Redes neurais convolucionais no diagnóstico de mielopatia cervical

Resumo

Objetivo As tecnologias de inteligência artificial são cada vez mais utilizadas em cirurgias de coluna como ferramentas diagnósticas. O objetivo do presente estudo foi avaliar a eficácia das redes neurais convolucionais no diagnóstico da mielopatia cervical (MC) em comparação à ressonância magnética (RM) cervical convencional.

Métodos O presente estudo foi transversal, descritivo e analítico. Cento e vinte e cinco participantes com diagnóstico clínico e radiológico de MC foram incluídos no estudo. Foram utilizadas imagens de RM sagital e axial em sequência ponderada em T2 da coluna cervical. Todas as imagens foram obtidas em 8 bits/pixel em duas categorias diferentes (MC e normal), tanto em vistas axiais quanto sagitais.

Resultados A validação transversal tripla evitou o sobreajuste (overfitting) durante o processo de treinamento. Duzentas e quarenta e duas imagens foram utilizadas para treinamento e teste do modelo criado para vistas axiais, que apresentou 97,44% de sensibilidade e 97,56% de especificidade. Duzentas e quarenta e nove imagens foram utilizadas para treinamento e teste do modelo criado para vistas sagitais, que apresentou 97,50% de sensibilidade e 97,67% de especificidade. Após o treinamento, a acurácia média foi de 96,7% (±1,53) para a vista axial e de 97,19% (±1,2) para a vista sagital.

Conclusão O deep learning (DL) apresentou grande melhora, especialmente na cirurgia de coluna. Observamos que a tecnologia de DL trabalha com maior acurácia do que em outros estudos na literatura para diagnóstico de MC.

Palavras-chave

aprendizado profundo - imagem por ressonância magnética - mielopatia - redes neurais artificiais - vértebras cervicais

Trabalho desenvolvido no Departamento de Medicina Física de Reabilitação, University of Health Sciences, Istanbul Kanuni Sultan Suleyman Training and Research Hospital, Istambul, Turquia.

Publikationsverlauf

Eingereicht: 08. März 2023

Angenommen: 05. Mai 2023

Artikel online veröffentlicht:
07. Dezember 2024

• Referências

• 1 Klineberg E. Cervical spondylotic myelopathy: a review of the evidence. Orthop Clin North Am 2010; 41 (02) 193-202
• 2 Cook C, Brown C, Isaacs R, Roman M, Davis S, Richardson W. Clustered clinical findings for diagnosis of cervical spine myelopathy. J Manual Manip Ther 2010; 18 (04) 175-180
• 3 Kadanka Z, Bednarík J, Vohánka S, Vlach O, Stejskal L, Chaloupka R. et al. Conservative treatment versus surgery in spondylotic cervical myelopathy: a prospective randomised study. Eur Spine J 2000; 9 (06) 538-544
• 4 Nagata K, Kiyonaga K, Ohashi T, Sagara M, Miyazaki S, Inoue A. Clinical value of magnetic resonance imaging for cervical myelopathy. Spine 1990; 15 (11) 1088-1096
• 5 Kanna RM, Kamal Y, Mahesh A, Venugopal P, Shetty AP, Rajasekaran S. The impact of routine whole spine MRI screening in the evaluation of spinal degenerative diseases. Eur Spine J 2017; 26 (08) 1993-1998
• 6 Maki S, Koda M, Kitamura M, Inada T, Kamiya K, Ota M. et al. Diffusion tensor imaging can predict surgical outcomes of patients with cervical compression myelopathy. Eur Spine J 2017; 26 (09) 2459-2466
• 7 Lee JG, Jun S, Cho YW, Lee H, Kim GB, Seo JB, Kim N. Deep Learning in Medical Imaging: General Overview. Korean J Radiol 2017; 18 (04) 570-584
• 8 Lundervold AS, Lundervold A. An overview of deep learning in medical imaging focusing on MRI. Z Med Phys 2019; 29 (02) 102-127
• 9 Galbusera F, Casaroli G, Bassani T. Artificial intelligence and machine learning in spine research. JOR Spine 2019; 2 (01) e1044
• 10 Abdelhafiz D, Yang C, Ammar R, Nabavi S. Deep convolutional neural networks for mammography: advances, challenges and applications. BMC Bioinformatics 2019; 20 (Suppl. 11) 281
• 11 Zhang S, Fukunaga T, Oka S, Orita H, Kaji S, Yube Y. et al. Concerns of quality, utility, and reliability of laparoscopic gastrectomy for gastric cancer in public video sharing platform. Ann Transl Med 2020; 8 (05) 196
• 12 Chang M, Canseco JA, Nicholson KJ, Patel N, Vaccaro AR. The role of machine learning in spine surgery: The future is now. Front Surg 2020; 7: 54
• 13 Schmidhuber J. Deep learning in neural networks: an overview. Neural Netw 2015; 61: 85-117
• 14 Kim JS, Merrill RK, Arvind V, Kaji D, Pasik SD, Nwachukwu CC. et al. Examining the ability of artificial neural networks machine learning models to accurately predict complications following posterior lumbar spine fusion. Spine 2018; 43 (12) 853-860
• 15 Al-Mefty O, Harkey LH, Middleton TH, Smith RR, Fox JL. Myelopathic cervical spondylotic lesions demonstrated by magnetic resonance imaging. J Neurosurg 1988; 68 (02) 217-222
• 16 Zhou Y, Kim SD, Vo K, Riew KD. Prevalence of cervical myelomalacia in adult patients requiring a cervical magnetic resonance imaging. Spine 2015; 40 (04) E248-E252
• 17 Emery SE. Cervical spondylotic myelopathy: diagnosis and treatment. J Am Acad Orthop Surg 2001; 9 (06) 376-388
• 18 Chen CJ, Lyu RK, Lee ST, Wong YC, Wang LJ. Intramedullary high signal intensity on T2-weighted MR images in cervical spondylotic myelopathy: prediction of prognosis with type of intensity. Radiology 2001; 221 (03) 789-794
• 19 Rajasekaran S, Vaccaro AR, Kanna RM, Schroeder GD, Oner FC, Vialle L. et al. The value of CT and MRI in the classification and surgical decision-making among spine surgeons in thoracolumbar spinal injuries. Eur Spine J 2017; 26 (05) 1463-1469
• 20 Hopkins BS, Weber II KA, Kesavabhotla K, Paliwal M, Cantrell DR, Smith ZA. Machine learning for the prediction of cervical spondylotic myelopathy: A post hoc pilot study of 28 participants. World Neurosurg 2019; 127: e436-e442
• 21 Jin R, Luk KD, Cheung JPY, Hu Y. Prognosis of cervical myelopathy based on diffusion tensor imaging with artificial intelligence methods. NMR Biomed 2019; 32 (08) e4114
• 22 Merali ZG, Witiw CD, Badhiwala JH, Wilson JR, Fehlings MG. Using a machine learning approach to predict outcome after surgery for degenerative cervical myelopathy. PLoS One 2019; 14 (04) e0215133