model parameters while constructing at pre-treatment step and while updating the model during the treatment. |
|---|---|
| ISSN: | 2008-4854 2783-2538 |