Is there agreement on what makes a good ultrasound image?

Abstract

With the introduction of national guidelines for ultrasound screening it might be assumed that there is agreement in the key features required in assessing image quality. The aim of this study was to determine whether it is possible that personal preferences may influence what is regarded as an acceptable image. Two image sets, one vascular and one obstetric, were taken to the British Medical Ultrasound Society ‘Ultrasound 2012’ conference for delegates to rate whether images were acceptable or not. Data collection took place on days one and two, with the results presented on day three of the conference. Images from a variety of ultrasound machines were used and ranked in order of the odds of producing an acceptable image using logistic regression. Agreement between observers was investigated and three images per person were repeated to look at agreement within observers. Audience feedback was used to record the reasons why images were regarded as acceptable or not. Eighty-two participants reviewed each of the two image sets. The machine rankings revealed that some machines were up to 12 times more likely to produce an acceptable image than other machines. Agreement amongst experts or non-experts was found, but disagreement between the subgroups of experts and non-experts. Agreement within observers was around 80% and similar results were found in each of the image sets. Despite image quality assessment, personal preferences and expertise may still affect judgement, and guidelines may not ensure agreement.