Preclinical Implementation of matRadiomics: A Case Study for Early Malformation Prediction in Zebrafish Model

Area of Science:

• Biomedical Imaging
• Machine Learning
• Radiomics

Background:

• Radiomics offers structured decision support but faces challenges with software interoperability.
• matRadiomics integrates the radiomics workflow into a single platform.

Purpose of the Study:

• Extend matRadiomics to preclinical settings.
• Validate the platform for early malformation differentiation in zebrafish.
• Optimize feature extraction for high-resolution preclinical images.

Main Methods:

• Integrated radiomics workflow using matRadiomics and Pyradiomics.
• Feature extraction, selection, and classification using LDA, KNN, and SVM.
• K-fold cross-validation and ROC curve analysis for performance evaluation.

Main Results:

• Feature extraction time optimization was critical for preclinical images.
• Support Vector Machines (SVM) achieved the highest performance (AUC = 0.723, accuracy = 0.72).
• The optimized plugin effectively predicted early biological outcomes.

Conclusions:

• matRadiomics is versatile and effective for preclinical radiomics.
• The platform aids in early biological outcome prediction.
• Applicable across various biomedical research fields.