Examination of Optimal Frequency Processing in X-ray Images for Occult Femoral Neck Fractures

Abstract

Osteosclerotic images are known as an image appearance of occult femoral neck fractures in X-ray images. The aim of this study is to investigate frequency enhancement processing that improves the visibility of the osteosclerotic image. We acquired three sclerotic bone images with different thicknesses, and self-made bone equivalent phantoms were set up on a pelvic phantom. The frequency processing type and high-density enhancement coefficients were applied to the X-ray images taken at RF-A(1.0, 2.0), C(2.0, 4.0), D(1.0), and H(2.0, 4.0). For the physical index, we compared the difference in signal values between the sclerotic and background normal bone. We evaluated the preference using Scheffé's paired comparison methods for the visual index. For the physical index, RF-C(4.0) had the most significant signal value difference for all 3 bone stiffness images. For the visual index, RF-C(4.0) showed the highest preference. Using frequency-enhanced processing, RF-C(4.0) was suggested to improve the visibility of the osteosclerosis image.