Machine Learning Algorithm: Texture Analysis in CNO and Application in Distinguishing CNO and Bone Marrow Growth-Related Changes on Whole-Body MRI

Abstract

Objective: The purpose of this study is to analyze the texture characteristics of chronic non-bacterial osteomyelitis (CNO) bone lesions, identified as areas of altered signal intensity on short tau inversion recovery (STIR) sequences, and to distinguish them from bone marrow growth-related changes through Machine Learning (ML) and Deep Learning (DL) analysis.

Materials and methods: We included a group of 66 patients with confirmed diagnosis of CNO and a group of 28 patients with suspected extra-skeletal systemic disease. All examinations were performed on a 1.5 T MRI scanner. Using the opensource 3D Slicer software version 4.10.2, the ROIs on CNO lesions and on the red bone marrow were sampled. Texture analysis (TA) was carried out using Pyradiomics. We applied an optimization search grid algorithm on nine classic ML classifiers and a Deep Learning (DL) Neural Network (NN). The model's performance was evaluated using Accuracy (ACC), AUC-ROC curves, F1-score, Positive Predictive Value (PPV), Mean Absolute Error (MAE) and Root-Mean-Square Error (RMSE). Furthermore, we used Shapley additive explanations to gain insight into the behavior of the prediction model.

Results: Most predictive characteristics were selected by Boruta algorithm for each combination of ROI sequences for the characterization and classification of the two types of signal hyperintensity. The overall best classification result was obtained by the NN with ACC = 0.91, AUC = 0.93 with 95% CI 0.91-0.94, F1-score = 0.94 and PPV = 93.8%. Between classic ML methods, ensemble learners showed high model performance; specifically, the best-performing classifier was the Stack (ST) with ACC = 0.85, AUC = 0.81 with 95% CI 0.8-0.84, F1-score = 0.9, PPV = 90%.

Conclusions: Our results show the potential of ML methods in discerning edema-like lesions, in particular by distinguishing CNO lesions from hematopoietic bone marrow changes in a pediatric population. The Neural Network showed the overall best results, while a Stacking classifier, based on Gradient Boosting and Random Forest as principal estimators and Logistic Regressor as final estimator, achieved the best results between the other ML methods.

Keywords: bone marrow; children; chronic non-bacterial osteomyelitis; machine learning; texture analysis; whole-body magnetic resonance imaging.

Figure 1

(a) WBMRI showing CNO hyperintense signal of femurs, tibias and a dorsal vertebrae. (b) WBMRI showing red marrow hyperintense signal of femurs, ilium and ischium.

Figure 2

CNO hyperintensity sampling: (a) shows a hyperintense signal in the right clavicle, (b) shows the corresponding ROI drawn manually using the 3D Slicer program on the corresponding hyperintensity.

Figure 3

Red marrow hyperintensity sampling: (a) shows a hyperintense signal in the right femur, (b) shows the corresponding ROI drawn manually using the 3D Slicer program on the corresponding hyperintensity.

Figure 4

Machine Learning classifiers tested in the present study. Non-ensemble learners included KNeighbors, Logistic Regressor and Decision Tree. Ensemble learners included boosting, Stacking and bagging classifiers.

Figure 5

Neural Network structure: input layer with 62 units, first hidden layer with 30 units and L2 regularization, first dropout layer, second hidden layer with 10 units and L2 regularization, second dropout layer, output layer with 1 unit and sigmoid activation.

Figure 6

SHAP reporting of the 20 ranking significative features according to global (a) and local (b) point of view of the method adopted.

Figure 6

SHAP reporting of the 20 ranking significative features according to global (a) and local (b) point of view of the method adopted.

Figure 7

Boxplot comparison between the nine ML classifiers used in terms of (a) F1-score, (b) Accuracy, (c) Positive Predictive Value, (d) ROC/AUC.