Niloufar Zarin, PhD, FHEA
LATEST PROJECTS
Project Area 1: Feasibility Assessment of High Resolution Myocardial Perfusion Quantification
The use of fully quantitative analysis of myocardial first-pass contrast-enhanced MRI allows the absolute quantification of Myocardial Blood Flow (MBF) and may permit more accurate and objective assessment of altered myocardial perfusion in patients with heart disease.
High resolution (voxel-wise) quantitative analysis has demonstrated the potential advantage of an early identification of ischemia arising from the subendocardial layers.
The purpose of this study was to enable high spatial resolution voxel-wise quantitative analysis of myocardial perfusion in dynamic contrast-enhanced cardiovascular MR, in particular by finding the most favorable quantification algorithm in this context. Four deconvolution algorithms—Fermi function modeling, deconvolution using B-spline basis, deconvolution using exponential basis, and autoregressive moving average modeling —were tested to calculate voxel-wise perfusion estimates. The algorithms were developed on synthetic data and validated against a true goldstandard using a hardware perfusion phantom.
The accuracy of each method was assessed for different levels of spatial averaging and perfusion rate. Finally, voxel-wise analysis was used to generate high resolution perfusion maps on real data acquired from patients with suspected coronary artery disease and healthy volunteers.
Project Area 2: The relationship between contrast enhancement, spatial resolution level and quatification accuracy
Quantitative assessment of perfusion CMR has never been hassle free. Several pitfalls can affect the accuracy of MBF quantification. Among these, contrast agent dispersion and computational errors in the fitting procedure required to calculate MBF estimates play a major role in defining the accuracy of the MBF measurements.
In this project, I developed an automated method for detection of contrast agent arrival time (tOnset). I demonstrated that measurements of the time delay between arterial and myocardial contrast enhancement play an important role in characterizing the collateral circulation and that the delayed arrival of a tracer can cause significant errors in the quantification of myocardial blood especially in collateral dependent regions and when high resolution MBF quantification is used.
Moreover, I used a new method to measure the quality of the fit and the ratio of the modelling error that occurs as a result of low SNR or reduced spatial resolution. These ratios, which are called fraction of residual information (FRI) and fraction of Modelled Information (FMI) allow extracting the modelling error from the residuals and is an estimate of the amount un-modelled information remaining in the residuals. I have demonstrated that decreasing the resolution level to improve the SNR of the data will result in losing important physiological information which can be possibly used for arriving at a clinical diagnosis.
Project Area 3: Modelling Parameter Role on Accuracy of Cardiac Perfusion Quantification
The goal of perfusion-CMR post- processing is to recover tissue impulse-response from observed signal intensity curves. While several deconvolution techniques are available for this purpose, all of them use models with varying parameters for the representation of the impulse-response. However this variation influences the accuracy of the deconvolution and introduces possible variations in the results. Using an appropriate order for quantification is essential to allow CMR-perfusion-quantification to develop into a useful clinical tool. The aim of this study was to evaluate the effect of parameter Variation in Fermi modelling, autoregressive moving-average model (ARMA), B-spline-basis and exponential-basis deconvolution.
Project Area 4: Multiclass imbalance learning: Improving classification ofPaediatric brain tumours from magnetic resonance spectroscopy
Classification of paediatric brain tumours from 1H Magnetic Resonance Spectroscopy (MRS) can aid diagnosis and management of brain tumours. However, varied incidence of the different tumour types leads to imbalanced class sizes and introduces difficulties in classifying rare tumour groups. This study assesses different imbalanced-multi-class learning techniques and compares the use of complete spectra and quantified metabolite-profiles for classification of three childhood brain tumour types.