Freehand real-time elastography: impact of scanning parameters on image quality and in vitro intra- and interobserver
Roald Flesland Havre1, Erlend Elde, Odd Helge Gilja
1National Centre for Ultrasound in Gastroenterology, Department of Medicine, Haukeland University Hospital, Bergen, Norway. roald.flesland.havre@helse-bergen.no
Abstract:
Real-time elastography is a method for visualization of the elastic properties of soft tissue and may potentially enable differentiation between malignant and benign pathologic lesions. Our aim was to validate the method on a tissue-mimicking (TM) phantom and to evaluate the influence of different scanning parameters and investigator variability. A TM-phantom containing eight spherical inclusions with known storage modulus was examined using two different transducers on an ultrasound (US) scanner equipped with software for real-time elasticity imaging. The ultrasound transducers were moved vertically in a repetitive manner to induce strain. Two investigators performed series of standardized elastography scans applying a 0-4 categorical quality scale to evaluate the influence of seven parameters: dynamic range of elasticity, region-of-interest, frequency of transducer movement, rejection of elastogram noise, frame rate, persistence and smoothing. Subsequently, repeated examinations of four selected inclusions were performed using a visual analog scale (VAS) where investigators marked a 100 mm horizontal line representing the span in image quality based on experience from the first examination. The hardest and softest inclusions were imaged more clearly than the inclusions with elasticity more similar to the background material. Intraobserver agreement on elastogram quality was good (kappa: 0.67 - 0.75) and interobserver agreement average (kappa: 0.55 - 0.56) when using the categorical scale. The subsequent VAS evaluation gave intraclass-correlation coefficients for the two observers of 0.98 and 0.93, respectively, and an interclass-correlation coefficient of 0.93. Real-time elastography adequately visualized isoechoic inclusions with different elastic properties in a TM-phantom with acceptable intra- and interobserver agreement. Dynamic range of elasticity was the parameter with most impact on the elastographic visualization of inclusions.
