Spatial localization of bone sarcomas on radiographs using a one-stage neural network model

The purpose of the research is to investigate the possibility of using the YOLOv8l‒seg single-stage neural network model for automatic spatial localization of bone sarcomas on radiographs.

Methods. To train the neural network model, a data set was synthesized based on annotated digitized radiographs with confirmed bone sarcomas and radiographs without pathology. The images in the dataset were preprocessed (standardization of size and contrast) and divided into training (1003 images) and validation (204 images) samples. To solve the problem of segmentation of bone sarcoma instances, the YOLOv8l-seg neural network model with pretrained weights on the COCO dataset was used. The retraining process was implemented using the deep transfer learning method. To prevent overfitting of the neural network model, an early stop method was applied that tracks the value of the validation loss function.

Results. The neural network model was tested on an independent test sample consisting of 148 digitized radio graphs with confirmed bone sarcomas and X-rays without pathology, which were not involved in the training and vali dation stages. The efficiency assessment showed that in 78.2% of cases, the model not only correctly classifies the tumor, but also outlines its spatial boundaries with acceptable accuracy, and the AUC-ROC metric value of 0.951 confirms the high discriminative ability of the model in distinguishing pathology from the norm. This result, although not a reference, confirms the applicability of the proposed approach to solving the problem of localization and con touring of bone sarcomas on radiographs.

Conclusion. Based on the results of the analysis, it can be concluded that, unlike models that provide only a probabilistic estimate, the main advantage of the proposed solution lies in the segmentation of bone sarcoma instances, which can be directly used by a radiologist as a decision support tool.

1. Aleskerova Kh.A., Romantsova O.M., Khairullova V.V. Differencial'naya diagnostika teleangiektaticheskoj osteosarkomy i anevrizmal'noj kostnoj kisty (obzor literatury). Rossiyskiy zhurnal detskoy gematologii i onkologii = Russian Journal of Pediatric Hematology and Oncology. 2024;11(2):67-73. (In Russ.) https://doi.org/10.21682/2311-1267-2024-11-2-67-73

2. Emelyanova N.V., Skvortsova A.V., Milovanova E.S., Yamshchikov O.N. Patofiziologicheskie osobennosti klinicheskoj diagnostiki hondrosarkom. International Journal of Medicine and Psychology. 2022;5(3):124-132. (In Russ.)

3. Mandat T.S. Metastatic High-Grade Intimal Sarcoma of the Middle Cranial Fossa. Journal on Oncology. 2022;2(2). https://doi.org/10.52768/2692-563x/1053

4. Kontorovich D.S., Pronkina E.V., Nudnov N.V. Vneskeletnaya sarkoma Yuinga malogo taza (obzor literatury i klinicheskoe nablyudenie). Meditsinskaya vizualizatsiya = Medical Visualization. 2021;(25):150-156. (In Russ.) https://doi.org/10.24835/1607-0763-924

5. Matcuk G., Waldman L., Fields B.K. Conventional radiography for the assessment of focal bone lesions of the appendicular skeleton: fundamental concepts in the modern imaging era. Skeletal Radiol. 2025;54:1391–1406. https://doi.org/10.1007/s00256-024-04854-6

6. Ochkurenko A.A., Petrichenko A.V., Romanov P.A., et al. Sovremennye vozmozhnosti i trudnosti diagnostiki kostnyh sarkom u detej. Pediatriya. Zhurnal im. G.N. Speranskogo = Pediatrics. Journal named after G.N. Speransky. 2020;99(4):56-62. (In Russ.) https://doi.org/10.24110/0031-403X-2020-99-4-56-62

7. Bulycheva I.V., Fedorova A.V. Osnovy algoritma rentgenomorfologicheskogo issledovaniya kostnyh preparatov. Onkopatologiya = Oncopathology. 2021;4(1):18-31. (In Russ.) https://doi.org/10.17650/2618-7019-2021-4-1-2-18-31

8. Kiselev A.V., Kuleshova E.A., Tanygin M.O., at al. Analiz effektivnosti primeneniya arhitektury U-net dlya klassifikacii i segmentacii gliomy na MRT-snimkah. Izvestiya YugoZapadnogo gosudarstvennogo universiteta. Serija: Upravlenie, vychislitel'naja tekhnika, informatika. Meditsinskoe priborostroenie = Proceedings of the Southwest State University. Series: Control, Computer Engineering, Information Science. Medical Instruments Engineering. 2024;14(3):104-120. (In Russ.) https://doi.org/10.21869/2223-1536-2024-14-3-104-120

9. Brezhnev A.V., Tomakova R.A., Chernykh E.V. Informacionnaya sistema dlya prognozirovaniya recidivov infarkta miokarda, realizovannaya kak mobil'noe prilozhenie. Informatsionnoye obshchestvo = Information Society. 2023;(1):116-126. (In Russ.) https://doi.org/10.52605/16059921_2023_01_116

10. Tomakova R.A., Dzyubin I.A., Brezhnev A.V. Metod i algoritm obucheniya svertochnoj nejronnoj seti, prednaznachennoj dlya intellektual'noj sistemy raspoznavaniya melanomy. Izvestiya Yugo-Zapadnogo gosudarstvennogo universiteta. Serija: Upravlenie, vychislitel'naja tekhnika, informatika. Meditsinskoe priborostroenie = Proceedings of the Southwest State University. Series: Control, Computer Engineering, Information Science. Medical Instruments Engineering. 2022;12(1):65-83. (In Russ.) https://doi.org/10.21869/2223-1536-2022-12-1-65-83

11. Mishin I.O., Tanygin M.O., Kiselev A.V. Metod analiza ocifrovannyh rentgenovskih snimkov grudnoj kletki dlya differencial'noj diagnostiki infekcionnyh zabolevanij dyhatel'noj sistemy. Vestnik Voronezhskogo gosudarstvennogo universiteta. Seriya: Sistemnyy analiz i informatsionnyye tekhnologii = Proceedings of Voronezh State University. Series: Systems Analysis and Information Technologies. 2024;(4):143-155. (In Russ.) https://doi.org/10.17308/sait/1995-5499/2024/4/143-155

12. Pan C., Lian L., Chen J., Huang R. Femur Tumor Net: Bone tumor classification in the proximal femur using DenseNet model based on radiographs. Journal of Bone Oncology. 2023;42:7639. https://doi.org/10.1016/j.jbo.2023.100504

13. Gawade S., Bhansali A., Patil K., Shaikh D. Application of the convolutional neural networks and supervised deep-learning methods for osteosarcoma bone cancer detection. Healthcare Analytics. 2023;3:100153. https://doi.org/10.1016/j.health.2023.100153

14. He Y., Pan I., Bao B., Halsey K., Chang M., et al. Deep learning-based classification of primary bone tumors on radiographs. EBioMedicine: A Preliminary Study. 2020;62:103121. https://doi.org/10.1016/j.ebiom.2020.103121

15. Yao S., Huang Y., Wang X., Zhang Y., Paixao I. C., Wang Z., et al. A Radiograph Dataset for the Classification, Localization, and Segmentation of Primary Bone Tumors. Scientific Data. 2024;(12):88. https://doi.org/10.1038/s41597-024-04311-y

16. Lapushkin A.G., Gavrilov D.A., Shchelkunov N.N., Bakeev R.N. Osnovnye podhody k podgotovke vizual'nyh dannyh dlya obucheniya nejrosetevyh algoritmov. Iskusstvennyy intellekt i prinyatiye resheniy = Artificial Intelligence and Decision Making. 2021;(4):62-74. (In Russ.) https://doi.org/10.14357/20718594210406

17. Iman M., Arabnia H.R., Rasheed K. A review of deep transfer learning and recent advancements.

18. Technologies. 2023;11(2):1-14. https://doi.org/10.3390/technologies11020040

19. Zhigalov A.A., Ivashchuk O.A., Biryukova T.K., Fedorov V.I. Metody nejrosetevoj detekcii sel'skohozyajstvennyh zhivotnyh v plotnyh dinamicheskih gruppah na izobrazheniyah. Iskusstvennyy intellekt i prinyatiye resheniy = Artificial Intelligence and Decision Making. 2023;(1):55-66. (In Russ.) https://doi.org/10.14357/20718594230106

20. Kiselev A.V., Brusentsev N.S., Kuleshova E.A. Analiz effektivnosti primeneniya dvuhetapnyh nejrosetevyh modelej dlya rannego obnaruzheniya lesnyh pozharov. Izvestiya Yugo-Zapadnogo gosudarstvennogo universiteta. Serija: Upravlenie, vychislitel'naja tekhnika, informatika. Meditsinskoe priborostroenie = Proceedings of the Southwest State University. Series: Control, Computer Engineering, Information Science. Medical Instruments Engineering. 2024;14(1):8-23. (In Russ.) https://doi.org/10.21869/2223-1536-2024-14-1-8-23

21. Kiselev A.V., Shatalova O.V., Protasova Z.U., et al. Modeli latentnyh prediktorov v intellektual'nyh sistemah prognozirovaniya sostoyaniya zhivyh sistem. Izvestiya YugoZapadnogo gosudarstvennogo universiteta. Serija: Upravlenie, vychislitel'naja tekhnika, informatika. Meditsinskoe priborostroenie = Proceedings of the Southwest State University. Series: Control, Computer Engineering, Information Science. Medical Instruments Engineering. 2020;10(1):114-133. (In Russ.)