Purpose A two‐layer cylinder (TLC) phantom was developed for simplifying film‐based isocenter verification of linear accelerators in radiotherapy. Methods and materials The phantom mainly consists of two parts: (1) two nested solid cylinders between which a radiochromic film can be inserted and irradiated; (2) a tungsten ball supported by a thin rod and located at the phantom center for alignment with the mechanical isocenter. In practice, the phantom was first positioned by the room laser to align the tungsten ball to the mechanical isocenter of the linear accelerator. Then, a radiochromic film was precisely inserted into the gap between the two cylinders of the phantom and irradiated by beams with preset gantry and couch angles. Later the irradiated film was scanned and processed by an in‐house developed analysis software. Finally, the offset of the radiation isocenter from the mechanical isocenter was determined by the built‐in three‐dimensional (3D) reconstruction algorithms. The accuracy of this method was evaluated by positioning the phantom with a known couch shift, then checking the residual error after couch shift correction. The reliability of this method was evaluated by comparing the calculated offset with the corresponding result determined by the traditional star‐shot method. Results For the accuracy test, the residual errors were −0.14 ± 0.03 mm, 0.05 ± 0.06 mm, and 0.05 ± 0.06 mm in the lateral, longitudinal, and vertical axes, respectively. For the reliability test, the differences between the calculated offset and the result determined by the star‐shot method were −0.10 mm, 0.12 mm, and 0.12 mm in the lateral, longitudinal, and vertical axes, respectively. Conclusion The proposed method is able to reconstruct beams in 3D with one film, which is more time‐saving and accurate. Additionally, with this design, the phantom positioning, film loading, beam delivery, and data analyzing are simpler. This phantom and analysis software provides an efficient and effective way to perform film‐based isocenter verification of linear accelerators in radiotherapy.

No abstract

Background: To develop an unsupervised anomaly detection method to identify suspicious error-prone treatment plans in radiotherapy. Methods: A total of 577 treatment plans of breast cancer patients were used in this study. They were labeled as either normal or abnormal plans by experienced clinicians. Multiple features of each plan were extracted and selected by the learning algorithms. The training set consisted of feature samples from 400 normal plans and the testing set consisted of feature samples from 158 normal plans and 19 abnormal plans. Using the k-means clustering algorithm in the training stage, 4 normal plan clusters were formed. The distance between the samples in the testing set and the cluster centers were then determined. To evaluate the effect of dimensionality reduction (DR) on detection accuracy, principal component analysis (PCA) and autoencoder (AE) methods were compared. Results: The sensitivity of the anomaly detection model based on PCA and AE methods were 84.2% (16/19) and 94.7% (18/19), respectively. The specificity of the anomaly detection model based on PCA and AE methods were 64.6% (102/158) and 69.0% (109/158), respectively. The areas under the receiver operating characteristic (ROC) curve (AUCs) based on PCA and AE methods were 0.81 and 0.90, respectively. Conclusions: The unsupervised learning method was effective for detecting anomalies from the feature samples. Accuracy could be improved with the introduction of AE-based DR technique. The combination of AE and k-means clustering methods provides an automated way to identify abnormal plans among clinical treatment plans in radiotherapy.