USING SIMULATED RADIOGRAPHIC IMAGES TO TEST THE ACCURACY OF AN UPPER CERVICAL X-RAY ANALYSIS SYSTEM.
Edward F. Owens, MS, DC, Life College, Marietta, GA
The value of X-ray analysis for producing listings for chiropractic adjustment has been questioned frequently in the literature. The reliability of x-ray marking systems for the cervical spine has been tested several times with generally promising results. Yet to be established is how well any upper cervical x-ray analysis predicts the actual positions of the skull and vertebrae with respect to each other. Grostic (1) used a model consisting of a dry skull with a movable atlas mounted beneath it to test the accuracy of the atlas laterality measure. The preliminary results showed very good agreement between actual and measured values. Grostic suggested the need for further studies to test the accuracy of rotational misalignments as well as the interrelationship between atlas rotation and laterality. The goal of this paper is to test the accuracy of the Grostic upper cervical analysis by comparing analysis findings to known misalignments displayed on simulated radiographs generated by a computer model.
METHODS Simulated radiographs have been generated using 3D Studio rendering and animation software (Autodesk, Inc.) on an MS-DOS compatible computer. The rendering software is first used to position a three-dimensional digitized model of the human skull and cervical spine. The model can be viewed from different viewpoints, and rendered with a semitransparent material to produce a fair likeness of a cervical radiograph. While the modeled bones have no internal trabecular structure, the surface models do well represent many of the commonly used reference points for cervical x-ray analysis.
In positioning the model, individual bones can be precisely rotated and translated to simulate misalignments of the atlas and axis. Renderings then represent radiographs of known misalignments. The rendering can be viewed on screen or printed on paper using a laser printer. Each rendering is generated at 2560x1920 pixels and takes two and one half hours to render (using a 486 DX2 running at 66 mHz).
For this study, five sets of simulated radiographs were generated, including Nasium, Vertex and Lateral views. These were sent to three chiropractors in practice locally for analysis. The simulated radiographs were coded so that none of the investigators knew the actual misalignments present.
RESULTS At the time of this writing, the results of the x-ray analyses are not yet available. The analysis results will be statistically evaluated for inter-examiner reliability and then compared to the known values to determine accuracy.
(1) Grostic JD, Marshall WG. Accuracy of an upper cervical measuring system: A validity study. Proceeding 1992 ICSM, Chicago, IL, pg 146-7.