Fast spectroscopic multiple analysis (FASMA) for brain tumor classification: a clinical decision support system...

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Clinical decision support systems (CDSS) can enhance brain tumor diagnosis and grading.
• A novel Fast Spectroscopic Multiple Analysis (FASMA) system was developed for brain tumor classification using multiparametric MRI data.

Purpose of the Study:

• To evaluate the efficacy of the FASMA system as a CDSS for brain tumor classification.
• To assess the performance of machine learning methods in grading gliomas and differentiating tumor infiltration patterns.

Main Methods:

• Acquired multiparametric MRI data (metabolic ratios, spectra, diffusion, perfusion) from 126 patients with intracranial tumors.
• Utilized machine learning algorithms, including support vector machines (SVM), to classify tumor types and grades.
• Integrated additional databases with MR parameters and tumor characteristics, developing a custom GUI for user-friendly classification.

Main Results:

• The SVM model combining all MR features achieved the highest classification performance.
• FASMA demonstrated high accuracy, correctly classifying 89% of intratumoral and 79% of peritumoral areas in an independent test set.
• The system could differentiate between infiltrative and non-infiltrative lesions, even in misclassified cases.

Conclusions:

• FASMA is a prototype CDSS integrating quantitative MR data for brain tumor characterization.
• The system serves as a diagnostic assistant, offering rapid analysis and classification of MR parameters.
• FASMA can function as a teaching tool for advanced MRI techniques, incorporating literature-derived tumor characteristics.