AUTHOR=Huang Peng , Shang Jiawen , Fan Yuhan , Hu Zhihui , Dai Jianrong , Liu Zhiqiang , Yan Hui 

TITLE=Unsupervised machine learning model for detecting anomalous volumetric modulated arc therapy plans for lung cancer patients

JOURNAL=Frontiers in Big Data

VOLUME=Volume 7 - 2024

YEAR=2024

URL=https://www.frontiersin.org/journals/big-data/articles/10.3389/fdata.2024.1462745

DOI=10.3389/fdata.2024.1462745

ISSN=2624-909X

ABSTRACT=Purpose: Volumetric modulated arc therapy (VMAT) is a new treatment modality in modern radiotherapy. To assure quality of the radiotherapy plan, the physics plan review is routinely performed by the senior clinicians, which is less efficient and accurate. In this study, a multi-task AutoEncoder (AE) is proposed to automate anomaly detection of VMAT plans for lung cancer patients.The feature maps are first extracted from a VMAT plan. Then, a multi-task AE is trained with the input of feature map and output of two features (beam aperture and prescribed dose). Based on the distribution of reconstruction errors on the training set, a detection threshold value is obtained. For a testing sample, its reconstruction error is calculated by the AE model and compared with the threshold value to determine its classes (anomaly or regular). The proposed multi-task AE model is compared with the other existing AE models including Vanilla AE, Contractive AE and Variational AE.The area under the receiver operating characteristic curve (AUC) and the other statistics are used to evaluate the performance of these models.Results: Among the four tested AE models, the proposed multi-task AE model achieves the highest values in AUC (0.964), accuracy (0.821), precision (0.471) and F1 score (0.632), and the lowest value in FPR (0.206).The proposed multi-task AE model using 2D feature maps can effectively detect anomaly in radiotherapy plans for lung cancer patients. Comparing with the other existing AE models, the multi-task AE is more accurate and efficient. The proposed model provides a feasible way to carry out automate anomaly detection of VMAT plans in radiotherapy.