Bone X-ray images have been used mainly for the qualitative analysis or the morphological measurement of bones. This paper proposes a method for quantitative characterization of femur X-ray images using a spectrum analysis technique. This method characterizes the trabecular pattern by the spectra (spatial frequency distributions) of trabeculae in the X-ray image. Though some methods have been proposed for the characterization of trabecular patterns, they were for the dried bones taken outside the body, and, therefore, could not easily be applied to the clinical X-ray images. This method deals with the X-ray images of the bones in vivo with a view to clinical use. As a spectrum analysis technique, Maximum Entropy Method (MEM) is used instead of the commonly used FFT technique. First, this method was applied to the X-ray image of the femoral head specimen (dried bone) to verify the effectiveness of this principle. The obtained spectra agreed well with the results of visual observation. Then, using this method, the trabecular patterns in clinical X-ray images were analyzed. Although the femora were in the body, the obtained spectra were found to reflect the characteristics of the trabecular pattern well. Particularly, the positions of spectral peaks stand for the line density of the trabecular pattern. In order to examine the usefulness of this method, it was applied to 11 clinical X-ray images of femoral heads varied from normal to severely deteriorated by osteoporosis. In the comparison between the line density estimated by this method and the diagnosis given by orthopedists (6 doctors), the following results were obtained. With the conventional grading method, large individual variations are inevitable due to subjective observations. And there is a close correlation (r = 0. 86) between the line density estimated by this method and the relative order given by the orthopedists.
|Number of pages||6|
|Journal||Japanese journal of medical electronics and biological engineering|
|Publication status||Published - 1986|
ASJC Scopus subject areas
- Biomedical Engineering