Statistical Shape Modeling of the Temporomandibular Joint
Disorders of the temporomandibular joint (TMJ) are widespread, afflicting between 5 and 40 percent of the adult population. TMJ disorder (TMJD) encompasses several conditions including orofacial pain, restricted mandibular movement, clicking and popping sounds from the jaw joint, and locked jaws. About 10 to 15 percent of TMJD patients have osteoarthritis, characterized by a degenerative joint which results from erosion of articular cartilage and degeneration of subchondral cortical and trabecular bone. Symptoms can become chronic and difficult to manage, severely reducing quality of life. However, the exact etiology for TMJD has not been fully understood. While TMJD is reported in virtually all populations and age groups, many studies have consistently indicated that TMJD is more prevalent for women than men. While the etiology of TMJD is clearly multifactorial, it is widely believed that local joint tissue level biomechanics resulting from daily functional mastication or other jaw motion plays a major role in the development and progression of TMJD. Several studies have investigated the correlation of TMJ anatomy with TMJD, revealing associations between TMJD and occlusal curvature, dental occlusion, and articular eminence inclination. However, significant controversy exists over these relationships, since other studies did not find correlations between anatomy and TMJD. We hypothesize that differences in TMJ anatomy between males and females could result in different biomechanics that leads to a higher prevalence of the disease in females. Furthermore, while individual anatomical measures may not show correlations with TMJD or sex, combinations of anatomical traits represented using a statistical shape and trait model (SSTM) may show correlations. Thus, the objective of this study was to elucidate sex differences in the anatomy of human temporomandibular joint mandibular condyles using an SSTM.
Mandibles were obtained from 16 human cadavers (9 males, 7 females, 79±13 years). The condyles were dissected at the point of the sigmoid notch concavity and scanned using micro-computed tomography with 27 micron resolution. An image processing algorithm was used to segment the bone and determine the border of the entire mandibular condyle and trabecular bone compartments. Triangulated meshes of the compartments were created. One subject was chosen as the template and was registered to the other individuals using a coherence point drift algorithm. This process positioned all vertices at corresponding anatomic locations. For the trabecular bone region, around each vertex position, the average bone image intensity, which is proportional to bone density, and microstructural traits, including trabecular bone volume fraction, thickness, separation, connectivity, and connectivity density were calculated. For the entire mandibular condyle mesh, the surface vertices were extracted to represent the overall anatomy of the condyle. Using an SSTM, the shape and trait information was reduced to a small set of independent and uncorrelated variables for each individual. Wilcoxon rank sum tests were used to test for differences in the variables between sexes. A lasso approach was used to determine a set of variables that differentiate between sexes.
Male condyles were on average larger than female condyles, with complex differences in the microstructural traits. Two out of 15 principal components were statistically different between males and females (p<0.1). The lasso approach determined a set of seven principal components that fully described the complex shape and trait differences between males and females. An SSTM was able to determine sex-dependent differences in the shape of the mandibular condyle. These differences may alter the biomechanics of the joint and contribute to developing temporomandibular joint disease.