Exploring the impact of network depth on 3D U-Net-based dose prediction for cervical cancer radiotherapy

Abstract

Purpose: The 3D U-Net deep neural network structure is widely employed for dose prediction in radiotherapy. However, the attention to the network depth and its impact on the accuracy and robustness of dose prediction remains inadequate.

Methods: 92 cervical cancer patients who underwent Volumetric Modulated Arc Therapy (VMAT) are geometrically augmented to investigate the effects of network depth on dose prediction by training and testing three different 3D U-Net structures with depths of 3, 4, and 5.

Results: For planning target volume (PTV), the differences between predicted and true values of D₉₈, D₉₉, and Homogeneity were statistically 1.00 ± 0.23, 0.32 ± 0.72, and -0.02 ± 0.02 for the model with a depth of 5, respectively. Compared to the other two models, these parameters were also better. For most of the organs at risk, the mean and maximum differences between the predicted values and the true values for the model with a depth of 5 were better than for the other two models.

Conclusions: The results reveal that the network model with a depth of 5 exhibits superior performance, albeit at the expense of the longest training time and maximum computational memory in the three models. A small server with two NVIDIA GeForce RTX 3090 GPUs with 24 G of memory was employed for this training. For the 3D U-Net model with a depth of more than 5 cannot be supported due to insufficient training memory, the 3D U-Net neural network with a depth of 5 is the commonly used and optimal choice for small servers.

Keywords: 3D U-Net; cervical cancer; dose prediction; network depth; radiotherapy.

Figure 1

The 3D U-Net architecture.

Figure 2

The flowchart of the training and testing process using 3D U-Net for dose prediction.

Figure 3

presents the prediction results obtained from the implementation of the 3D U-Net neural network models with different depths of 3, 4 and 5. The results of the models for depths 3, 4 and 5 are analyzed from top to bottom. From left to right, the input terms are contours, the true dose distribution map, the predicted dose distribution map, the difference map and the DVH comparison between the true dose and the predicted dose. (– Refers to the true dose distribution, - Refers to the predicted dose distribution).

Figure 4

shows the absolute dose error results of 10 testing plans for mean and maximum doses in each OAR and PTV. Where (A, B) are the mean and maximum violin distribution of absolute dose error values for the 3 D U-Net model with depth 3, (C, D) are the mean and maximum violin distribution of absolute dose error values for the 3 D U-Net model with depth 4, (E, F) are the mean and maximum violin distribution of absolute dose error values for the 3 D U-Net model with depth 5, respectively.