Tibial Injury Detection using Convolutional Neural Networks

• Matheus Bonfim da Rocha Federal University of Technology – Paraná (UTFPR)
• Bruno Uhlmann Marcato Federal University of Technology – Paraná (UTFPR)
• Wally auf der Strasse Pontifical Catholic University of Paraná (PUCPR)
• Maiara Mitiko Taniguchi State University of Maringá (UEM)
• José Luis Seixas Junior State University of Paraná (UNESPAR)
• Daniel Prado Campos Federal University of Technology – Paraná (UTFPR)
• Rafael Gomes Mantovani Federal University of Technology – Paraná (UTFPR)

DOI: https://doi.org/10.5753/kdmile.2025.247506

Abstract

Bone fractures are common traumas in hospital orthopedic departments. Thermal images in an orthopedic emergency setting indicate the exact location of the traumatic injury, facilitating the acquisition of radiological images and the correct patient positioning, avoiding the acquisition of complementary images. Despite significant progress in the area, there is still a need to develop thermal image automated techniques that provide robust, accurate, and detailed classification. Most studies segment manually regions of interest and establish threshold temperature values using specific thermal image processing software. Thus, in this study, we evaluated the use and effectiveness of convolutional neural networks for tibia injury detection with thermographic images. Experiments were performed with a real dataset developed by UTFPR/UFPR universities under the ethical guidelines of Resolution 466/12, with the approval of the Research Ethics Committees of the Federal and Hospital das Clínicas of the Federal University of Paraná (UFPR). The results were promising, showing that VGG19 could accurately recognize healthy and unhealthy patients with an average F-Score of 0.894. Although not statistically accurate like VGG results, traditional ML baselines could unveil some important image features that could explain the decision process, most related to the red channel values, saturation, and image texture.

Keywords: Deep Learning, Medical Diagnosis, Supervised Classification, Thermal Images

References

Aggarwal, C. C. Neural Networks and Deep Learning: A Textbook. Springer International Publishing, Cham, 2018.

Aguiar, G. J., Mantovani, R. G., Mastelini, S. M., de Carvalho, A. C., Campos, G. F., and Junior, S. B. A meta-learning approach for selecting image segmentation algorithm. Pattern Recognition Letters vol. 128, pp. 480–487, 2019.

Ahalya, R. K. and Snekhalatha, U. Cnn transformer for the automated detection of rheumatoid arthritis in hand thermal images. In Artificial Intelligence over Infrared Images for Medical Applications. Lecture Notes in Computer Science, vol. 15279. Springer, Cham, pp. 29–40, 2025.

Almigdad, A., Mustafa, A., Alazaydeh, S., Alshawish, M., Bani Mustafa, M., and Alfukaha, H. Bone fracture patterns and distributions according to trauma energy. Advances in Orthopedics 2022 (1): 8695916, 2022.

Bixel, M., Sivaraj, K., Timmen, M., Mohanakrishnan, V., Aravamudhan, A., Adams, S., and Adams, R. Angiogenesis is uncoupled from osteogenesis during calvarial bone regeneration. Nature Communications 15 (1): 4575, 2024.

Cakir, M., Tulum, G., Cuce, F., Yilmaz, K., Aralasmak, A., Isik, M., and Canbolat, H. Differential diagnosis of diabetic foot osteomyelitis and charcot neuropathic osteoarthropathy with deep learning methods. Journal of Imaging Informatics in Medicine 37 (5): 2454–2465, 2024.

der Strasse, W. A., Campos, D. P., Mendonça, C. J. A., et al. Detecting bone lesions in the emergency room with medical infrared thermography. BioMedical Engineering OnLine 21 (1): 35, 2022.

der Strasse, W. A., Campos, D. P., Mendonça, C. J. A., Soni, J. F., Mendes, J., and Nohama, P. Evaluation of tibia bone healing by infrared thermography: A case study. Journal of Multidisciplinary Healthcare vol. 14, pp. 3161–3175, 2021.

der Strasse, W. A., Campos, D. P., Mendonça, C. J. A., Soni, J. F., Tuon, F., Mendes, J., and Nohama, P. Evaluating physiological progression of chronic tibial osteomyelitis using infrared thermography. Research on Biomedical Engineering, 2022.

Ergene, M. C., Bayrak, A., and and, M. C. A new deep learning based end-to-end pipeline for hamstring injury detection in thermal images of professional football player. Quantitative InfraRed Thermography Journal 0 (0): 1–18, 2024.

Etehadtavakol, M., Sirati-Amsheh, M., Moallem, G., and Ng, E. Enhancing thyroid nodule classification: A comprehensive analysis of feature selection in thermography. Infrared Physics & Technology, 2025.

Gawade, S., Bhansali, A., Patil, K., and Shaikh, D. Application of the convolutional neural networks and supervised deep-learning methods for osteosarcoma bone cancer detection. Healthcare Analytics vol. 3, pp. 100153, 2023.

Khandakar, A., Chowdhury, M., Reaz, M., Ali, S., Hasan, M., Kiranyaz, S., and Malik, R. A machine learning model for early detection of diabetic foot using thermogram images. Computers in Biology and Medicine vol. 137, pp. 104838, 2021.

Magalhães, C., Tavares, J., Mendes, J., and Vardasca, R. Comparison of machine learning strategies for infrared thermography of skin cancer. Biomedical Signal Processing and Control vol. 69, pp. 102872, 2021.

Marsland, S. Machine learning: an algorithmic perspective. CRC press, 2015.

Reed, C., Saatchi, R., Burke, D., and Ramlakhan, S. Infrared thermal imaging as a screening tool for paediatric wrist fractures. Medical & Biological Engineering & Computing vol. 58, pp. 1549–1563, 2020.

Ribeiro, J., Mileski, M., Seixas Junior, J. L., Carvalho, L. F., and Mantovani, R. G. Image classification for precision agriculture: A coffee study case. In Anais do Computer on the Beach 2024, 2024.

Senalp, F. and Ceylan, M. Effects of the deep learning-based super-resolution method on thermal image classification applications. Multimedia Tools and Applications 81 (7): 9313–9330, 2022.

Shobayo, O., Saatchi, R., and Ramlakhan, S. Convolutional neural network to classify infrared thermal images of fractured wrists in pediatrics. Healthcare 12 (10): 994, 2024.

Simonyan, K. and Zisserman, A. Very deep convolutional networks for large-scale image recognition. In International Conference on Learning Representations. CoRR, San Diego, CA, USA, 2015.

Taspinar, Y. S. Light weight convolutional neural network and low-dimensional images transformation approach for classification of thermal images. Case Studies in Thermal Engineering vol. 41, pp. 102670, 2023.

Trejo-Chavez, O., Amezquita-Sanchez, J. P., Huerta-Rosales, J. R., et al. Automatic knee injury identification through thermal image processing and convolutional neural networks. Electronics 11 (23), 2022.