BPAT-UNet: Boundary preserving assembled transformer UNet for ultrasound thyroid nodule segmentation

Background and Objective: Accurate and efficient segmentation of thyroid nodules on ultrasound im-ages is critical for computer-aided nodule diagnosis and treatment. For ultrasound images, Convolutional neural networks (CNNs) and Transformers, which are widely used in natural images, cannot obtain sat-isfactory segmentation results, because they either cannot obtain precise boundaries or segment small objects. Methods: To address these issues, we propose a novel Boundary-preserving assembly Trans-former UNet (BPAT-UNet) for ultrasound thyroid nodule segmentation. In the proposed network, a Bound-ary point supervision module (BPSM), which adopts two novel self-attention pooling approaches, is de-signed to enhance boundary features and generate ideal boundary points through a novel method. Mean-while, an Adaptive multi-scale feature fusion module (AMFFM) is constructed to fuse features and chan-nel information at different scales. Finally, to fully integrate the characteristics of high-frequency local and low-frequency global, the Assembled transformer module (ATM) is placed at the bottleneck of the network. The correlation between deformable features and features-among computation is characterized by introducing them into the above two modules of AMFFM and ATM. As the design goal and even-tually demonstrated, BPSM and ATM promote the proposed BPAT-UNet to further constrain boundaries, whereas AMFFM assists to detect small objects. Results: Compared to other classical segmentation net-works, the proposed BPAT-UNet displays superior segmentation performance in visualization results and evaluation metrics. Significant improvement of segmentation accuracy was shown on the public thyroid dataset of TN3k with Dice similarity coefficient (DSC) of 81.64 % and 95th percentage of the asymmet-ric Hausdorff distance (HD95) of 14.06, whereas those on our private dataset were with DSC of 85.63 % and HD95 of 14.53, respectively. Conclusions: This paper presents a method for thyroid ultrasound im-age segmentation, which achieves high accuracy and meets the clinical requirements. Code is available at https://github.com/ccjcv/BPAT-UNet .& COPY; 2023 Elsevier B.V. All rights reserved.