DCE-Qnet: deep network quantification of dynamic contrast enhanced (DCE) MRI

Area of Science:

• Medical Imaging
• Artificial Intelligence in Medicine
• Quantitative MRI

Background:

• Dynamic contrast-enhanced MRI (DCE-MRI) offers clinical insights but faces challenges in pharmacokinetic modeling for widespread adoption.
• Accurate quantification of DCE-MRI parameters is crucial for reliable clinical interpretation.

Purpose of the Study:

• To develop and validate a deep learning approach (DCE-Qnet) for comprehensive DCE-MRI quantification.
• To assess the performance of DCE-Qnet compared to conventional methods and evaluate its reproducibility.

Main Methods:

• A 7-layer neural network, DCE-Qnet, was trained on simulated DCE-MRI data using the Extended Tofts model and Parker arterial input function.
• Network training incorporated B1 inhomogeneities to estimate perfusion (Ktrans, vp, ve), T1 relaxation, proton density, and bolus arrival time (BAT).
• Validation involved digital phantom testing against nonlinear least-squares fitting (NLSQ) and in vivo studies in healthy subjects and a cervical cancer patient.

Main Results:

• DCE-Qnet demonstrated superior performance over NLSQ in phantom studies.
• In vivo, inter-subject variability in healthy cervix ranged from 5-51% for different parameters.
• Reproducibility in tumor regions showed coefficients of variation between 1-47%.

Conclusions:

• DCE-Qnet enables comprehensive DCE-MRI quantification from a single acquisition, integrating perfusion, T1 relaxation, and BAT estimation.
• This approach reduces scan time by approximately 10 minutes and enhances quantification accuracy.
• The method shows promise for improved clinical utility in oncology and other applications.