One-Stage Deep Learning Approach for the Detection of Fractures in Cervical Spine

Fractures occurring in the cervical spine are a severe medical condition that can have fatal consequences or result in permanent paralysis that lasts a lifetime. Rapid identification and location of vertebral fractures are critical to prevent neurological deterioration and paralysis after trauma. Cervical spine fractures are injuries that must be handled with care. Some are stable, some are not, and mismanagement and misdiagnosis can lead to life-altering consequences. This paper proposes a YOLOv7(You Only Look Once v7) based method to locate the area that may contain fracture. Experimental results on RSNA 2022 [1] Cervical Spine Fracture Detection competition dataset have proven that the suggested model performs better than the most advanced methods in terms of higher detection efficiency, faster detection times, and high accuracy.