DOIONLINE 
DOIONLINE Publish In |
International Journal of Advances in Computer Science and Cloud Computing (IJACSCC)-IJACSCC |
 Journal Home Volume Issue |
||||||||
Issue |
Volume-12,Issue-1 ( May, 2024 ) | |||||||||
Paper Title |
Scaphoid Fracture Classification and Detection Using Convolutional Neural Network | |||||||||
Author Name |
Ming-Huwi Horng, Tai-Hua Yang | |||||||||
Affilition |
||||||||||
Pages |
16-20 | |||||||||
Abstract |
Scaphoid fractures are common injuries that occur in the wrist, particularly in young adults. Due to the location and nature of scaphoid fractures, they can be challenging to detect on standard X-rays. Convolutional Neural Networks (CNNs) are a type of deep learning algorithm commonly used for image recognition and analysis tasks. In the context of detecting scaphoid fractures from medical imaging such as X-rays, CNNs can play a crucial role in automating the detection process and assisting healthcare professionals in accurate diagnosis. In this paper, a two-stage fracture detection of scaphoid radiographs. The first one is the scaphoid region detection by using the Yolo v4 or Faster RCNN, the another is the fractures detection convolutional network. A powerful fracture detection CNN consists of ResNet, spatial feature pyramid and convolutional block attention module. Additionally, the Laws texture analysis is used to extract powerful features incorporating to feature maps to enhance the performances of fracture classification. Experimental results showed that the proposed CNN achieved high detection performances of precision, recall, F1-score and mAP are 0.843, 0.771, 0.835 and 0.796. The results reveal that the integration of Laws texture features and CNN feature maps can improve the fracture detection of radiographs. Keywords - Scaphoid fractures, Convolutional Neural Networks, Yolo v4, Spatial feature pyramid, Convolutional block attention module | |||||||||
| View Paper | ||||||||||
