MAEMC-NET: a hybrid self-supervised learning method for predicting the malignancy of solitary pulmonary nodules from CT images
IntroductionPulmonary granulomatous nodules (PGN) often exhibit similar CT morphological features to solid lung adenocarcinomas (SLA), making preoperative differentiation challenging. This study aims to address this diagnostic challenge by developing a novel deep learning model.MethodsThis study pro...
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2025-02-01
|
| Series: | Frontiers in Medicine |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fmed.2025.1507258/full |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1823856566496919552 |
|---|---|
| author | Tianhu Zhao Tianhu Zhao Yong Yue Hang Sun Jingxu Li Yanhua Wen Yudong Yao Wei Qian Yubao Guan Shouliang Qi Shouliang Qi |
| author_facet | Tianhu Zhao Tianhu Zhao Yong Yue Hang Sun Jingxu Li Yanhua Wen Yudong Yao Wei Qian Yubao Guan Shouliang Qi Shouliang Qi |
| author_sort | Tianhu Zhao |
| collection | DOAJ |
| description | IntroductionPulmonary granulomatous nodules (PGN) often exhibit similar CT morphological features to solid lung adenocarcinomas (SLA), making preoperative differentiation challenging. This study aims to address this diagnostic challenge by developing a novel deep learning model.MethodsThis study proposes MAEMC-NET, a model integrating generative (Masked AutoEncoder) and contrastive (Momentum Contrast) self-supervised learning to learn CT image representations of intra- and inter-solitary nodules. A generative self-supervised task of reconstructing masked axial CT patches containing lesions was designed to learn intra- and inter-slice image representations. Contrastive momentum is used to link the encoder in axial-CT-patch path with the momentum encoder in coronal-CT-patch path. A total of 494 patients from two centers were included.ResultsMAEMC-NET achieved an area under curve (95% Confidence Interval) of 0.962 (0.934–0.973). These results not only significantly surpass the joint diagnosis by two experienced chest radiologists (77.3% accuracy) but also outperform the current state-of-the-art methods. The model performs best on medical images with a 50% mask ratio, showing a 1.4% increase in accuracy compared to the optimal 75% mask ratio on natural images.DiscussionThe proposed MAEMC-NET effectively distinguishes between benign and malignant solitary pulmonary nodules and holds significant potential to assist radiologists in improving the diagnostic accuracy of PGN and SLA. |
| format | Article |
| id | doaj-art-2ab6c4df836e46c689bbb6ce954ba40e |
| institution | Kabale University |
| issn | 2296-858X |
| language | English |
| publishDate | 2025-02-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Medicine |
| spelling | doaj-art-2ab6c4df836e46c689bbb6ce954ba40e2025-02-12T07:25:42ZengFrontiers Media S.A.Frontiers in Medicine2296-858X2025-02-011210.3389/fmed.2025.15072581507258MAEMC-NET: a hybrid self-supervised learning method for predicting the malignancy of solitary pulmonary nodules from CT imagesTianhu Zhao0Tianhu Zhao1Yong Yue2Hang Sun3Jingxu Li4Yanhua Wen5Yudong Yao6Wei Qian7Yubao Guan8Shouliang Qi9Shouliang Qi10College of Medicine and Biological Information Engineering, Northeastern University, Shenyang, ChinaKey Laboratory of Intelligent Computing in Medical Image, Ministry of Education, Northeastern University, Shenyang, ChinaDepartment of Radiology, Shengjing Hospital of China Medical University, Shenyang, ChinaSchool of Information Science and Engineering, Shenyang Ligong University, Shenyang, ChinaDepartment of Radiology, The First Affiliated Hospital of Guangzhou Medical University, Guangzhou, ChinaDepartment of Radiology, The Fifth Affiliated Hospital of Guangzhou Medical University, Guangzhou, ChinaDepartment of Electrical and Computer Engineering, Stevens Institute of Technology, Hoboken, NJ, United StatesCollege of Medicine and Biological Information Engineering, Northeastern University, Shenyang, ChinaDepartment of Radiology, The Fifth Affiliated Hospital of Guangzhou Medical University, Guangzhou, ChinaCollege of Medicine and Biological Information Engineering, Northeastern University, Shenyang, ChinaKey Laboratory of Intelligent Computing in Medical Image, Ministry of Education, Northeastern University, Shenyang, ChinaIntroductionPulmonary granulomatous nodules (PGN) often exhibit similar CT morphological features to solid lung adenocarcinomas (SLA), making preoperative differentiation challenging. This study aims to address this diagnostic challenge by developing a novel deep learning model.MethodsThis study proposes MAEMC-NET, a model integrating generative (Masked AutoEncoder) and contrastive (Momentum Contrast) self-supervised learning to learn CT image representations of intra- and inter-solitary nodules. A generative self-supervised task of reconstructing masked axial CT patches containing lesions was designed to learn intra- and inter-slice image representations. Contrastive momentum is used to link the encoder in axial-CT-patch path with the momentum encoder in coronal-CT-patch path. A total of 494 patients from two centers were included.ResultsMAEMC-NET achieved an area under curve (95% Confidence Interval) of 0.962 (0.934–0.973). These results not only significantly surpass the joint diagnosis by two experienced chest radiologists (77.3% accuracy) but also outperform the current state-of-the-art methods. The model performs best on medical images with a 50% mask ratio, showing a 1.4% increase in accuracy compared to the optimal 75% mask ratio on natural images.DiscussionThe proposed MAEMC-NET effectively distinguishes between benign and malignant solitary pulmonary nodules and holds significant potential to assist radiologists in improving the diagnostic accuracy of PGN and SLA.https://www.frontiersin.org/articles/10.3389/fmed.2025.1507258/fulllung cancerpulmonary granulomatous nodulesolid lung adenocarcinomasCT imageself-supervised learningmasked autoencoder |
| spellingShingle | Tianhu Zhao Tianhu Zhao Yong Yue Hang Sun Jingxu Li Yanhua Wen Yudong Yao Wei Qian Yubao Guan Shouliang Qi Shouliang Qi MAEMC-NET: a hybrid self-supervised learning method for predicting the malignancy of solitary pulmonary nodules from CT images Frontiers in Medicine lung cancer pulmonary granulomatous nodule solid lung adenocarcinomas CT image self-supervised learning masked autoencoder |
| title | MAEMC-NET: a hybrid self-supervised learning method for predicting the malignancy of solitary pulmonary nodules from CT images |
| title_full | MAEMC-NET: a hybrid self-supervised learning method for predicting the malignancy of solitary pulmonary nodules from CT images |
| title_fullStr | MAEMC-NET: a hybrid self-supervised learning method for predicting the malignancy of solitary pulmonary nodules from CT images |
| title_full_unstemmed | MAEMC-NET: a hybrid self-supervised learning method for predicting the malignancy of solitary pulmonary nodules from CT images |
| title_short | MAEMC-NET: a hybrid self-supervised learning method for predicting the malignancy of solitary pulmonary nodules from CT images |
| title_sort | maemc net a hybrid self supervised learning method for predicting the malignancy of solitary pulmonary nodules from ct images |
| topic | lung cancer pulmonary granulomatous nodule solid lung adenocarcinomas CT image self-supervised learning masked autoencoder |
| url | https://www.frontiersin.org/articles/10.3389/fmed.2025.1507258/full |
| work_keys_str_mv | AT tianhuzhao maemcnetahybridselfsupervisedlearningmethodforpredictingthemalignancyofsolitarypulmonarynodulesfromctimages AT tianhuzhao maemcnetahybridselfsupervisedlearningmethodforpredictingthemalignancyofsolitarypulmonarynodulesfromctimages AT yongyue maemcnetahybridselfsupervisedlearningmethodforpredictingthemalignancyofsolitarypulmonarynodulesfromctimages AT hangsun maemcnetahybridselfsupervisedlearningmethodforpredictingthemalignancyofsolitarypulmonarynodulesfromctimages AT jingxuli maemcnetahybridselfsupervisedlearningmethodforpredictingthemalignancyofsolitarypulmonarynodulesfromctimages AT yanhuawen maemcnetahybridselfsupervisedlearningmethodforpredictingthemalignancyofsolitarypulmonarynodulesfromctimages AT yudongyao maemcnetahybridselfsupervisedlearningmethodforpredictingthemalignancyofsolitarypulmonarynodulesfromctimages AT weiqian maemcnetahybridselfsupervisedlearningmethodforpredictingthemalignancyofsolitarypulmonarynodulesfromctimages AT yubaoguan maemcnetahybridselfsupervisedlearningmethodforpredictingthemalignancyofsolitarypulmonarynodulesfromctimages AT shouliangqi maemcnetahybridselfsupervisedlearningmethodforpredictingthemalignancyofsolitarypulmonarynodulesfromctimages AT shouliangqi maemcnetahybridselfsupervisedlearningmethodforpredictingthemalignancyofsolitarypulmonarynodulesfromctimages |