Comparison of Resampling Methods and Radiomic Machine Learning Classifiers for Predicting Bone Quality Using Dual-Energy X-Ray Absorptiometry

Abstract

Background/Objectives: This study presents a novel approach, based on a combination of radiomic feature extraction, data resampling techniques, and machine learning algorithms, for the detection of degraded bone structures in Dual X-ray Absorptiometry (DXA) images. This comprehensive approach, which addresses the critical aspects of the problem, distinguishes this work from previous studies, improving the performance achieved by the most similar studies. The primary aim is to provide clinicians with an accessible tool for quality bone assessment, which is currently limited. Methods: A dataset of 1531 spine DXA images was automatically segmented and labelled based on Trabecular Bone Score (TBS) values. Radiomic features were extracted using Pyradiomics, and various resampling techniques were employed to address class imbalance. Three machine learning classifiers (Logistic Regression, Support Vector Machine (SVM), and XGBoost) were trained and evaluated using standard performance metrics. Results: The SVM classifier outperformed the other classifiers. The highest F-score of 97.5% was achieved using the Grey Level Dependence Matrix and Grey Level Run Length Matrix feature combination with SMOTEENN resampling, which proved to be the most effective resampling technique, while the undersampling method yielded the lowest performance. Conclusions: This research demonstrates the potential of radiomic texture features, resampling techniques, and machine learning methods for classifying DXA images into healthy or degraded bone structures, which potentially leads to improved clinical diagnosis and treatment.

Keywords: data resampling; dual energy X-ray absorptiometry; machine learning; radiomics; trabecular bone score.

Figure 1

The left panel displays a DXA image with ROIs defined by the equipment’s software, and the right panel shows the corresponding automatically generated mask. In the left figure, the lateral contour is delineated by black pixels, while white lines individually divide the four vertebrae of interest.

Figure 2

Resampled Class Distribution in the WFS Dataset. Each bar represents a resampling technique, showing the final number of samples for each class. The INITIAL bar indicates the original, imbalanced distribution.

Figure 3

Application of five resampling techniques to each radiomic dataset results in 45 new, more balanced datasets.

Figure 4

Work pipeline: (1) Image acquisition and labelling as healthy or degraded; (2) Image segmentation and binary mask generation; (3) Feature extraction using Pyradiomics; (4) Resampling techniques application; (5) ML classification using the proposed classifiers.

Figure 5

Classification results. The X-axis represents the dataset used for model training and testing, while the colour indicates the resampling technique. The symbol shape represents the ML model employed. The Y-axis displays the F-score value obtained in each case.

Figure 6

F-score comparison of SVM classifier performance on the three combined feature sets (WFS, GLDM-GLRLM-DS and GLDM-GLSZM-GLRLM-DS) using five different resampling techniques: UNDERSAMPLING, SMOTE, ADASYN, SMOTEENN and SMOTETomek.