Category Classification for Lung Computed Tomography of COVID-19 by Natural Language Processing in Japanese Radiology Report
Purpose: We screened patients admitted for coronavirus disease 2019 (COVID-19) via lung computed tomography (CT) using our own five-level categorization of imaging findings. We postulated that natural language processing (NLP) and machine learning (ML) could predict categorization using Japanese rad...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Society of Tokyo Women's Medical University
2023-12-01
|
| Series: | Tokyo Women's Medical University Journal |
| Subjects: | |
| Online Access: | https://www.jstage.jst.go.jp/article/twmuj/7/0/7_2023010/_pdf/-char/en |
| _version_ | 1797376672525189120 |
|---|---|
| author | Kazufumi Suzuki Yurie Shirai Tomohiro Kawaji Shuji Sakai |
| author_facet | Kazufumi Suzuki Yurie Shirai Tomohiro Kawaji Shuji Sakai |
| author_sort | Kazufumi Suzuki |
| collection | DOAJ |
| description | Purpose: We screened patients admitted for coronavirus disease 2019 (COVID-19) via lung computed tomography (CT) using our own five-level categorization of imaging findings. We postulated that natural language processing (NLP) and machine learning (ML) could predict categorization using Japanese radiology reports. Methods: We screened 528 patients, including 40 polymerase chain reaction (PCR) test-positive patients. We built ML models to predict these categories and the results of PCR tests using a CoreML 3 framework. Results: When categories 1-3 were considered positive predictions, the precision of the probability of PCR results predicted by radiologists was 0.24 with recall of 0.65; specificity of 0.83; accuracy of 0.82; and F1 score of 0.35. The precision of the ML models was 0.62 with recall if 0.53; specificity of 0.88; accuracy of 0.78; and F1 score of 0.57. The macro-averaged accuracy of the reproducibility of the ML models for classification was 0.47. The area under the curve of receiver operating curve for PCR tests was 0.644, whereas that for categories 1-3 was 0.680. Conclusion: Although the understanding of Japanese radiology reports by NLP is still limited, the use of categorization may increase its usefulness in screening for COVID-19. |
| first_indexed | 2024-03-08T19:42:05Z |
| format | Article |
| id | doaj.art-fe30edb15b0d4a36ba6d9802e0073ac1 |
| institution | Directory Open Access Journal |
| issn | 2432-6186 |
| language | English |
| last_indexed | 2024-03-08T19:42:05Z |
| publishDate | 2023-12-01 |
| publisher | Society of Tokyo Women's Medical University |
| record_format | Article |
| series | Tokyo Women's Medical University Journal |
| spelling | doaj.art-fe30edb15b0d4a36ba6d9802e0073ac12023-12-25T06:02:02ZengSociety of Tokyo Women's Medical UniversityTokyo Women's Medical University Journal2432-61862023-12-017010911410.24488/twmuj.2023010twmujCategory Classification for Lung Computed Tomography of COVID-19 by Natural Language Processing in Japanese Radiology ReportKazufumi Suzuki0Yurie Shirai1Tomohiro Kawaji2Shuji Sakai3Division of Diagnostic Imaging and Nuclear Medicine, Department of Radiology, Tokyo Women's Medical UniversityDivision of Diagnostic Imaging and Nuclear Medicine, Department of Radiology, Tokyo Women's Medical UniversityDivision of Diagnostic Imaging and Nuclear Medicine, Department of Radiology, Tokyo Women's Medical UniversityDivision of Diagnostic Imaging and Nuclear Medicine, Department of Radiology, Tokyo Women's Medical UniversityPurpose: We screened patients admitted for coronavirus disease 2019 (COVID-19) via lung computed tomography (CT) using our own five-level categorization of imaging findings. We postulated that natural language processing (NLP) and machine learning (ML) could predict categorization using Japanese radiology reports. Methods: We screened 528 patients, including 40 polymerase chain reaction (PCR) test-positive patients. We built ML models to predict these categories and the results of PCR tests using a CoreML 3 framework. Results: When categories 1-3 were considered positive predictions, the precision of the probability of PCR results predicted by radiologists was 0.24 with recall of 0.65; specificity of 0.83; accuracy of 0.82; and F1 score of 0.35. The precision of the ML models was 0.62 with recall if 0.53; specificity of 0.88; accuracy of 0.78; and F1 score of 0.57. The macro-averaged accuracy of the reproducibility of the ML models for classification was 0.47. The area under the curve of receiver operating curve for PCR tests was 0.644, whereas that for categories 1-3 was 0.680. Conclusion: Although the understanding of Japanese radiology reports by NLP is still limited, the use of categorization may increase its usefulness in screening for COVID-19.https://www.jstage.jst.go.jp/article/twmuj/7/0/7_2023010/_pdf/-char/encomputed tomographycovid-19machine learningnatural language processing |
| spellingShingle | Kazufumi Suzuki Yurie Shirai Tomohiro Kawaji Shuji Sakai Category Classification for Lung Computed Tomography of COVID-19 by Natural Language Processing in Japanese Radiology Report Tokyo Women's Medical University Journal computed tomography covid-19 machine learning natural language processing |
| title | Category Classification for Lung Computed Tomography of COVID-19 by Natural Language Processing in Japanese Radiology Report |
| title_full | Category Classification for Lung Computed Tomography of COVID-19 by Natural Language Processing in Japanese Radiology Report |
| title_fullStr | Category Classification for Lung Computed Tomography of COVID-19 by Natural Language Processing in Japanese Radiology Report |
| title_full_unstemmed | Category Classification for Lung Computed Tomography of COVID-19 by Natural Language Processing in Japanese Radiology Report |
| title_short | Category Classification for Lung Computed Tomography of COVID-19 by Natural Language Processing in Japanese Radiology Report |
| title_sort | category classification for lung computed tomography of covid 19 by natural language processing in japanese radiology report |
| topic | computed tomography covid-19 machine learning natural language processing |
| url | https://www.jstage.jst.go.jp/article/twmuj/7/0/7_2023010/_pdf/-char/en |
| work_keys_str_mv | AT kazufumisuzuki categoryclassificationforlungcomputedtomographyofcovid19bynaturallanguageprocessinginjapaneseradiologyreport AT yurieshirai categoryclassificationforlungcomputedtomographyofcovid19bynaturallanguageprocessinginjapaneseradiologyreport AT tomohirokawaji categoryclassificationforlungcomputedtomographyofcovid19bynaturallanguageprocessinginjapaneseradiologyreport AT shujisakai categoryclassificationforlungcomputedtomographyofcovid19bynaturallanguageprocessinginjapaneseradiologyreport |