Motion Correction and Parameter Estimation in dceMRI Sequences: Application to Colorectal Cancer

Abstract

Abstract. We present a novel Bayesian framework for non-rigid motion correction and pharmacokinetic parameter estimation in dceMRI sequences which incorporates a physiological image formation model into the similarity measure used for motion correction. The similarity measure is based on the maximization of the joint posterior probability of the transformations which need to be applied to each image in the dataset to bring all images into alignment, and the physiological parameters which best explain the data. The deformation framework used to deform each image is based on the diffeomorphic logDemons algorithm. We then use this method to co-register images from simulated and real dceMRI datasets and show that the method leads to an improvement in the estimation of physiological parameters as well as improved alignment of the images.

A comparison of Bayesian and non-linear regression methods for robust estimation of pharmacokinetics in DCE-MRI and how it affects cancer diagnosis

“…Bayesian inference for nonlinear model algorithms have been proposed before in the literature [10,11,12] to estimate unbiased quantitative tracer kinetic parameters. The proposed scheme is similar to the one suggested by Schmid et al [10], the main difference is on the estimation of the onset time.…”

“…Bayesian algorithms can model the noise of the measured concentration of the contrast agent and have a theoretical guarantee to converge if run long enough [13]. This work suggests a Bayesian inference for nonlinear model algorithm similar to the ones proposed by other groups [10,11,12] and evaluates its robustness and diagnostic value against the LevenbergMarquardt and the simplex algorithms on two separate cohorts of patients: i) a cohort of 76 men, 20 of whom had significant prostate cancer in the peripheral zone ii) a cohort of 9 healthy volunteers and 24 patients with squamous cell carcinoma.…”

“…The suggested Bayesian algorithm similar to [10,11,12] maximizes the posterior probability distribution function p(k,σ|CTIC) as a function of k and σ…”

“…To overcome these issues, Bayesian inference for nonlinear models were suggested [10,11,12]. Bayesian algorithms can model the noise of the measured concentration of the contrast agent and have a theoretical guarantee to converge if run long enough [13].…”

Registration of Free-Breathing 3D+t Abdominal Perfusion CT Images via Co-segmentation

“…Standard choices such as mutual information are not effective in DCE-CT sequences [5]. In [6], Bhushan et al used the pharmacokinetic model fitting error as registration criterion for DCE-MR, thus coupling the two tasks of sequence stabilisation and parameter estimation. However, the latter is a highly non-convex problem: including additional unknowns (namely the pharmacokinetic parameters) increases the risk of falling in local minima.…”