AUTHOR=Liu Zhikai , Liu Fangjie , Chen Wanqi , Liu Xia , Hou Xiaorong , Shen Jing , Guan Hui , Zhen Hongnan , Wang Shaobin , Chen Qi , Chen Yu , Zhang Fuquan 

TITLE=Automatic Segmentation of Clinical Target Volumes for Post-Modified Radical Mastectomy Radiotherapy Using Convolutional Neural Networks

JOURNAL=Frontiers in Oncology

VOLUME=Volume 10 - 2020

YEAR=2021

URL=https://www.frontiersin.org/journals/oncology/articles/10.3389/fonc.2020.581347

DOI=10.3389/fonc.2020.581347

ISSN=2234-943X

ABSTRACT=Background: This study aimed to construct and validate a model based on a convolutional neural network (CNN), which can fulfill the automatic segmentation of the clinical target volume (CTV) of breast cancer for radiotherapy. 
Methods: In this work, computed tomography (CT) scans of 110 patients who underwent modified radical mastectomies were collected. The CTV contours were confirmed by two experienced oncologists. A novel CNN was constructed to delineate the CTV automatically. Quantitative evaluation metrics were calculated, and a clinical evaluation was conducted to evaluate the performance of our model.
Results: The mean Dice similarity coefficient (DSC) of the proposed model was 0.90, and the 95th percentile Hausdorff distance (95HD) was 5.65 mm. The evaluation results of two clinicians showed that 99.3% of the chest wall CTV slices could be accepted by clinician A, and this number was 98.9% for clinician B. In addition, 9/10 of patients that have all slices acceptable by clinician A, while 7/10 can be accepted by clinician B. The score differences between the AI (artificial intelligence) group and the GT (ground truth) group show no statistically significant differences for both clinicians. However, the score differences in the AI group are significantly different between the two clinicians. The Kappa consistency index is 0.259. It took 3.45 s to delineate chest wall CTV using the model.
Conclusion: Our model could automatically generate the CTV of breast cancer. AI-generated structures by the proposed model have a trend to be comparable well or even better than those of human-generated structures. Further multi-center evaluations should be performed for adequate validation before it could be completely applied in clinical practice.