Kuo Yung-Ming, Sheng Jia-Chun, Lo Chen-Hsuan, Wu You-Jie, Huang Chun-Rong
Department of Electronic EngineeringNational Formosa University Yunlin County 632 Taiwan.
Department of Computer Science and EngineeringNational Chung Hsing University Taichung 402 Taiwan.
IEEE Open J Eng Med Biol. 2025 Mar 28;6:413-419. doi: 10.1109/OJEMB.2025.3555818. eCollection 2025.
To assess the degree of adenocarcinoma, pathologists need to manually review pathology images. To reduce their burdens and achieve good inter-observer as well as intra-observer reproducibility, instance segmentation methods can help pathologists quantify shapes of gland cells and provide an automatic solution for computer-assisted grading of adenocarcinoma. However, segmenting individual gland cells of different sizes remains a difficult challenge in computer aided diagnosis. A novel cross-scale guidance integration transformer is proposed for gland cell instance segmentation. Our network contains a cross-scale guidance integration module to integrate multi-scale features learned from the pathology image. By using the integrated features from different field-of-views, the decoder with mask attention can better segment individual gland cells. Compared with recent task-specific deep learning methods, our method can achieve state-of-the-art performance in two public gland cell datasets. By imposing cross-scale encoder information, our method can retrieve accurate gland cell segmentation to assist the pathologists for computer-assisted grading of adenocarcinoma.
为了评估腺癌的程度，病理学家需要人工查看病理图像。为了减轻他们的负担并实现良好的观察者间以及观察者内的可重复性，实例分割方法可以帮助病理学家量化腺细胞的形状，并为腺癌的计算机辅助分级提供自动解决方案。然而，在计算机辅助诊断中，分割不同大小的单个腺细胞仍然是一项艰巨的挑战。本文提出了一种用于腺细胞实例分割的新型跨尺度引导集成变换器。我们的网络包含一个跨尺度引导集成模块，用于整合从病理图像中学习到的多尺度特征。通过使用来自不同视野的集成特征，带有掩码注意力的解码器可以更好地分割单个腺细胞。与近期特定任务的深度学习方法相比，我们的方法在两个公共腺细胞数据集中能够实现领先的性能。通过引入跨尺度编码器信息，我们的方法可以获得准确的腺细胞分割结果，以协助病理学家进行腺癌的计算机辅助分级。
