Efficacy evaluation of 2D, 3D U-Net semantic segmentation and atlas-based segmentation of normal lungs excluding the trachea and main bronchi

• Published in: Journal of radiation research Vol. 61; no. 2; pp. 257 - 264
• Main Authors: Nemoto, Takafumi, Futakami, Natsumi, Yagi, Masamichi, Kumabe, Atsuhiro, Takeda, Atsuya, Kunieda, Etsuo, Shigematsu, Naoyuki
• Format: Journal Article
• Language: English
• Published: England Oxford University Press 23.03.2020
• Subjects: Algorithms; Bronchi; Bronchi - diagnostic imaging; Cancer research; Care and treatment; CAT scans; Computed tomography; Contouring; Deep learning; Digital imaging; Durvalumab; Effectiveness; Humans; Image Processing, Computer-Assisted; Image segmentation; Imaging, Three-Dimensional; Lung - diagnostic imaging; Lung cancer; Machine learning; Medical imaging; Medical imaging equipment; Methods; Non-small cell lung cancer; Radiation; Radiation (Physics); Radiation therapy; Radiotherapy; Rank tests; Regular Paper; Semantic segmentation; Semantics; Test sets; Tomography; Tomography, X-Ray Computed; Trachea; Trachea - diagnostic imaging; Japan; lung cancer; semantic segmentation; trachea; main bronchi; U-Net

This study aimed to examine the efficacy of semantic segmentation implemented by deep learning and to confirm whether this method is more effective than a commercially dominant auto-segmentation tool with regards to delineating normal lung excluding the trachea and main bronchi. A total of 232 non-s...