Abstract
The aim of this feasibility study was to assess the impact that image processing of abdominal aortic ultrasound (US) images had on the intra-observer reliability of the diameter measurement. The study compared variability between inner-to-inner (ITI), outer-to-outer (OTO) and outer-to-inner (OTI) wall diameter measurements and their resilience to image processing. Three US images of transverse abdominal aortas were manipulated in 13 different ways using functions from Image J software (National Institutes of Health, Bethesda, MD, USA). Blinded measurements were performed of the aortic diameters from each image; this was repeated for ITI, OTO and OTI. Profiles of each image were produced and sets of rules developed to provide detailed instructions as to where, on the profile, the callipers should be placed to correspond with the actual image. The reliability of the diameter measurements compared to the original diameter measurement was least affected by adjusting the brightness and contrast of the US images (better than ± 1.5 mm). Using the functions 'Sharpen' and 'Find Edges' created the largest difference (up to -5 mm). The ITI measurements had the widest spread of variability, whereas the OTI measurements proved to be the most repeatable and resilient to image processing. This study suggests the precision of the measurements can be kept within satisfactory levels even after image manipulation. It also showed the most reliable measuring guideline was OTI, in contrast to the guideline currently used by the NHS Abdominal Aortic Aneurysm Screening Programme. Further research is needed to transfer the findings into the clinical setting of the National Screening Programme to increase its reliability.