YOLOv8-ENAMR: An Improved YOLOv8 model with an efficient attention module for femoral neck fracture image detection

Abstract

The Internet of Things (IoT) and deep learning are being increasingly utilized in heal- thcare. Among these, the YOLOv8 (You Only Look Once version 8) target detection network plays a key role. As an advanced target detection algorithm,YOLOV8 enhan-ces the accuracy of bone and neck fracture image detection when integrated with the ENAMR module. The ENAMR module integrates with the residual network using an attentional mechanism. The aim of this study is to enhance the module's ability to deeply learn key features in complex medical imaging. By accurately capturing and interpreting these key features, the model is able to recognize subtle but critical signs of fractures, significantly boosting diagnostic accuracy and providing robust support to clinicians. Extensive experiments demonstrate that, compared to YOLOv8, the proposed method achieves a significantly improved detection performance in terms of accuracy and [email protected]. Additionally, there are notable improvements in mAP50- 95 and other metrics, indicating that YOLOv8-ENAMR provides superior detection capability for femoral neck data.

Ähnliche Arbeiten