The Performance of Artificial Intelligence in Classifying Molecular Markers in Adult-Type Gliomas Using...

Area of Science:

• Neuro-oncology
• Computational pathology
• Medical artificial intelligence

Background:

• Adult-type gliomas are aggressive primary brain tumors requiring accurate molecular classification for prognosis.
• Isocitrate dehydrogenase (IDH) mutations and 1p/19q codeletions are key diagnostic and prognostic markers.
• Histopathological image analysis using artificial intelligence (AI) offers a novel approach to molecular classification.

Purpose of the Study:

• To systematically evaluate the performance of AI models in detecting IDH mutation status and 1p/19q codeletion in adult-type gliomas.
• To assess AI's diagnostic accuracy using histopathological images for these critical molecular markers.

Main Methods:

• A systematic review adhering to PRISMA-DTA guidelines was performed.
• Searched seven databases for studies from 2015-2025 using AI on histopathology images for glioma molecular classification.
• Data extraction and risk of bias assessment were conducted by two independent reviewers.

Main Results:

• Twenty-two studies involving 2453 reports met inclusion criteria, with pooled average accuracy of 85.46%.
• Hybrid AI models achieved the highest performance (accuracy 92.80%).
• AI models showed better performance for IDH mutations than 1p/19q codeletions.

Conclusions:

• AI models demonstrate significant potential as adjunct tools for molecular classification of adult-type gliomas via histopathology.
• Further research with larger datasets and multimodal AI is recommended for broader applicability.
• Integration of AI diagnostic support with clinical expertise is crucial.