AttR2U-Net: A Fully Automated Model for MRI Nasopharyngeal Carcinoma Segmentation Based on Spatial Attention and Residual Recurrent Convolution

Abstract

Radiotherapy is an essential method for treating nasopharyngeal carcinoma (NPC), and the segmentation of NPC is a crucial process affecting the treatment. However, manual segmentation of NPC is inefficient. Besides, the segmentation results of different doctors might vary considerably. To improve the efficiency and the consistency of NPC segmentation, we propose a novel AttR2U-Net model which automatically and accurately segments nasopharyngeal carcinoma from MRI images. This model is based on the classic U-Net and incorporates advanced mechanisms such as spatial attention, residual connection, recurrent convolution, and normalization to improve the segmentation performance. Our model features recurrent convolution and residual connections in each layer to improve its ability to extract details. Moreover, spatial attention is fused into the network by skip connections to pinpoint cancer areas more accurately. Our model achieves a DSC value of 0.816 on the NPC segmentation task and obtains the best performance compared with six other state-of-the-art image segmentation models.

Convolutional neural networks in auto-segmentation of nasopharyngeal carcinoma tumor — a systematic review and meta-analysis

“…(4) Flow and timing (sufficient time between index and reference, all data points included in analysis) [17], 3D Res-UNet [17,27,28], modified 3D Res-UNet [14,17], a mix of 2D and 3D Res-UNet [29,30], 3D VNet [15], 3D SI-UNet [18], 3D Nested UNet [14,19,31], 3D AttR2-UNet [14,21,31], 3D LW-UNet [32], and 3D DE-UNet [33]. Magnetic resonance imagining modality studies used hospital data [21-24, 26, 28, 29, 31-35], and CT studies often used the 2019 MICCAI StructSeg data [14,15,17,19] and hospital data [16,18,27,30].…”

Deep learning in MRI‐guided radiation therapy: A systematic review

“…Dense U‐Net densely connects convolutional layers in blocks, 48 ResU‐Net includes residual connections, 49 Retina U‐Net is a two‐stage network, RU‐Net includes recurrent connections, R2U‐Net adds residual recurrent connections 50 . Attention modules have also been added at the skip connections 51,52 . Both V‐Net 53 and nnUNet 54 were designed with 3D convolutional layers with nnUNet additionally automating preprocessing and learning parameter optimization.…”

“… 50 Attention modules have also been added at the skip connections. 51 , 52 Both V‐Net 53 and nnUNet 54 were designed with 3D convolutional layers with nnUNet additionally automating preprocessing and learning parameter optimization. Pix2pix uses U‐Net as the generator with a convolutional discriminator (PatchGAN).…”

CAFS: An Attention-Based Co-Segmentation Semi-Supervised Method for Nasopharyngeal Carcinoma Segmentation

“…After that, many segmentation algorithms for medical images were adapted from U-Net. Some scholars combined mechanisms such as attention mechanism and residual connectivity with U-Net to improve segmentation performance and segment the nasopharyngeal carcinoma [40][41][42]. In order to accommodate the volume segmentation of medical images, many U-Net-based 3D models have been developed as well [43,44].…”