Classification of chest X-ray images by incorporation of medical domain knowledge into operation branch networks
Abstract Background This study was conducted to alleviate a common difficulty in chest X-ray image diagnosis: The attention region in a convolutional neural network (CNN) does not often match the doctor’s point of focus. The method presented herein, which guides the area of attention in CNN to a med...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
BMC
2023-05-01
|
| Series: | BMC Medical Imaging |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s12880-023-01019-0 |
| _version_ | 1797827216586833920 |
|---|---|
| author | Takumasa Tsuji Yukina Hirata Kenya Kusunose Masataka Sata Shinobu Kumagai Kenshiro Shiraishi Jun’ichi Kotoku |
| author_facet | Takumasa Tsuji Yukina Hirata Kenya Kusunose Masataka Sata Shinobu Kumagai Kenshiro Shiraishi Jun’ichi Kotoku |
| author_sort | Takumasa Tsuji |
| collection | DOAJ |
| description | Abstract Background This study was conducted to alleviate a common difficulty in chest X-ray image diagnosis: The attention region in a convolutional neural network (CNN) does not often match the doctor’s point of focus. The method presented herein, which guides the area of attention in CNN to a medically plausible region, can thereby improve diagnostic capabilities. Methods The model is based on an attention branch network, which has excellent interpretability of the classification model. This model has an additional new operation branch that guides the attention region to the lung field and heart in chest X-ray images. We also used three chest X-ray image datasets (Teikyo, Tokushima, and ChestX-ray14) to evaluate the CNN attention area of interest in these fields. Additionally, after devising a quantitative method of evaluating improvement of a CNN’s region of interest, we applied it to evaluation of the proposed model. Results Operation branch networks maintain or improve the area under the curve to a greater degree than conventional CNNs do. Furthermore, the network better emphasizes reasonable anatomical parts in chest X-ray images. Conclusions The proposed network better emphasizes the reasonable anatomical parts in chest X-ray images. This method can enhance capabilities for image interpretation based on judgment. |
| first_indexed | 2024-04-09T12:45:47Z |
| format | Article |
| id | doaj.art-0629e299231843059371ce9592363142 |
| institution | Directory Open Access Journal |
| issn | 1471-2342 |
| language | English |
| last_indexed | 2024-04-09T12:45:47Z |
| publishDate | 2023-05-01 |
| publisher | BMC |
| record_format | Article |
| series | BMC Medical Imaging |
| spelling | doaj.art-0629e299231843059371ce95923631422023-05-14T11:31:37ZengBMCBMC Medical Imaging1471-23422023-05-0123111810.1186/s12880-023-01019-0Classification of chest X-ray images by incorporation of medical domain knowledge into operation branch networksTakumasa Tsuji0Yukina Hirata1Kenya Kusunose2Masataka Sata3Shinobu Kumagai4Kenshiro Shiraishi5Jun’ichi Kotoku6Graduate School of Medical Care and Technology, Teikyo UniversityUltrasound Examination Center, Tokushima University HospitalDepartment of Cardiovascular Medicine, Tokushima University HospitalDepartment of Cardiovascular Medicine, Tokushima University HospitalCentral Radiology Division, Teikyo University HospitalDepartment of Radiology, Teikyo University School of MedicineGraduate School of Medical Care and Technology, Teikyo UniversityAbstract Background This study was conducted to alleviate a common difficulty in chest X-ray image diagnosis: The attention region in a convolutional neural network (CNN) does not often match the doctor’s point of focus. The method presented herein, which guides the area of attention in CNN to a medically plausible region, can thereby improve diagnostic capabilities. Methods The model is based on an attention branch network, which has excellent interpretability of the classification model. This model has an additional new operation branch that guides the attention region to the lung field and heart in chest X-ray images. We also used three chest X-ray image datasets (Teikyo, Tokushima, and ChestX-ray14) to evaluate the CNN attention area of interest in these fields. Additionally, after devising a quantitative method of evaluating improvement of a CNN’s region of interest, we applied it to evaluation of the proposed model. Results Operation branch networks maintain or improve the area under the curve to a greater degree than conventional CNNs do. Furthermore, the network better emphasizes reasonable anatomical parts in chest X-ray images. Conclusions The proposed network better emphasizes the reasonable anatomical parts in chest X-ray images. This method can enhance capabilities for image interpretation based on judgment.https://doi.org/10.1186/s12880-023-01019-0Attention mechanismChest X-ray imagesConvolutional neural networksDeep learningExplainable AI |
| spellingShingle | Takumasa Tsuji Yukina Hirata Kenya Kusunose Masataka Sata Shinobu Kumagai Kenshiro Shiraishi Jun’ichi Kotoku Classification of chest X-ray images by incorporation of medical domain knowledge into operation branch networks BMC Medical Imaging Attention mechanism Chest X-ray images Convolutional neural networks Deep learning Explainable AI |
| title | Classification of chest X-ray images by incorporation of medical domain knowledge into operation branch networks |
| title_full | Classification of chest X-ray images by incorporation of medical domain knowledge into operation branch networks |
| title_fullStr | Classification of chest X-ray images by incorporation of medical domain knowledge into operation branch networks |
| title_full_unstemmed | Classification of chest X-ray images by incorporation of medical domain knowledge into operation branch networks |
| title_short | Classification of chest X-ray images by incorporation of medical domain knowledge into operation branch networks |
| title_sort | classification of chest x ray images by incorporation of medical domain knowledge into operation branch networks |
| topic | Attention mechanism Chest X-ray images Convolutional neural networks Deep learning Explainable AI |
| url | https://doi.org/10.1186/s12880-023-01019-0 |
| work_keys_str_mv | AT takumasatsuji classificationofchestxrayimagesbyincorporationofmedicaldomainknowledgeintooperationbranchnetworks AT yukinahirata classificationofchestxrayimagesbyincorporationofmedicaldomainknowledgeintooperationbranchnetworks AT kenyakusunose classificationofchestxrayimagesbyincorporationofmedicaldomainknowledgeintooperationbranchnetworks AT masatakasata classificationofchestxrayimagesbyincorporationofmedicaldomainknowledgeintooperationbranchnetworks AT shinobukumagai classificationofchestxrayimagesbyincorporationofmedicaldomainknowledgeintooperationbranchnetworks AT kenshiroshiraishi classificationofchestxrayimagesbyincorporationofmedicaldomainknowledgeintooperationbranchnetworks AT junichikotoku classificationofchestxrayimagesbyincorporationofmedicaldomainknowledgeintooperationbranchnetworks |