IADR Abstract Archives
Textural Biomarkers of the Temporomandibular Joint Subchondral Bone
Objectives: Osteoarthritis (OA) appears in 42.6% of patients with TMJ disorders, but its early diagnosis remains controversial. The purpose of this study is to demonstrate that it is possible to compute textural biomarkers of subchondral bone as early markers of OA in hr-CBCT, which is an image modality that can be used in-vivo patients. For that purpose we compared textural biomarkers computed in hr-CBCT to those computed in µCT scans as our Gold Standard.
Methods: We used 16 specimens of condylar bones obtained from condylectomies that were scanned using µCT and hr-CBCT. First, 3D surface models for all samples were computed. Then, models and scans for each sample were superimposed. Nine 3D textural biomarkers were then computed in the volume of interest defined by the condylar head: (i) Four co-occurrence features: Entropy, Energy, Contrast and Homogeneity; and (ii) Five Run-length features: short run emphasis (SRE), long run emphasis (LRE), grey-level non-uniformity (GLN), run length non-uniformity (RLN) and run percentage (RP). Finally, Pearson correlation coefficients between the textural values in µCT and hr-CBCT were calculated.
Results: Positive correlations between µCT and hr-CBCT for Energy and Contrast features were found. This indicates that the uniform lattice-shaped spicules that form the subchondral bone (Energy) and also the differences between tissue signal intensities (Contrast) can be measured reliably in hr-CBCT. Positive correlations between µCT and hr-CBCT computed textural biomarkers for GLN and RLN were also found indicating that the gray-level distributions and size (GLN and RLN respectively) can be measured reliably in hr-CBCT.
Conclusions: Subchondral bone structure is preserved in hr-CBCT and can be reliably represented by 3D textural biomarkers. The results suggest that 3D texture analysis has great potential to provide quantitative information on the spatial distribution of subchondral bone internal architecture that is not captured by assessment of bone density or volume.
Methods: We used 16 specimens of condylar bones obtained from condylectomies that were scanned using µCT and hr-CBCT. First, 3D surface models for all samples were computed. Then, models and scans for each sample were superimposed. Nine 3D textural biomarkers were then computed in the volume of interest defined by the condylar head: (i) Four co-occurrence features: Entropy, Energy, Contrast and Homogeneity; and (ii) Five Run-length features: short run emphasis (SRE), long run emphasis (LRE), grey-level non-uniformity (GLN), run length non-uniformity (RLN) and run percentage (RP). Finally, Pearson correlation coefficients between the textural values in µCT and hr-CBCT were calculated.
Results: Positive correlations between µCT and hr-CBCT for Energy and Contrast features were found. This indicates that the uniform lattice-shaped spicules that form the subchondral bone (Energy) and also the differences between tissue signal intensities (Contrast) can be measured reliably in hr-CBCT. Positive correlations between µCT and hr-CBCT computed textural biomarkers for GLN and RLN were also found indicating that the gray-level distributions and size (GLN and RLN respectively) can be measured reliably in hr-CBCT.
Conclusions: Subchondral bone structure is preserved in hr-CBCT and can be reliably represented by 3D textural biomarkers. The results suggest that 3D texture analysis has great potential to provide quantitative information on the spatial distribution of subchondral bone internal architecture that is not captured by assessment of bone density or volume.