Lumbar Spine Vertebral Compression Fracture Case Diagnosis Using Machine Learning Methods on CT images

摘要

Fracture with partial collapse of vertebral bodies is generally known as vertebral compression fracture (VCF) which is vary considerably in types and causes and usually results from severity of trauma condition such as osteoporosis. Due to the medical error and the need to have sufficient experience to diagnose this disease, many researchers are thinking of using intelligent diagnostic methods by modeling the knowledge of radiologists using machine learning algorithms. The location and the severity of abnormalities on the vertebral body and its height are determined by the radiologist. In this auto-diagnostic system, we use morphometric features and measure three parts of vertebral body to diagnosis the vertebral compression and the location of the anomaly by segmentation and finding vertebral edges. We use support vector machine and k nearest neighbor as classifier and gained 88.3% accuracy and 92.5% sensitivity in diagnosis vertebral compression fracture and also achieved an accurate diagnosis of 86.2% to detect an abnormal location on 25 CT scans clinical images from the spinal column. Finally, we also compared our method with some other studies by implementing them on our used dataset.