2024

Permanent URI for this collection

https://escholar.umt.edu.pk/handle/123456789/7502

Browse

Browsing 2024 by Author "HABIBA AFZAL, KHADIJA SAIF, NOORIA MAQSOOD and USMAN SARFARAZ"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    X-ray Vision 1.0: Deep Learning’s Role in Bone Fracture Identification
    (UMT, Lahore, 2024) HABIBA AFZAL, KHADIJA SAIF, NOORIA MAQSOOD and USMAN SARFARAZ
    Bone fractures are a common injury that requires accurate identification for proper diagnosis and treatment. In this paper, we propose a deep learning-based approach for bone fracture identification in X-ray images using various state-of-the-art architectures including ResNet50, DesNet201, VGG16 Architecture, Inception V3 architecture, and EfficientNetV2. These architectures have shown outstanding performance in image recognition tasks and are well-suited for the challenging task of bone fracture identification. Each CNN model is trained and applied on the augmented dataset, and the performance is regularly evaluated through metrics as such accuracy and precision. It not only assesses the overall classification performance but also has the ability to identify fractures types and location. To evaluate the performance of these models, we have collected our own dataset consisting of X-ray images from various sources, spanning nine different classes of bone fractures including chest, foot bone, hand, wrist, hip bone, leg, shoulder bone, ulna, radius, upper arm fracture and spinal cord. To enhance the robustness of the models and address the limited availability of training data, data augmentation techniques are employed. The dataset is preprocessed to ensure standardized input and to remove any noise or artifacts that may impact the performance of the models. Training and validation splits are established to train the models and assess their performance. Experimental results demonstrate the effectiveness of our approach, with the trained models achieving high accuracy in bone fracture detection. Furthermore, comparative analysis of the different architectures reveals variations in their performance, with some models outperforming others in terms of accuracy and efficiency. Our findings suggest that deep learning models, especially ResNet50 and EfficientNetV2, can be successfully employed for accurate bone fracture identification in X-ray images. Overall, this paper highlights the development and evaluation of deep learning models for bone fracture identification in X-ray images.