5 Key Takeaways

• 1

  AI models may enable earlier noninvasive identification of pathogenic variants in CNS tumors.

• 2

  The study developed two machine-learning models using mutation data and MRI images from CNS tumor patients.

• 3

  The models classified samples into glial tumors, nonglial CNS tumors, and normal tissue with strong reliability.

• 4

  The dense neural network achieved a Matthews Correlation Coefficient of 0.8822, while the MRI model achieved 0.8525.

• 5

  Combining model outputs improved performance in identifying pathogenic variants and tumor types.