{"title":"[Evaluation of Auto-segmentation in the Head and Neck Region].","authors":"Issei Matsuo, Masayuki Tachibana, Hirofumi Honda, Hiroki Namba","doi":"10.6009/jjrt.25-1475","DOIUrl":null,"url":null,"abstract":"<p><strong>Purpose: </strong>The purpose of this study was to perform auto-segmentation on cases in the head and neck region and to elucidate the characteristics of the extraction performance.</p><p><strong>Methods: </strong>Auto-segmentation was performed on 100 cases in the head and neck region in Japan. The Auto-segmentation contours were compared with the clinical contours. The evaluation structures were the brain, brainstem, spinal cord, mandible, parotid gland, larynx, and eyeballs. DSC and HD were used for similarity evaluation.</p><p><strong>Results: </strong>Overall, there was a high similarity in the eyeball and spinal cord. However, the lower the contrast with neighboring objects and the greater the variation in shape, the lower the similarity. In addition, metal artifacts lowered the similarity in some areas.</p><p><strong>Conclusion: </strong>When auto-segmentation extraction was performed, differences were observed in the degree of similarity between organs. Depending on the case and morphology, the accuracy of auto-segmentation extraction varied from site to site, even for the same organ. The characteristics of the extractability found in this study make it useful for the manual modification of contouring.</p>","PeriodicalId":74309,"journal":{"name":"Nihon Hoshasen Gijutsu Gakkai zasshi","volume":"81 3","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Nihon Hoshasen Gijutsu Gakkai zasshi","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.6009/jjrt.25-1475","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
Purpose: The purpose of this study was to perform auto-segmentation on cases in the head and neck region and to elucidate the characteristics of the extraction performance.
Methods: Auto-segmentation was performed on 100 cases in the head and neck region in Japan. The Auto-segmentation contours were compared with the clinical contours. The evaluation structures were the brain, brainstem, spinal cord, mandible, parotid gland, larynx, and eyeballs. DSC and HD were used for similarity evaluation.
Results: Overall, there was a high similarity in the eyeball and spinal cord. However, the lower the contrast with neighboring objects and the greater the variation in shape, the lower the similarity. In addition, metal artifacts lowered the similarity in some areas.
Conclusion: When auto-segmentation extraction was performed, differences were observed in the degree of similarity between organs. Depending on the case and morphology, the accuracy of auto-segmentation extraction varied from site to site, even for the same organ. The characteristics of the extractability found in this study make it useful for the manual modification of contouring.