Measurement Process: MR Data Collection and Image Analysis

The process of MRI measurement is analysed in some detail; it consists of two distinct processes: MR data collection and image analysis. Data collection requires a static field B₀, static field gradients that can be switched very quickly; a radiofrequency transmit field B₁ ⁺; and a receive sensitivity field B₁ ^-. B₁ ⁺ mapping methods are summarised. B₁ ^- could in the past be measured using the reciprocity principle; on modern MRI machines, this is mostly invalid; this has implications for the measurement of proton density and spectroscopic metabolite concentrations. Image noise and scanner stability are considered. Image analysis can be carried out using regions of interest (ROI’s), histograms or voxel-based group mapping. Methods to optimism reproducibility of ROI’s and histograms are presented. Statistical analysis is considered, with group comparison using t-tests, positive and negative results, correlation, clinical scores, classification and ROC curves analysed. Multi-parametric measurements are likely to give better indications of biological status. In this chapter, a table of recommended statistics books is given, and there are nearly 100 references to methods of measurement.