Deep learning characterization of brain tumours with diffusion weighted imaging.

Area of Science:

• Computational biology
• Medical imaging
• Artificial intelligence

Background:

• Glioblastoma multiforme (GBM) is a highly aggressive brain cancer.
• Accurate characterization of GBM is crucial for predicting progression and treatment response.
• Existing proliferation-invasion (PI) models require patient-specific parameters that are difficult to estimate.

Purpose of the Study:

• To develop and apply a deep learning model for estimating key GBM proliferation-invasion parameters.
• To predict glioblastoma tumor progression curves using patient-specific data.
• To address the challenge of personalized GBM forecasting.

Main Methods:

• A deep learning model was developed to estimate tumor cell diffusivity and proliferation rate.
• Multi-sequence MRI data was utilized, including brain tumor segmentation and conversion to tumor cellularity.
• The model was validated on synthetic tumors and applied to a clinical dataset of five GBM patients.

Main Results:

• The deep learning model accurately estimated PI model parameters for GBM.
• The model generated predictions of tumor progression curves with associated parameter uncertainties.
• Evidence-based, patient-specific parameter estimates were derived for all clinical cases.

Conclusions:

• The developed deep learning method offers a novel and generalizable approach for estimating patient-specific GBM parameters.
• This method can enhance the accuracy of personalized glioblastoma treatment planning.
• Improved parameter estimation facilitates better prediction of tumor behavior and therapeutic outcomes.