Current techniques for determining the existence of pathologic morphology in the TMJ, particularly disc derangement disorders, include conventional radiographs, arthrography, and CT. Arthrography is considered by most clinicians to provide the best diagnostic information on soft tissue pathology and has shown high correlation with surgical findings. However, it is painful, invasive and uses moderate levels of ionizing radiation. MRI is an alternative technique which is non-invasive and has no known adverse health effects. It displays the anatomy of most body tissues, but in contrast to x-ray techniques, is particularly effective in imaging soft tissue structures. Further, it is possible that variations in MR signals may be associated with differences in tissue characteristics. We seek to test the hypothesis that MRI is at least as accurate as arthrography in identification of disc derangement. An observer performance study will be conducted utilizing 2 panel of expert readers to independently interpret arthrographic and MR images derived from a population of patients with symptoms of disc derangement. Observer diagnoses will be compared to direct verification in patients who proceed to surgery. To avoid the effects of sample bias on observer performance, data sets will consist of roughly equal proportions of 'normal' and abnormal cases. This investigation will provide quantitative evidence of comparative accuracy, sensitivity, specificity, and inter and intra observer reliability for each method. Further analysis of responses on a range of diagnostic features will identify the specific information which is uniquely or jointly derived from each technique. The effect of MR imaging parameters on observer detection and characterization of TMJ pathology will be studied in equal subsets of normal and abnormal subjects who will be scanned using 4 different sets of pulse sequences. Randomly sequenced images will be read twice by the MRI panel. Accuracy and inter and intra reader reliability will be compared across imaging protocols. Finally, the relationship between MR signal characteristics and histopathology will be studied in a sample of excised TMJ tissue specimens derived from the surgery patients. Tissue relaxation times (T1 and T2) will be correlated with the type and degree of pathologic change observed via histologic examination of the tissue. The results will determine if relaxation time can be used to infer information about meniscal pathology.
