Cells Grouping Detection and Confusing Labels Correction on Cervical Pathology Images DOI Creative Commons
Wenbo Pang, Yi Ma,

Huiyan Jiang

и другие.

Bioengineering, Год журнала: 2024, Номер 12(1), С. 23 - 23

Опубликована: Дек. 30, 2024

Cervical cancer is one of the most prevalent cancers among women, posing a significant threat to their health. Early screening can detect cervical precancerous lesions in timely manner, thereby enabling prevention or treatment disease. The use pathological image analysis technology automatically interpret cells slices hot topic digital medicine research, as it reduce substantial effort required from pathologists identify and improve diagnostic efficiency accuracy. Therefore, we propose cell detection network based on collecting prior knowledge correcting confusing labels, called PGCC-Net. Specifically, utilize clinical break down task into multiple sub-tasks for grouping detection, aiming more effectively learn specific structure cells. Subsequently, merge region proposals achieve refined detection. In addition, according Bethesda system, definitions various categories abnormal are complex, boundaries ambiguous. Differences assessment criteria result ambiguously labeled cells, which poses challenge deep learning networks. To address this issue, perform labels correction module with feature similarity by constructing centers typical each category. Then, that easily confused mapped these order update cells' annotations. Accurate labeling greatly aids classification head network. We conducted experimental validation public dataset 7410 images private 13,526 images. results indicate our model outperforms state-of-the-art methods.

Язык: Английский

REHRSeg: Unleashing the power of self-supervised super-resolution for Resource-Efficient 3D MRI Segmentation DOI
Zhiyun Song,

Yinjie Zhao,

Xiaomin Li

и другие.

Neurocomputing, Год журнала: 2025, Номер unknown, С. 129425 - 129425

Опубликована: Янв. 1, 2025

Язык: Английский

Процитировано

0

Central loss guides coordinated Transformer for reliable anatomical landmark detection DOI
Qikui Zhu,

Yihui Bi,

Jie Chen

и другие.

Neural Networks, Год журнала: 2025, Номер unknown, С. 107391 - 107391

Опубликована: Март 1, 2025

Язык: Английский

Процитировано

0

An efficient framework based on large foundation model for cervical cytopathology whole slide image screening DOI

Jialong Huang,

Gaojie Li, Shichao Kan

и другие.

Biomedical Signal Processing and Control, Год журнала: 2025, Номер 107, С. 107859 - 107859

Опубликована: Март 29, 2025

Язык: Английский

Процитировано

0

Cells Grouping Detection and Confusing Labels Correction on Cervical Pathology Images DOI Creative Commons
Wenbo Pang, Yi Ma,

Huiyan Jiang

и другие.

Bioengineering, Год журнала: 2024, Номер 12(1), С. 23 - 23

Опубликована: Дек. 30, 2024

Cervical cancer is one of the most prevalent cancers among women, posing a significant threat to their health. Early screening can detect cervical precancerous lesions in timely manner, thereby enabling prevention or treatment disease. The use pathological image analysis technology automatically interpret cells slices hot topic digital medicine research, as it reduce substantial effort required from pathologists identify and improve diagnostic efficiency accuracy. Therefore, we propose cell detection network based on collecting prior knowledge correcting confusing labels, called PGCC-Net. Specifically, utilize clinical break down task into multiple sub-tasks for grouping detection, aiming more effectively learn specific structure cells. Subsequently, merge region proposals achieve refined detection. In addition, according Bethesda system, definitions various categories abnormal are complex, boundaries ambiguous. Differences assessment criteria result ambiguously labeled cells, which poses challenge deep learning networks. To address this issue, perform labels correction module with feature similarity by constructing centers typical each category. Then, that easily confused mapped these order update cells' annotations. Accurate labeling greatly aids classification head network. We conducted experimental validation public dataset 7410 images private 13,526 images. results indicate our model outperforms state-of-the-art methods.

Язык: Английский

Процитировано

0