Multimodal cardiomegaly classification with image-derived digital biomarkers
We investigate the problem of automatic cardiomegaly diagnosis. We approach this by developing classifiers using multimodal data enhanced by two image-derived digital biomarkers, the cardiothoracic ratio (CTR) and the cardiopulmonary area ratio (CPAR). The CTR and CPAR values are estimated using seg...
| Main Authors: | , , , , , |
|---|---|
| Format: | Conference item |
| Language: | English |
| Published: |
Springer
2022
|
| _version_ | 1797107904664305664 |
|---|---|
| author | Duvieusart, B Krones, F Parsons, G Tarassenko, L Papież, BW Mahdi, A |
| author_facet | Duvieusart, B Krones, F Parsons, G Tarassenko, L Papież, BW Mahdi, A |
| author_sort | Duvieusart, B |
| collection | OXFORD |
| description | We investigate the problem of automatic cardiomegaly diagnosis. We approach this by developing classifiers using multimodal data enhanced by two image-derived digital biomarkers, the cardiothoracic ratio (CTR) and the cardiopulmonary area ratio (CPAR). The CTR and CPAR values are estimated using segmentation and detection models. These are then integrated into a multimodal network trained simultaneously on chest radiographs and ICU data (vital sign values, laboratory values and metadata). We compare the predictive power of different data configurations with and without the digital biomarkers. There was a negligible performance difference between the XGBoost model containing only CTR and CPAR (accuracy 81.4%, F1 0.859, AUC 0.810) and black-box models which included full images (ResNet-50: accuracy 81.9%, F1 0.874, AUC 0.767; Multimodal: 81.9%, F1 0.873, AUC 0.768). We concluded that models incorporating domain knowledge-based digital biomarkers CTR and CPAR provide comparable performance to black-box multimodal approaches with the former providing better clinical explainability. |
| first_indexed | 2024-03-07T07:22:15Z |
| format | Conference item |
| id | oxford-uuid:066b746c-fe8e-420c-a870-13f95f14cca3 |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-03-07T07:22:15Z |
| publishDate | 2022 |
| publisher | Springer |
| record_format | dspace |
| spelling | oxford-uuid:066b746c-fe8e-420c-a870-13f95f14cca32022-10-28T13:16:43ZMultimodal cardiomegaly classification with image-derived digital biomarkersConference itemhttp://purl.org/coar/resource_type/c_5794uuid:066b746c-fe8e-420c-a870-13f95f14cca3EnglishSymplectic ElementsSpringer2022Duvieusart, BKrones, FParsons, GTarassenko, LPapież, BWMahdi, AWe investigate the problem of automatic cardiomegaly diagnosis. We approach this by developing classifiers using multimodal data enhanced by two image-derived digital biomarkers, the cardiothoracic ratio (CTR) and the cardiopulmonary area ratio (CPAR). The CTR and CPAR values are estimated using segmentation and detection models. These are then integrated into a multimodal network trained simultaneously on chest radiographs and ICU data (vital sign values, laboratory values and metadata). We compare the predictive power of different data configurations with and without the digital biomarkers. There was a negligible performance difference between the XGBoost model containing only CTR and CPAR (accuracy 81.4%, F1 0.859, AUC 0.810) and black-box models which included full images (ResNet-50: accuracy 81.9%, F1 0.874, AUC 0.767; Multimodal: 81.9%, F1 0.873, AUC 0.768). We concluded that models incorporating domain knowledge-based digital biomarkers CTR and CPAR provide comparable performance to black-box multimodal approaches with the former providing better clinical explainability. |
| spellingShingle | Duvieusart, B Krones, F Parsons, G Tarassenko, L Papież, BW Mahdi, A Multimodal cardiomegaly classification with image-derived digital biomarkers |
| title | Multimodal cardiomegaly classification with image-derived digital biomarkers |
| title_full | Multimodal cardiomegaly classification with image-derived digital biomarkers |
| title_fullStr | Multimodal cardiomegaly classification with image-derived digital biomarkers |
| title_full_unstemmed | Multimodal cardiomegaly classification with image-derived digital biomarkers |
| title_short | Multimodal cardiomegaly classification with image-derived digital biomarkers |
| title_sort | multimodal cardiomegaly classification with image derived digital biomarkers |
| work_keys_str_mv | AT duvieusartb multimodalcardiomegalyclassificationwithimagederiveddigitalbiomarkers AT kronesf multimodalcardiomegalyclassificationwithimagederiveddigitalbiomarkers AT parsonsg multimodalcardiomegalyclassificationwithimagederiveddigitalbiomarkers AT tarassenkol multimodalcardiomegalyclassificationwithimagederiveddigitalbiomarkers AT papiezbw multimodalcardiomegalyclassificationwithimagederiveddigitalbiomarkers AT mahdia multimodalcardiomegalyclassificationwithimagederiveddigitalbiomarkers |