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.

Lateral Cephalometric Landmark Annotation Using Histogram Oriented Gradients Extracted from Region of Interest Patches

Introduction

Two-dimensional cephalometric image analysis plays a crucial role in orthodontic diagnosis and treatment planning. While deep learning-based algorithms have emerged to automate the laborious task of anatomical landmark annotation, their effectiveness is hampered by the challenges of acquiring and labelling clinical data. In this study, we propose a model that leverages conventional machine learning techniques to enhance the accuracy of landmark detection using limited dataset.

Materials and methods

Our methodology involves coarse localization through region of interest (ROI) extraction and fine localization utilizing histogram-oriented gradient (HOG) feature. The image patch containing landmark pixels is classified using the light gradient boosting machine (LGBM) algorithm. To evaluate our model's performance, we conducted rigorous tests on the ISBI Cephalometric dataset and Dental Cepha dataset, aiming to achieve accuracy within a 2 mm radial precision range. We also employed cross-validation to assess our approach, providing a robust evaluation.

Results

Our model's performance on the ISBI Cephalometric dataset showed an accuracy rate of 77.11% within the desired 2 mm radial precision range. The cross-validation results further confirmed the effectiveness of our approach, yielding a mean accuracy of 78.17%. Additionally, we applied our model to the Dental Cepha dataset, where we achieved a remarkable landmark detection accuracy of 84%.

Conclusion

The results demonstrate that traditional machine learning techniques can be effective for accurate landmark detection in cephalometric images, even with limited data. Our findings highlight the potential of these techniques for clinical applications, where large datasets of labelled images may not be available.

Artificial Intelligence for Detecting Cephalometric Landmarks: A Systematic Review and Meta-analysis

Using computer vision through artificial intelligence (AI) is one of the main technological advances in dentistry. However, the existing literature on the practical application of AI for detecting cephalometric landmarks of orthodontic interest in digital images is heterogeneous, and there is no consensus regarding accuracy and precision. Thus, this review evaluated the use of artificial intelligence for detecting cephalometric landmarks in digital imaging examinations and compared it to manual annotation of landmarks. An electronic search was performed in nine databases to find studies that analyzed the detection of cephalometric landmarks in digital imaging examinations with AI and manual landmarking. Two reviewers selected the studies, extracted the data, and assessed the risk of bias using QUADAS-2. Random-effects meta-analyses determined the agreement and precision of AI compared to manual detection at a 95% confidence interval. The electronic search located 7410 studies, of which 40 were included. Only three studies presented a low risk of bias for all domains evaluated. The meta-analysis showed AI agreement rates of 79% (95% CI: 76-82%, I2 = 99%) and 90% (95% CI: 87-92%, I2 = 99%) for the thresholds of 2 and 3 mm, respectively, with a mean divergence of 2.05 (95% CI: 1.41-2.69, I2 = 10%) compared to manual landmarking. The menton cephalometric landmark showed the lowest divergence between both methods (SMD, 1.17; 95% CI, 0.82; 1.53; I2 = 0%). Based on very low certainty of evidence, the application of AI was promising for automatically detecting cephalometric landmarks, but further studies should focus on testing its strength and validity in different samples.

Automatic localization of three-dimensional cephalometric landmarks on CBCT images by extracting symmetry features of the skull

To propose an algorithm for automatic localization of 3D cephalometric landmarks on CBCT data, those are useful for both cephalometric and upper airway volumetric analysis. 20 landmarks were targeted for automatic detection, of which 12 landmarks exist on the mid-sagittal plane. Automatic detection of mid-sagittal plane from the volume is a challenging task. Mid-sagittal plane is detected by extraction of statistical parameters of the symmetrical features of the skull. The mid-sagittal plane is partitioned into four quadrants based on the boundary definitions extracted from the human anatomy. Template matching algorithm is applied on the mid-sagittal plane to identify the region of interest ROI, further the edge features are extracted, to form contours in the individual regions. The landmarks are automatically localized by using the extracted knowledge of anatomical definitions of the landmarks. The overall mean error for detection of 20 landmarks was 1.88 mm with a standard deviation of 1.10 mm. The cephalometric land marks on CBCT data were detected automatically with in the mean error less than 2 mm.

A pilot study for segmentation of pharyngeal and sino-nasal airway subregions by automatic contour initialization

PURPOSE

The objective of the present study is to put forward a novel automatic segmentation algorithm to segment pharyngeal and sino-nasal airway subregions on 3D CBCT imaging datasets.

METHODS

A fully automatic segmentation of sino-nasal and pharyngeal airway subregions was implemented in MATLAB programing environment. The novelty of the algorithm is automatic initialization of contours in upper airway subregions. The algorithm is based on boundary definitions of the human anatomy along with shape constraints with an automatic initialization of contours to develop a complete algorithm which has a potential to enhance utility at clinical level. Post-initialization; five segmentation techniques: Chan-Vese level set (CVL), localized Chan-Vese level set (LCVL), Bhattacharya distance level set (BDL), Grow Cut (GC), and Sparse Field method (SFM) were used to test the robustness of automatic initialization.

RESULTS

Precision and F-score were found to be greater than 80% for all the regions with all five segmentation methods. High precision and low recall were observed with BDL and GC techniques indicating an under segmentation. Low precision and high recall values were observed with CVL and SFM methods indicating an over segmentation. A Larger F-score value was observed with SFM method for all the subregions. Minimum F-score value was observed for naso-ethmoidal and sphenoidal air sinus region, whereas a maximum F-score was observed in maxillary air sinuses region. The contour initialization was more accurate for maxillary air sinuses region in comparison with sphenoidal and naso-ethmoid regions.

CONCLUSION

The overall F-score was found to be greater than 80% for all the airway subregions using five segmentation techniques, indicating accurate contour initialization. Robustness of the algorithm needs to be further tested on severely deformed cases and on cases with different races and ethnicity for it to have global acceptance in Katradental radKatraiology workflow.