Deep Neural Network and Transfer Learning for DVH prediction in VMAT prostate treatments
OC-0222
Abstract
Deep Neural Network and Transfer Learning for DVH prediction in VMAT prostate treatments
Authors: Ambroa Rey|, Eva Maria(1)*[eva.ambroa@gmail.com];Pérez-Alija|, Jaime(2);Gallego|, Pedro(2);
(1)Consorci Sanitari de Terrassa, Medical Physics Unit- Radiation Oncology, Terrassa, Spain;(2)Hospital de la Santa Creu i Sant Pau, Medical Physics, Barcelona, Spain;
Show Affiliations
Hide Affiliations
Purpose or Objective
Volumetric modulated arc therapy (VMAT) has been used widely to provide highly conformal plans. However, treatment planning has increased in complexity and has become a time-consuming process.
The purpose of this work is to establish a convolutional neural network (CNN) model for the prediction of rectum and bladder dose-volume histograms (DVH) in prostate patients treated with a VMAT technique.
Material and Methods
A total of 145 patients with intermediate or high-risk prostate cancer treated with simultaneous integrated boost (SIB) were selected for this study. The prescribed dose was 50.4 Gy and 70Gy in 28 fractions to the pelvic and prostate area, respectively. Data were split into two sets: 120 and 25 patients, respectively. Besides, the first set was partitioned in training, validation, and test, each with 100, 10, and 10 patients. The second set was used for final validation.
We use transfer learning in combination with a VGG-16 network. VGG-16 is a deep CNN pre-trained with the ImageNet image dataset (1.2 million natural images of 1000 object categories). We dropped the fully-connected layers from the VGG-16 and added a new fully-connected neural network.
The inputs for the CNN were a 2D image of the volumes contoured in the CT. We only retained the geometrical information of every CT. The outputs were the corresponding dose maps. Rectum and bladder DVHs were computed for each patient summing up all the dose-volume information in every slice. To ensure the quality of the data, we selected all potential outliers and proceeded to re-optimize them.
The already trained CNN was tested using the second test of patients (25). All patients were re-optimized by the same operator unfamiliar with the results of the prediction. A confusion matrix was used to report the number of false positives, false negatives, true positives, and true negatives.
Results
Figure I shows the clinically approved, replanned, and predicted bladder and rectum DVHs for three representative cases. The results demonstrated that for most cases the actual DVH, either clinically approved or replanned, was within the DVH predicted by our model.
Our algorithm achieved 100% and 81.25% of true positive and true negative prediction, respectively. We have an overall accuracy of 87.5%, a misclassification rate of 27% and a precision of 100%.
Conclusion
We have successfully developed a model for reliable prediction of rectum and bladder DVH in prostate patients treated with a VMAT technique by applying a CNN pre-trained previously on a set of natural images. Our model demonstrates excellent performance, and its results validate the ability not only to detect sub-optimal plans retrospectively but also to predict achievable DVHs as a reliable guide and an optimal target for treatment planning. To our knowledge, this is the first attempt to apply transfer learning to the prediction of DVHs that accounts for the ground truth problem while training the model.