Exploring the Potential of Machine Learning Algorithms to Improve Diffusion Nuclear Magnetic Resonance Imaging Models Analysis

Abstract

Purpose: This paper explores different machine learning (ML) algorithms for analyzing diffusion nuclear magnetic resonance imaging (dMRI) models when analytical fitting shows restrictions. It reviews various ML techniques for dMRI analysis and evaluates their performance on different b-values range datasets, comparing them with analytical methods.

Materials and methods: After standard fitting for reference, four sets of diffusion-weighted nuclear magnetic resonance images were used to train/test various ML algorithms for prediction of diffusion coefficient (D), pseudo-diffusion coefficient (D*), perfusion fraction (f), and kurtosis (K). ML classification algorithms, including extra-tree classifier (ETC), logistic regression, C-support vector, extra-gradient boost, and multilayer perceptron (MLP), were used to determine the existence of diffusion parameters (D, D*, f, and K) within single voxels. Regression algorithms, including linear regression, polynomial regression, ridge, lasso, random forest (RF), elastic-net, and support-vector machines, were used to estimate the value of the diffusion parameters. Performance was evaluated using accuracy (ACC), area under the curve (AUC) tests, and cross-validation root mean square error (RMSE_CV). Computational timing was also assessed.

Results: ETC and MLP were the best classifiers, with 94.1% and 91.7%, respectively, for the ACC test and 98.7% and 96.3% for the AUC test. For parameter estimation, RF algorithm yielded the most accurate results The RMSE_CV percentages were: 8.39% for D, 3.57% for D*, 4.52% for f, and 3.53% for K. After the training phase, the ML methods demonstrated a substantial decrease in computational time, being approximately 232 times faster than the conventional methods.

Conclusions: The findings suggest that ML algorithms can enhance the efficiency of dMRI model analysis and offer new perspectives on the microstructural and functional organization of biological tissues. This paper also discusses the limitations and future directions of ML-based dMRI analysis.

Keywords: Diffusion magnetic resonance imaging; intravoxel incoherent motion; kurtosis; machine learning.

Figure 1

Flowchart for determining each voxel’s characteristic parameters (diffusion, pseudo-diffusion, perfusion fraction, and kurtosis)

Figure 2

A flowchart that outlines the process for determining the fitting for any specific voxel. It provides a step-by-step visual guide, simplifying the understanding of the fitting determination procedure

Figure 3

Illustration of the process for deriving input and output vectors from the r² and modified-r² maps. The modification is implemented through a proposed “smoothing filter,” which enhances the diversity of the data quality. The figure visually depicts the derivation of these vectors, which are subsequently used for training and testing the machine learning models

Figure 4

Comprehensive evaluation of machine learning (ML) models used in this study. The training and testing of the models are conducted using input–output vectors, which are analytically derived beforehand. The effectiveness of the models is demonstrated by comparing the predicted results with the respective ML outcomes

Figure 5

Normalized confusion matrices of the algorithms implemented for classification based on the r² value. Color code is used for faster visual analysis. ETC: Extra-tree classifier, LR: Logistic regression, SVC: C-Support Vector Classification, XGB: Extragradient boost, MLP: Multilayer perceptron, TN: True negative, TP: True positive, FP: False positive, FN: False negative

Figure 6

Receiver operating characteristic curves of the implemented classification algorithms for the classification of r², compared with a random classifier. ETC: Extra-tree classifier, ROC: Receiver operating characteristic, AUC: Area under the curve, LR: Logistic regression, SVC: C-Support Vector Classification, MLP: Multilayer perceptron, XGBL: ExtraGradient Boost

Figure 7

Root mean square error cross-validated scores for every algorithm implemented for the prediction of (a) diffusion coefficient, (b) pseudo-diffusion coefficient, (c) perfusion fraction, and (d) kurtosis. LR: Logistic regression, SVR: Support vector regression, RF: Random forest, XGB: Extragradient boost, EN: ElasticNet, RMSE: Root mean square error