Quantification of missing and overlapping data in multiple breath hold abdominal imaging

Abstract

Magnetic resonance imaging (MRI) of the abdomen is often performed in multiple breath holds which are designed to contiguously cover the region of interest. This technique may result in a failure to image all the appropriate area, and the extent of this failure is difficult to appreciate on a set of 2D slices. With reference to three patient cases, we present a method to quantify the extent of this problem and suggest a solution. First, we manually delineate the region of interest on a single breath hold fast spoiled gradient echo (FSPGR) sequence. Subsequently, we align images acquired in separate breath holds to this reference volume. A coloured 3D presentation makes the extent of unimaged and repeatedly imaged areas clearly visible to the clinician. The alignment also helps radiologists to accurately determine the location of individual slices. The described method can easily be automated and is ideally implemented at the scanner console, ensuring the availability of contiguously sampled datasets to radiologists with minimum user interaction from the radiographer. Such datasets enable the deployment of robust 3D analysis algorithms.