Tumor detection on bronchoscopic images by unsupervised learning
Abstract The diagnosis and early identification of intratracheal tumors relies on the experience of the operators and the specialists. Operations by physicians with insufficient experience may lead to misdiagnosis or misjudgment of tumors. To address this issue, a datasets for intratracheal tumor de...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-01-01
|
| Series: | Scientific Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1038/s41598-024-81786-0 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841559603772391424 |
|---|---|
| author | Qingqing Liu Haoliang Zheng Zhiwei Jia Zhihui Shi |
| author_facet | Qingqing Liu Haoliang Zheng Zhiwei Jia Zhihui Shi |
| author_sort | Qingqing Liu |
| collection | DOAJ |
| description | Abstract The diagnosis and early identification of intratracheal tumors relies on the experience of the operators and the specialists. Operations by physicians with insufficient experience may lead to misdiagnosis or misjudgment of tumors. To address this issue, a datasets for intratracheal tumor detection has been constructed to simulate the diagnostic level of experienced specialists, and a Knowledge Distillation-based Memory Feature Unsupervised Anomaly Detection (KD-MFAD) model was proposed to learn from this simulated experience. The unsupervised training approach could effectively deal with the irregular features of the tumorous appearance. The Downward Deformable Convolution Module (DDC) allowed the encoding phase to provide more detailed internal airway environment features. The Memory Matrix based on Convolutional Block focusing (CB-Mem) helped the student model store more meaningful normal sample features during training and disrupted the reconstruction of “tumor” images. Our model achieved an AUC-ROC of 97.60%, Acc of 93.33%, and F1-score of 94.94% on our self-built intratracheal endoscopy datasets, improving baseline performance by 5 to 10%. Our model also demonstrated superior performance over existing models in the public datasets in the same field. |
| format | Article |
| id | doaj-art-779529600f6d4f599795c911dd2744f0 |
| institution | Kabale University |
| issn | 2045-2322 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj-art-779529600f6d4f599795c911dd2744f02025-01-05T12:22:05ZengNature PortfolioScientific Reports2045-23222025-01-0115111410.1038/s41598-024-81786-0Tumor detection on bronchoscopic images by unsupervised learningQingqing Liu0Haoliang Zheng1Zhiwei Jia2Zhihui Shi3Department of Pulmonary and Critical Care Medicine, The Second Xiangya Hospital, Central South UniversitySchool of Electrical and Information Engineering, Changsha University of Science and TechnologySchool of Electrical and Information Engineering, Changsha University of Science and TechnologyDepartment of Pulmonary and Critical Care Medicine, The Second Xiangya Hospital, Central South UniversityAbstract The diagnosis and early identification of intratracheal tumors relies on the experience of the operators and the specialists. Operations by physicians with insufficient experience may lead to misdiagnosis or misjudgment of tumors. To address this issue, a datasets for intratracheal tumor detection has been constructed to simulate the diagnostic level of experienced specialists, and a Knowledge Distillation-based Memory Feature Unsupervised Anomaly Detection (KD-MFAD) model was proposed to learn from this simulated experience. The unsupervised training approach could effectively deal with the irregular features of the tumorous appearance. The Downward Deformable Convolution Module (DDC) allowed the encoding phase to provide more detailed internal airway environment features. The Memory Matrix based on Convolutional Block focusing (CB-Mem) helped the student model store more meaningful normal sample features during training and disrupted the reconstruction of “tumor” images. Our model achieved an AUC-ROC of 97.60%, Acc of 93.33%, and F1-score of 94.94% on our self-built intratracheal endoscopy datasets, improving baseline performance by 5 to 10%. Our model also demonstrated superior performance over existing models in the public datasets in the same field.https://doi.org/10.1038/s41598-024-81786-0Medical image recognitionIntratracheal tumorArtificial IntelligenceUnsupervised learningKnowledge distillation |
| spellingShingle | Qingqing Liu Haoliang Zheng Zhiwei Jia Zhihui Shi Tumor detection on bronchoscopic images by unsupervised learning Scientific Reports Medical image recognition Intratracheal tumor Artificial Intelligence Unsupervised learning Knowledge distillation |
| title | Tumor detection on bronchoscopic images by unsupervised learning |
| title_full | Tumor detection on bronchoscopic images by unsupervised learning |
| title_fullStr | Tumor detection on bronchoscopic images by unsupervised learning |
| title_full_unstemmed | Tumor detection on bronchoscopic images by unsupervised learning |
| title_short | Tumor detection on bronchoscopic images by unsupervised learning |
| title_sort | tumor detection on bronchoscopic images by unsupervised learning |
| topic | Medical image recognition Intratracheal tumor Artificial Intelligence Unsupervised learning Knowledge distillation |
| url | https://doi.org/10.1038/s41598-024-81786-0 |
| work_keys_str_mv | AT qingqingliu tumordetectiononbronchoscopicimagesbyunsupervisedlearning AT haoliangzheng tumordetectiononbronchoscopicimagesbyunsupervisedlearning AT zhiweijia tumordetectiononbronchoscopicimagesbyunsupervisedlearning AT zhihuishi tumordetectiononbronchoscopicimagesbyunsupervisedlearning |