Frequency of abnormal findings on chest radiographs: Analysis of chest X-ray reports in the metropolis

Introduction: To achieve the highest diagnostic accuracy of AI services in radiology, it is necessary to test and validate them on data sets that consider the balance of classes of various abnormalities. Information about the frequency of abnormal findings in practical healthcare is essential for creation of such datasets.

Objective: To establish the frequency of chest X-ray abnormalities using big data from the healthcare system of a Russian metropolis.

Materials and methods: We conducted an observational multicenter retrospective sample study by retrieving 562,077 chest X-ray reports dated February 18, 2021 to June 11, 2021 from the Unified Radiological Information Service of the Unified Medical Information Analysis System of the city of Moscow, which were then analyzed and automatically labeled using the Medlabel tool. The results were processed in Microsoft Excel and using the Python 3.9 programming language. Group differences were determined using the chi-square test.

Results: Among all analyzed reports, cardiomegaly was the most frequent abnormal finding (12.23 %), while the proportion of other abnormalities did not exceed 3.0 %. Among all abnormal chest X-rays, 79.60 % showed only one abnormality with cardiomegaly found in 80.78 % of cases. Among the reports with two or more abnormal findings, cardiomegaly was detected in only 43.36 % of cases, whereas opacities (64.98 %) and infiltration/consolidation (64.50 %) prevailed.

Conclusions: The proportion of abnormal chest X-rays was 16.7 %. In terms of the frequency of occurrence, cardiomegaly ranked first followed by focal pulmonary opacity and infiltration/consolidation. It is worth noting that the frequency of certain types of abnormalities varied significantly between the tests with one and several (two or more) abnormal findings, which should be taken into account when training and testing AI services.

1. Akhter Y, Singh R, Vatsa M. AI-based radiodiagnosis using chest X-rays: A review. Front Big Data. 2023;6:1120989. doi: 10.3389/fdata.2023.1120989

2. Fanni SC, Marcucci A, Volpi F, Valentino S, Neri E, Romei C. Artificial intelligence-based software with CE mark for chest X-ray interpretation: Opportunities and challenges. Diagnostics (Basel). 2023;13(12):2020. doi: 10.3390/diagnostics13122020

3. Ahn JS, Ebrahimian S, McDermott S, et al. Association of artificial intelligence-aided chest radiograph interpretation with reader performance and efficiency. JAMA Netw Open. 2022;5(8):e2229289. doi: 10.1001/jamanetworkopen.2022.29289

4. Wu JT, Wong KCL, Gur Y, et al. Comparison of chest radiograph interpretations by artificial intelligence algorithm vs radiology residents. JAMA Netw Open. 2020;3(10):e2022779. doi: 10.1001/jamanetworkopen.2020.22779

5. Rangarajan K, Muku S, Garg AK, et al. Artificial intelligence-assisted chest X-ray assessment scheme for COVID-19. Eur Radiol. 2021;31(8):6039-6048. doi: 10.1007/s00330-020-07628-5

6. Vasilev YA, Tyrov IA, Vladzymyrskyy AV, Arzamasov KM, Pestrenin LD, Shulkin IM. A New Model of Organizing Mass Screening Based on Stand-Alone Artificial Intelligence Used for Fluorography Image Triage. Public Health and Life Environment – PH&LE. 2023;31(11):23-32. (In Russ.) doi: 10.35627/2219-5238/2023-31-11-23-32

7. Tripathi S, Gabriel K, Dheer S, et al. Understanding biases and disparities in radiology AI datasets: A review. J Am Coll Radiol. 2023;20(9):836-841. doi: 10.1016/j.jacr.2023.06.015

8. Chen Y, Wan Y, Pan F. Enhancing multi-disease diagnosis of chest X-rays with advanced deep-learning networks in real-world data. J Digit Imaging. 2023;36(4):1332- 1347. doi: 10.1007/s10278-023-00801-4

9. Johnson JM, Khoshgoftaar TM. Survey on deep learning with class imbalance. J Big Data. 2019;6(1):27. doi: 10.1186/s40537-019-0192-5

10. Heye T, Segeroth M, Franzeck F, Vosshenrich J. Turning radiology reports into epidemiological data to track seasonal pulmonary infections and the COVID-19 pandemic. Eur Radiol. 2024;34(6):3624-3634. doi: 10.1007/s00330-023-10424-6

11. Vasilev YA, Goncharova IV, Vladzymyrskyy AV, Arzamasov KM, Pestrenin LD. Study of emphysematous changes in the population of Moscow using automated evaluation of radiological examinations. Menedzher Zdravookhraneniya. 2023;(9):23-36. (In Russ.) doi: 10.21045/1811-0185-2023-9-23-36

12. Watanabe Y, Nakagawa T, Fukai K, et al. Descriptive study of chest x-ray examination in mandatory annual health examinations at the workplace in Japan. PLoS One. 2022;17(1):e0262404. doi: 10.1371/journal.pone.0262404

13. Ronzhin LV, Astanin PA, Kokina DYu, Semenov SS, Arzamasov KM, Rauzina SE. Semantic analysis methods in the system for automated marking of the unstructured radiological chest examination protocols. Sotsial’nye Aspekty Zdorov’ya Naseleniya. 2023;69(1):12. (In Russ.) doi: 10.21045/2071-5021-2023-69-1-12

14. Kokina DYu, Gombolevskiy VA, Arzamasov KM, Andreychenko AE, Morozov SP. Possibilities and limitations of using machine text-processing tools in Russian radiology reports. Digital Diagnostics. 2022;3(4):374-383. (In Russ.) doi: 10.17816/DD101099

15. Twabi HH, Semphere R, Mukoka M, et al. Pattern of abnormalities amongst chest X-rays of adults undergoing computer-assisted digital chest X-ray screening for tuberculosis in Peri–Urban Blantyre, Malawi: A cross-sectional study. Trop Med Int Health. 2021;26(11):1427-1437. doi: 10.1111/tmi.13658

16. Watanabe Y, Nakagawa T, Fukai K, et al. Descriptive study of chest x-ray examination in mandatory annual health examinations at the workplace in Japan. PLoS One. 2022;17(1):e0262404. doi: 10.1371/journal.pone.0262404

17. McKee JL, Ferrier K. Is cardiomegaly on chest radiograph representative of true cardiomegaly: A cross-sectional observational study comparing cardiac size on chest radiograph to that on echocardiography. N Z Med J. 2017;130(1464):57-63.

18. Biharas Monfared A, Agha Farajollah S, Sabour F, Farzanegan R, Taghdisi S. Comparison of radiological findings of chest x-ray with echocardiography in determination of the heart size. Iran Red Crescent Med J. 2015;17(1):e18242. doi: 10.5812/ircmj.18242

19. Vasilev Y, Vladzymyrskyy A, Omelyanskaya O, Blokhin I, Kirpichev Y, Arzamasov K. AI-based CXR first reading: Current limitations to ensure practical value. Diagnostics (Basel). 2023;13(8):1430. doi: 10.3390/diagnostics13081430

20. Khalifa M, Albadawy M. AI in diagnostic imaging: revolutionising accuracy and efficiency. Computer Methods and Programs in Biomedicine Update. 2024;5:1-12. doi: 10.1016/j.cmpbup.2024.100146