Convex-Optimization-Based Compartmental Pharmacokinetic Analysis for Prostate Tumor Characterization Using DCE-MRI.

Area of Science:

• Medical Imaging
• Biophysics
• Computational Biology

Background:

• Dynamic contrast-enhanced MRI (DCE-MRI) is crucial for pharmacokinetic analysis in suspected cancer tissues.
• Prostate cancer diagnosis and prognosis rely on accurate pharmacokinetic (PK) analysis, including time activity curves (TACs) and kinetic parameters (KPs).

Purpose of the Study:

• To develop a novel blind source separation algorithm, the convex-optimization-based KPs estimation (COKE) algorithm.
• To enhance prostate tumor detection and quantification using DCE-MRI data through improved PK analysis.

Main Methods:

• The COKE algorithm employs compartmental modeling for DCE-MRI data.
• It identifies representative pixels for plasma, fast-flow, and slow-flow TACs.
• The algorithm reformulates the non-convex flux rate constants (FRCs) estimation into two convex optimization problems and uses pixel-wise constrained curve-fitting for KPs estimation.

Main Results:

• The COKE algorithm reliably estimates FRCs by exploiting matrix structures.
• Accurate FRCs estimation leads to effective KPs estimation for cancer and normal tissue mapping.
• Simulation and patient data evaluations show the COKE algorithm's efficacy and consistency with clinical observations.

Conclusions:

• The COKE algorithm provides an effective method for PK analysis in prostate cancer DCE-MRI.
• This approach enhances the accuracy of kinetic parameter estimation for improved tumor detection and quantification.
• The developed algorithm shows promise for clinical application in prostate cancer management.