基于目标定位的眼睑肿瘤自动诊断-《中国医学物理学杂志》

文章信息/Info

作者:: 蒋杰伟1; 刘海洋1; 蔺彤彤1; 裴梦杰2; 魏戌盟3; 巩稼民1; 4; 李中文5; 1.西安邮电大学电子工程学院，陕西西安 710121； 2.西安邮电大学计算机学院，陕西西安 710121； 3.西安邮电大学通信与信息工程学院，陕西西安 710121； 4.西安邮电大学现代邮政学院，陕西西安 710121； 5.温州医科大学宁波市眼科医院，浙江宁波 315000

Author(s):: JIANG Jiewei1; LIU Haiyang1; LIN Tongtong1; PEI Mengjie2; WEI Xumeng3; GONG Jiamin1; 4; LI Zhongwen5; 1. School of Electronic Engineering, Xian University of Posts and Telecommunications, Xian 710121, China 2. School of Computer Science and Technology, Xian University of Posts and Telecommunications, Xian 710121, China 3. School of Communications and Information Engineering, Xian University of Posts and Telecommunications, Xian 710121, China 4. School of Modern Post, Xian University of Posts and Telecommunications, Xian 710121, China 5. Ningbo Eye Hospital, Wenzhou Medical University, Ningbo 315000, China

Keywords:: Keywords: eyelid tumor fine-grained localization dual attention mechanism residual network

摘要:: 眼睑肿瘤是导致视力下降甚至致盲的严重眼病，良恶性结构的相似性导致缺乏临床经验的眼科医生不易区分。针对此问题，提出一种基于两阶段目标定位算法和融合双重注意力机制的残差网络，以实现眼睑肿瘤良恶性的自动诊断。首先，利用FCOS算法自动定位眼眶的整体轮廓，去除背景区域和周围噪声；然后，在眼眶内部精细化定位眼睑肿瘤病灶区域；最后，将病灶区域输入到融合双重注意力机制的残差网络（ResNet101_CBAM），实现良恶性的自动诊断。实验结果表明目标定位算法对眼睑肿瘤病灶的定位平均精度为0.821；与ResNet101相比，ResNet101_CBAM在眼睑肿瘤分类中的敏感度和准确率分别提高4.7%和3.0%，表明该模型在眼睑肿瘤良恶性自动诊断中表现出较优性能。

Abstract:: Abstract: Eyelid tumor is a serious eye disease that leads to vision loss or even blindness. The similarity between benign and malignant characteristics makes it difficult for ophthalmologists lacking clinical experience to distinguish between them. To address the problem, a method (ResNet101_CBAM) based on two-stage target localization using fully convolutional one-stage object detection (FCOS) and residual network incorporating a dual attention mechanism is proposed to realize the automatic diagnosis of benign and malignant eyelid tumors. FCOS is used to automatically localize the overall contour of the orbit, removing the background and surrounding noise, and then finely localize the tumor lesion inside the orbit. The obtained lesion region is input into ResNet101_CBAM for the automatic diagnosis of benign and malignant eyelid tumors. The experimental results show that the average precision of the target localization algorithm for tumor lesion is 0.821, and that compared with ResNet101, ResNet101_CBAM improves the sensitivity and accuracy in eyelid tumor classification by 4.7% and 3.0%, respectively, indicating that the proposed model has superior performances in the automatic diagnosis of benign and malignant eyelid tumors.

备注/Memo:: 【收稿日期】2023-08-06 【基金项目】国家自然科学基金（62276210，82201148）；陕西省自然科学研究计划（2022JM-380）；陕西省大学生创新创业训练项目（S202311664128X）；浙江省自然科学基金（LQ22H120002）；浙江省医药卫生科技项目（2022RC069, 2023KY1140）；宁波市自然科学基金（2023J390）；西安邮电大学研究生创新基金（CXJJZL2022008）【作者简介】蒋杰伟，博士，硕士生导师，研究方向：深度学习、机器学习和医疗图像处理，E-mail: jiangjw924@126.com 【通信作者】巩稼民，博士，教授，研究方向：光通信技术与图像处理，E-mail: gjm@xupt.edu.cn