Improving apparent diffusion coefficient estimates and elucidating tumor heterogeneity using Bayesian adaptive...

Area of Science:

• Biomedical Imaging
• Radiology
• Computational Biology

Background:

• Tumor apparent diffusion coefficient (ADC) estimates show spatial heterogeneity due to true biological variation and measurement error.
• Accurate ADC mapping is crucial for cancer diagnosis, staging, and treatment monitoring.

Purpose of the Study:

• To develop and validate an adaptive Bayesian adaptive smoothing (BAS) model for improved ADC estimation in tumors.
• To assess the performance of the BAS model against a diffusion data gold standard and simulations.

Main Methods:

• An adaptive Bayesian adaptive smoothing (BAS) model incorporating a Markov random field was developed for ADC estimation.
• The BAS model was applied to in vivo diffusion MRI data from two murine tumor models.
• Performance was evaluated against diffusion data acquired with four averages (empirical gold standard) and computational simulations.

Main Results:

• BAS-derived ADC estimates showed significantly closer agreement with the gold standard data.
• Analysis of uncertainty estimates indicated that the BAS model potentially outperformed the gold standard.
• A novel measure of tumor ADC heterogeneity revealed differences between cell lines, correlating with histological variations.

Conclusions:

• The adaptive BAS postprocessing technique improves ADC estimation accuracy in tumors.
• BAS offers a method to enhance ADC estimates without compromising spatial resolution, signal-to-noise ratio, or requiring hardware modifications.
• The developed heterogeneity measure can distinguish between tumor types, reflecting underlying microenvironmental differences.