Learning Cephalometric Landmarks for Diagnostic Features Using Regression Trees

Assessing the Co-relation between Mandibular Flare and Thickness of Lingual Cortex in Relation to the Third Molars: A Retrospective Cross-sectional Study

Aim

To assess the co-relation between mandibular flare and thickness of lingual cortex in relation to the third molars.

Materials and Methods

Retrospectively obtained computed tomography (CT) data of 26 patients was used after classifying them into respective skeletal malocclusion groups (classes I, II, and III). Thickness of lingual cortex was measured at crestal, middle, and apical levels in mandibular third molar region. Two angular and two linear measurements were used to measure mandibular flare. Angular measurements included the angle between condylion (Co) and menton (Me), and between gonion (Go) and menton (Me). Linear measurements included bigonial and bicondylion widths.

Results

The two angular measurements did not differ significantly among the three skeletal malocclusion groups. Contrastingly, bicondylion width differed significantly among the three groups. Class II group showed no significant correlation between mandibular flare and lingual cortical thickness. Class III group demonstrated a significant negative correlation of linear and angular measurements with cortical bone thickness. Bicondylion width was significantly more in Class III group than in other skeletal groups, which proved an increased mandibular flare in patients with Class III malocclusion.

Conclusion

Increase in mandibular flare was associated with decreased thickness of lingual cortical bone.

Landmark annotation through feature combinations: a comparative study on cephalometric images with in-depth analysis of model's explainability

OBJECTIVES

The objectives of this study are to explore and evaluate the automation of anatomical landmark localization in cephalometric images using machine learning techniques, with a focus on feature extraction and combinations, contextual analysis, and model interpretability through Shapley Additive exPlanations (SHAP) values.

METHODS

We conducted extensive experimentation on a private dataset of 300 lateral cephalograms to thoroughly study the annotation results obtained using pixel feature descriptors including raw pixel, gradient magnitude, gradient direction, and histogram-oriented gradient (HOG) values. The study includes evaluation and comparison of these feature descriptions calculated at different contexts namely local, pyramid, and global. The feature descriptor obtained using individual combinations is used to discern between landmark and nonlandmark pixels using classification method. Additionally, this study addresses the opacity of LGBM ensemble tree models across landmarks, introducing SHAP values to enhance interpretability.

RESULTS

The performance of feature combinations was assessed using metrics like mean radial error, standard deviation, success detection rate (SDR) (2 mm), and test time. Remarkably, among all the combinations explored, both the HOG and gradient direction operations demonstrated significant performance across all context combinations. At the contextual level, the global texture outperformed the others, although it came with the trade-off of increased test time. The HOG in the local context emerged as the top performer with an SDR of 75.84% compared to others.

CONCLUSIONS

The presented analysis enhances the understanding of the significance of different features and their combinations in the realm of landmark annotation but also paves the way for further exploration of landmark-specific feature combination methods, facilitated by explainability.