高级检索
当前位置: 首页 > 详情页

PatchCL-AE: Anomaly detection for medical images using patch-wise contrastive learning-based auto-encoder

文献详情

资源类型:
WOS体系:
Pubmed体系:

收录情况: ◇ SCIE

机构: [1]Beijing Institute of Technology, Beijing, 100081, China [2]Beijing Institute of Ophthalmology, Beijing Tongren Hospital, Capital Medical University, Beijing Ophthalmology & Visual Science Key Lab, Beijing, 100005, China
出处:
ISSN:

关键词: Medical anomaly detection Contrastive learning Patch loss

摘要:
Anomaly detection is an important yet challenging task in medical image analysis. Most anomaly detection methods are based on reconstruction, but the performance of reconstruction-based methods is limited due to over-reliance on pixel-level losses. To address the limitation, we propose a patch-wise contrastive learning-based auto-encoder for medical anomaly detection. The key contribution is the patch-wise contrastive learning loss that provides supervision on local semantics to enforce semantic consistency between corresponding input-output patches. Contrastive learning pulls corresponding patch pairs closer while pushing non-corresponding ones apart between input and output, enabling the model to learn local normal features better and improve discriminability on anomalous regions. Additionally, we design an anomaly score based on local semantic discrepancies to pinpoint abnormalities by comparing feature difference rather than pixel variations. Extensive experiments on three public datasets (i.e., brain MRI, retinal OCT, and chest X-ray) achieve state-of-the-art performance, with our method achieving over 99% AUC on retinal and brain images. Both the contrastive patch-wise supervision and patch-discrepancy score provide targeted advancements to overcome the weaknesses in existing approaches.Copyright © 2024 Elsevier Ltd. All rights reserved.

基金:
语种:
被引次数:
WOS:
PubmedID:
中科院(CAS)分区:
出版当年[2023]版:
大类 | 2 区 医学
小类 | 2 区 工程:生物医学 2 区 核医学
最新[2025]版:
大类 | 2 区 医学
小类 | 2 区 工程:生物医学 2 区 核医学
JCR分区:
出版当年[2022]版:
Q1 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING Q2 ENGINEERING, BIOMEDICAL
最新[2024]版:
Q1 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING Q2 ENGINEERING, BIOMEDICAL

影响因子: 最新[2024版] 最新五年平均 出版当年[2022版] 出版当年五年平均 出版前一年[2021版] 出版后一年[2023版]

第一作者:
第一作者机构: [1]Beijing Institute of Technology, Beijing, 100081, China
通讯作者:
通讯机构: [1]Beijing Institute of Technology, Beijing, 100081, China [2]Beijing Institute of Ophthalmology, Beijing Tongren Hospital, Capital Medical University, Beijing Ophthalmology & Visual Science Key Lab, Beijing, 100005, China
推荐引用方式(GB/T 7714):
APA:
MLA:

资源点击量:25910 今日访问量:0 总访问量:1514 更新日期:2025-07-01 建议使用谷歌、火狐浏览器 常见问题

版权所有©2020 首都医科大学附属北京同仁医院 技术支持:重庆聚合科技有限公司 地址:北京市东城区东交民巷1号(100730)