EVALUATION OF DIAGNOSTIC ACCURACY OF THE SYSTEM FOR PULMONARY TUBERCULOSIS SCREENING BASED ON ARTIFICIAL NEURAL NETWORKS

The objective of the study: to evaluate the applicability of the automated system for detection of chest diseases during a regular mass screening of the population through assessment of universe parameters of diagnostic accuracy.

Subjects and methods. A retrospective diagnostic study was conducted. The index-test (the method being studied) implied distinction and analysis of X-ray films using the software based on convolutional neural networks of U-NET type, which were modified and trained for specific purposes. The reference method used was the double revision of the previously classified X-ray films by two qualified roentgenologists with work experience of 8-10 years. Two depersonalized samplings of digital X-ray films were used: Sample 1 (n = 140), the ratio of the norm and pathology made 50 : 50; Sample 2 (n = 150), the ratio of the norm and pathology made 95 : 5.

Results. The following parameters were set up for Samples 1 and 2 respectively: sensitivity ‒ 87.2 and 75.0%, specificity ‒ 60.0 and 53.5%, the prognostic value of the positive result ‒ 68.6 and 8.3%, the prognostic value of the negative result ‒ 82.4 and 97.5%, the area under characteristic curve ‒ 0.74 and 0.64.

Conclusions. The index test can be used only for mass regular screening in the population with low pre-test chances of pathology, which is confirmed by the prognostic value of the negative result (97.5%). This technology was recommended for the semiautomatic formation of pulmonary tuberculosis risk groups for consequent verification of the results by a roentgenologist.

1. Vasilyeva I.А., Belilovskiy E.M., Borisov S.E., Sterlikov S.А. Incidence, mortality and prevalence as indicators of tuberculosis burden in WHO regions, countries of the world and the Russian Federation. Part 2. Tuberculosis mortality Tuberculosis and Lung Diseases, 2017, vol. 95, no. 7, pp. 8-16. (In Russ.)

2. doi: 10.21292/2075-1230-2017-95-7-8-16.

3. Gusev A.V. Perspectives of neural networks and deep machine learning in the making decisions in healthcare. Vrach i Informatsionnye Tekhnologii, 2017, no. 3, pp. 92-105. (In Russ.)

4. Lange S., Walsh G. Luchevaya diagnostika zabolevaniy organov grudnoy kletki: rukovodstvo, atlas. (Russ. ed.: Sebastian Lange, Geraldine Walsh. Radiology of chest diseases). S.K. Ternovy, A.I. Shekhter, transl. and eds., Moscow, GEOTAR-Media Publ., 2010, 432 p.

5. Morozov S.P. Unified radiological information system. Meditsina: Tselevye Proekty, 2016, no. 24, pp. 54-55. (In Russ.)

6. Aguiar F.S., Torres R.C., Pinto J.V., Kritski A.L., Seixas J.M., Mello F.C. Development of two artificial neural network models to support the diagnosis of pulmonary tuberculosis in hospitalized patients in Rio de Janeiro, Brazil. Med. Biol. Eng. Comput., 2016, vol. 54, no. 11, pp. 1751-1759. doi:10.1007/s11517-016-1465-1.

7. Becker A.S., Blüthgen C., Phi van V.D., Sekaggya-Wiltshire C., Castelnuovo B., Kambugu A., Fehr J., Frauenfelder T. Detection of tuberculosis patterns in digital photographs of chest X-ray images using Deep Learning: feasibility study. Int. J. Tuberc. Lung Dis., 2018, vol. 22, no. 3, pp. 328-335. doi: 10.5588/ijtld.17.0520.

8. Bossuyt P.M., Reitsma J.B., Bruns D.E., Gatsonis C.A., Glasziou P.P., Irwig L., Lijmer J.G., Moher D., Rennie D., de Vet H.C., Kressel H.Y., Rifai N., Golub R.M., Altman D.G., Hooft L., Korevaar D.A., Cohen J.F. STARD 2015: An Updated List of Essential Items for Reporting Diagnostic Accuracy Studies. Radiology, 2015, 151516.

9. Dande P., Samant P. Acquaintance to Artificial Neural Networks and use of artificial intelligence as a diagnostic tool for tuberculosis: A review. Tuberculosis (Edinb.), 2018, vol. 108, pp. 1-9. doi: 10.1016/j.tube.2017.09.006.

10. Darsey J.A., Griffin W.O., Joginipelli S., Melapu V.K. Architecture and biological applications of artificial neural networks: a tuberculosis perspective. Methods Mol. Biol., 2015, no. 1260, pp. 269-283. doi: 10.1007/978-1-4939-2239-0_17.

11. de O. Souza Filho J.B., de Seixas J.M., Galliez R., de Bragança Pereira B., de Q. Mello F.C., Dos Santos A.M., Kritski A.L. A screening system for smear-negative pulmonary tuberculosis using artificial neural networks. Int. J. Infect. Dis., 2016, vol. 49, pp. 33-39. doi: 10.1016/j.ijid.2016.05.019.

12. El-Solh A.A., Hsiao C.B., Goodnough S., Serghani J., Grant B.J. Predicting active pulmonary tuberculosis using an artificial neural network. Chest, 1999, vol. 116, no. 4, pp. 968-973.

13. Elveren E., Yumuşak N. Tuberculosis disease diagnosis using artificial neural network trained with genetic algorithm. J. Med. Syst., 2011, vol. 35, no. 3, pp. 329-332. doi: 10.1007/s10916-009-9369-3.

14. Er O., Temurtas F., Tanrikulu A.C. Tuberculosis disease diagnosis using artificial neural networks. J. Med. Syst., 2010, vol. 34, no. 3, pp. 299-302.

15. Gao X.W., Qian Y. Prediction of Multidrug-Resistant TB from CT Pulmonary Images Based on Deep Learning Techniques. Mol. Pharm., 2018, doi: 10.1021/acs.molpharmaceut.7b00875.

16. Hwang S., Kim H.E., Jeong J., Kim H.J. A novel approach for tuberculosis screening based on deep convolutional neural networks. Proc. SPIE 9785. Med. Imaging, 2016, Computer-Aided Diagnosis, 97852W. doi:10.1117/12.2216198.

17. Lakhani P., Sundaram B. Deep learning at chest radiography: automated classification of pulmonary tuberculosis by using convolutional neural networks. Radiology, 2017, vol. 284, no. 2, pp. 574-582. doi: 10.1148/radiol.2017162326.

18. Lopes U.K., Valiati J.F. Pre-trained convolutional neural networks as feature extractors for tuberculosis detection. Comput Biol Med., 2017, vol. 89, pp. 135-143. doi: 10.1016/j.compbiomed.2017.08.001.

19. Mohamed E.I., Mohamed M.A., Moustafa M.H., Abdel-Mageed S.M., Moro A.M., Baess A.I., El-Kholy S.M. Qualitative analysis of biological tuberculosis samples by an electronic nose-based artificial neural network. Int. J. Tuberc. Lung Dis., 2017, vol. 21, no. 7, pp. 810-817. doi: 10.5588/ijtld.16.0677.

20. Morozov S., Ranschaert E. Why Europeans must care more about AI, machine learning (2017) – Available at: https://www.auntminnieeurope.com/index.aspx?sec=def&pno=24 (accessed 7 May 2018)

21. Seixas J.M., Faria J., Souza Filho J.B., Vieira A.F., Kritski A., Trajman A. Artificial neural network models to support the diagnosis of pleural tuberculosis in adult patients. Int. J. Tuberc. Lung Dis., 2013, vol. 17, no. 5, pp. 682-686.

22. Shamshirband S., Hessam S., Javidnia H., Amiribesheli M., Vahdat S., Petković D., Gani A., Kiah M.L.M. Tuberculosis Disease Diagnosis Using Artificial Immune Recognition System. Int. J. Med. Sci., 2014, vol. 11, no. 5, pp. 508-514. doi: 10.7150/ijms.8249.

23. Ting D.S.W., Yi P.H., Hui F. Clinical Applicability of Deep Learning System in Detecting Tuberculosis with Chest Radiography. Radiology, 2018, vol. 286, no. 2, pp. 729-731. doi: 10.1148/radiol.2017172407.

24. Viktor H.L., Cloete I., Beyers N. Extraction of rules for tuberculosis diagnosis using an artificial neural network. Methods Inf. Med., 1997, vol. 36, no. 2, pp. 160-162.

25. Vladzymyrskyy A., Mozgovoy V., Bondarenko S. A telemedicine network for managing multidrug-resistant tuberculosis. J. Telemed. Telecare, 2014, vol. 20, no. 2, pp. 113-114. doi:10.1177/1357633X13519054.