Sequential and Iterative Auto-Segmentation of High-Risk Clinical Target Volume for Radiotherapy of Nasopharyngeal Carcinoma in Planning CT Images

Abstract

Background: Accurate segmentation of tumor targets is critical for maximizing tumor control and minimizing normal tissue toxicity. We proposed a sequential and iterative U-Net (SI-Net) deep learning method to auto-segment the high-risk primary tumor clinical target volume (CTVp1) for treatment planning of nasopharyngeal carcinoma (NPC) radiotherapy. Methods: The SI-Net is a variant of the U-Net architecture. The input of SI-Net includes one CT image, the CTVp1 contour on this image, and the next CT image. The output is the predicted CTVp1 contour on the next CT image. We designed the SI-Net, using the left side to learn the volumetric features and the right to localize the contour on the next image. Two prediction directions, one from inferior to superior (forward direction) and the other from superior to inferior (backward direction), were tested. The performance was compared between the SI-Net and the U-Net using Dice similarity coefficient (DSC), Jaccard index (JI), average surface distance (ASD), and Hausdorff distance (HD) metrics. Results: The DSC and JI values from the forward direction SI-Net model were 5 and 6% higher than those from the U-Net model (0.84 ± 0.04 vs. 0.80 ± 0.05 and 0.74 ± 0.05 vs. 0.69 ± 0.05, p < 0.001). The smaller ASD and HD values also indicated a better performance (2.8 ± 1.0 vs. 3.3 ± 1.0 mm and 8.7 ± 2.5 vs. 9.7 ± 2.7 mm, p < 0.01) for the SI-Net model. For the backward direction SI-Net model, the DSC and JI values were still better than those from the U-Net model ( p < 0.01), although there were no significant differences in ASD and HD. Conclusions: The SI-Net model preserved the continuity between adjacent images and thus improved the segmentation accuracy compared with the conventional U-Net model. This model has potential of improving the efficiency and consistence of CTVp1 contouring for NPC patients.

Automated clinical target volume delineation using deep 3D neural networks in radiation therapy of Non-small Cell Lung Cancer

“…This result was expected because of the 3D convolution kernels used in this study. Similar results were reported in [17] even with a semi 3D approach. Different from clinicians (or uniform-expansion in TPS) who handle the CT images and delineate the CTV contours slice-by-slice, our trained model took the full 3D volumetric information into account and naturally enforced the smoothness of the predicted contours in axial direction.…”

“…In the other words, this AM-GM inequality based function alone handled imbalance well. Similar to our approach, a dice coefficient based loss function to replace BCE for UNet was reported in [16] , [17] .…”

Application of Artificial Intelligence for Nasopharyngeal Carcinoma Management – A Systematic Review

“… 2. The paper mainly focused on the data level approach using re-sampling technique Xue et al (2020) 69 (China) NPC 150 (Combined CT and MRI images) Deep learning (Auto-contouring) 1. Segmentation: Deeplabv3+ 2.…”

“…MRI images were not used for training the model. Xue et al (2020) 46 (China) NPC 150 (Combined CT and MRI images) Deep learning (Auto-contouring) 1. Segmentation: SI-net To assess the model’s ability to segment high-risk tumors 1.…”

A Comprehensive Review on Radiomics and Deep Learning for Nasopharyngeal Carcinoma Imaging

“…Xue [ 139 ] proposed a sequential and iterative U-Net (SI-Net) to automatically segment the target volume of the primary tumour and compared it with a conventional U-Net. It was considered that the performance of the SI-Net was better than that of the U-Net (DSCs were 0.84 and 0.80, respectively).…”