Comparative clinical evaluation of atlas and deep-learning-based auto-segmentation of organ structures in liver cancer

• Published in: Radiation oncology (London, England) Vol. 14; no. 1; pp. 213 - 13
• Main Authors: Ahn, Sang Hee, Yeo, Adam Unjin, Kim, Kwang Hyeon, Kim, Chankyu, Goh, Youngmoon, Cho, Shinhaeng, Lee, Se Byeong, Lim, Young Kyung, Kim, Haksoo, Shin, Dongho, Kim, Taeyoon, Kim, Tae Hyun, Youn, Sang Hee, Oh, Eun Sang, Jeong, Jong Hwi
• Format: Journal Article
• Language: English
• Published: London BioMed Central 27.11.2019; BioMed Central Ltd; Springer Nature B.V; BMC
• Subjects: Accuracy; Aged; Artificial neural networks; Atlas-based auto-segmentation; Biomedical and Life Sciences; Biomedicine; Cable television broadcasting industry; Cancer; Cancer patients; Cancer Research; Cancer therapies; Care and treatment; Clinical Radiation Oncology; Comparative analysis; Computed tomography; Contouring; Convolution; Datasets; Deep convolution neural network (DCNN); Deep Learning; Deep-learning-based auto-segmentation; Female; Heart; Humans; Image contrast; Image processing; Image Processing, Computer-Assisted - methods; Image segmentation; Imaging; Kidneys; Libraries; Liver; Liver cancer; Liver Neoplasms - diagnostic imaging; Liver Neoplasms - radiotherapy; Machine learning; Male; Medical imaging; Medical imaging equipment; Methods; Metric space; Middle Aged; Neural networks; Neural Networks, Computer; Observer Variation; Oncology; Organs; Organs at Risk; Patients; Physicians; Radiation; Radiation (Physics); Radiation therapy; Radiology; Radiotherapy; Radiotherapy Planning, Computer-Assisted - methods; Reproducibility of Results; Republic of Korea; Software; Stomach; Tomography; Tomography, X-Ray Computed; Republic of Korea; United States; United Kingdom > UK; United States > US; Contouring; Deep convolution neural network (DCNN); Atlas-based auto-segmentation; Deep-learning-based auto-segmentation
• ISSN: 1748-717X; 1748-717X
• DOI: 10.1186/s13014-019-1392-z

Background Accurate and standardized descriptions of organs at risk (OARs) are essential in radiation therapy for treatment planning and evaluation. Traditionally, physicians have contoured patient images manually, which, is time-consuming and subject to inter-observer variability. This study aims to a) investigate whether customized, deep-learning-based auto-segmentation could overcome the limitations of manual contouring and b) compare its performance against a typical, atlas-based auto-segmentation method organ structures in liver cancer. Methods On - contrast computer tomography image sets of 70 liver cancer patients were used, and four OARs (heart, liver, kidney, and stomach) were manually delineated by three experienced physicians as reference structures. Atlas and deep learning auto-segmentations were respectively performed with MIM Maestro 6.5 (MIM Software Inc., Cleveland, OH) and, with a deep convolution neural network (DCNN). The Hausdorff distance (HD) and, dice similarity coefficient (DSC), volume overlap error (VOE), and relative volume difference (RVD) were used to quantitatively evaluate the four different methods in the case of the reference set of the four OAR structures. Results The atlas-based method yielded the following average DSC and standard deviation values (SD) for the heart, liver, right kidney, left kidney, and stomach: 0.92 ± 0.04 (DSC ± SD), 0.93 ± 0.02, 0.86 ± 0.07, 0.85 ± 0.11, and 0.60 ± 0.13 respectively. The deep-learning-based method yielded corresponding values for the OARs of 0.94 ± 0.01, 0.93 ± 0.01, 0.88 ± 0.03, 0.86 ± 0.03, and 0.73 ± 0.09. The segmentation results show that the deep learning framework is superior to the atlas-based framwork except in the case of the liver. Specifically, in the case of the stomach, the DSC, VOE, and RVD showed a maximum difference of 21.67, 25.11, 28.80% respectively. Conclusions In this study, we demonstrated that a deep learning framework could be used more effectively and efficiently compared to atlas-based auto-segmentation for most OARs in human liver cancer. Extended use of the deep-learning-based framework is anticipated for auto-segmentations of other body sites.