Ulus Travma Acil Cerrahi Derg. 2022; 28(2): 196-201 | DOI: 10.14744/tjtes.2020.06944

Use of deep learning methods for hand fracture detection from plain hand radiographs

Kemal Üreten¹, Hüseyin Fatih Sevinç², Ufuk Ýðdeli³, Aslýhan Onay⁴, Yüksel Maraþ^5
1Department of Rheumatology, Kýrýkkale University Faculty of Medicine, Kýrýkkale-Turkey
²Department of Orthopedics and Traumatology, Nevþehir State Hospital, Nevþehir-Turkey
³Department of Internal Medicine, Kýrýkkale University Faculty of Medicine, Kýrýkkale-Turkey
⁴Department of Radiology, Baþkent University Faculty of Medicine, Ankara-Turkey
⁵Department of Rheumatology, Ankara City Hospital, Ankara-Turkey

BACKGROUND: Patients with hand trauma are usually examined in emergency departments of hospitals. Hand fractures are frequently observed in patients with hand trauma. Here, we aim to develop a computer-aided diagnosis (CAD) method to assist physicians in the diagnosis of hand fractures using deep learning methods.
METHODS: In this study, Convolutional Neural Networks (CNN) were used and the transfer learning method was applied. There were 275 fractured wrists, 257 fractured phalanx, and 270 normal hand radiographs in the raw dataset. CNN, a deep learning method, were used in this study. In order to increase the performance of the model, transfer learning was applied with the pre-trained VGG-16, GoogLeNet, and ResNet-50 networks.
RESULTS: The accuracy, sensitivity, specificity, and precision results in Group 1 (wrist fracture and normal hand) dataset were 93.3%, 96.8%, 90.3%, and 89.7%, respectively, with VGG-16, were 88.9%, 94.9%, 84.2%, and 82.4%, respectively, with Resnet-50, and were 88.1%, 90.6%, 85.9%, and 85.3%, respectively, with GoogLeNet. The accuracy, sensitivity, specificity, and precision results in Group 2 (phalanx fracture and normal hand) dataset were 84.0%, 84.1%, 83.8%, and 82.8%, respectively, with VGG-16, were 79.4%, 78.5%, 80.3%, and 79.7%, respectively, with Resnet-50, and were 81.7%, 81.3%, 82.1%, and 81.3%, respectively, with GoogLeNet.
CONCLUSION: We achieved promising results in this CAD method, which we developed by applying methods such as transfer learning, data augmentation, which are state-of-the-art practices in deep learning applications. This CAD method can assist physicians working in the emergency departments of small hospitals when interpreting hand radiographs, especially when it is difficult to reach qualified colleagues, such as night shifts and weekends.

Keywords: Computer-aided diagnosis, convolutional neural networks; data augmentation; deep learning; hand fractures; transfer learning.

---

Düz el radyografilerinden el kýrýklarýnýn tespiti için derin öðrenme yöntemlerinin kullanýlmasý

Kemal Üreten¹, Hüseyin Fatih Sevinç², Ufuk Ýðdeli³, Aslýhan Onay⁴, Yüksel Maraþ^5
1Kýrýkkale Üniversitesi Týp Fakültesi, Romatoloji Bilim Dalý, Kýrýkkale
²Nevþehir Devlet Hastanesi, Ortopedi ve Travmatoloji Kliniði, Nevþehir
³Kýrýkkale Üniversitesi Týp Fakültesi, Ýç Hastalýklarý Kliniði, Kýrýkkale
⁴Baþkent Üniversitesi Týp Fakültesi, Radyoloji Anabilim Dalý, Ankara
⁵Ankara Þehir Hastanesi, Romatoloji Bölümü, Ankara

AMAÇ: El travmasý olan hastalar genellikle hastanelerin acil servislerinde muayene edilir. El travmalarýnda el kemik kýrýklarý sýklýkla görülür. Bu çalýþmada, derin öðrenme yöntemlerini kullanarak el kýrýklarýnýn tanýsýnda hekimlere yardýmcý olmak için bilgisayar destekli bir yöntem geliþtirmeyi hedefledik.
GEREÇ VE YÖNTEM: Bu çalýþmada, konvolüsyonel sinir aðlarý kullanýlmýþ ve öðrenme transferi yöntemi uygulanmýþtýr. Veri kümesinde 275 el bilek kýrýðý, 257 falanks kýrýðý ve 270 normal el radyografisi vardý. Bu çalýþmada derin öðrenme yöntemi olan konvolüsyonel sinir aðlarý kullanýlmýþtýr. Modelin performansýný artýrmak için önceden eðitilmiþ VGG-16, GoogLeNet ve ResNet-50 aðlarý ile öðrenme transferi uygulanmýþtýr.
BULGULAR: Grup 1 (el bilek kýrýðý ve normal el) veri setindeki doðruluk, duyarlýlýk, özgüllük ve kesinlik sonuçlarý VGG-16 ile sýrasýyla %93.3, %96.8, %90.3 ve %89.7, ResNet-50 ile sýrasýyla %88.9, %94.9, %84.2 ve %82.4 ve GoogLeNet ile sýrasýyla %88.1, %90.6, %85.9 ve %85.3 idi. Grup 2 (falanks kýrýðý & normal el) veri setindeki doðruluk, duyarlýlýk, özgüllük ve kesinlik sonuçlarý, VGG-16 ile sýrasýyla %84.0, %84.1, %83.8 ve %82.8, Resnet-50 ile sýrasýyla %79.4, %78.5, %80.3 ve %79.7 ve GoogLeNet ile sýrasýyla %81.7, %81.3, %82.1 ve %81.3 idi.
TARTIÞMA: Derin öðrenme uygulamalarýnda son teknoloji uygulamalar olan transfer öðrenme, veri artýrma gibi yöntemler uygulayarak geliþtirdiðimiz bu bilgisayar destekli taný yönteminde umut verici sonuçlar elde ettik. Bu bilgisayar destekli taný yöntemi, el radyografilerini yorumlarken, özellikle gece vardiyalarý ve hafta sonlarý gibi uzman meslektaþlara ulaþmak zor olduðunda, küçük hastanelerin acil servislerinde çalýþan hekimlere yardýmcý olabilir.

Anahtar Kelimeler: Bilgisayar destekli taný, derin öðrenme; el kýrýklarý; konvolüsyonel sinir aðlarý; öðrenme transferi; veri artýrma.

---

CITE