Assessment of Inter-Expert Variability and of an Automated Segmentation Method of 40 and 60 MHz IVUS Images of Coronary Arteries

Abstract

The objectives were to compare the performance of a segmentation algorithm, based on the minimization of an uncertainty function, to delineate contours of external elastic membrane and lumen of human coronary arteries imaged with 40 and 60 MHz IVUS, and to use values of this function to delineate portions of contours with highest uncertainty. For 8 patients, 40 and 60 MHz IVUS coronary data acquired pre- and post-interventions were used, for a total of 68,516 images. Manual segmentations of contours (on 2312 images) performed by experts at three core laboratories were the gold-standards. Inter-expert variability was highest on contour points with largest values of the uncertainty function (p < 0.001). Inter-expert variability was lower at 60 than 40 MHz for external elastic membrane (p = 0.013) and lumen (p = 0.024). Average differences in plaque (and atheroma) burden between algorithmic contours and experts’ contours were within inter-expert variability (p < 0.001).

Volume electron microscopy

“…Image-to-image and multimodal registrations should be published and/or publicly deposited with accompanying landmark and transformation files for reproducing such registrations. Segmentation is still largely a manual process, despite progress in the development of machine learning algorithms, and known inter-expert variations on segmentation tasks raise questions around the fidelity of ground truth 274 , 275 . Inter-expert variation is often due to inadequacies in the datasets; for example, sections may be too thick to resolve very thin axons, or to identify synapses that are sectioned parallel to the postsynaptic density, and missing sections and specimen contamination contributes to errors.…”

Diagnostic Performance of 60 MHz High-Definition Intravascular Ultrasound versus Fourier Domain Optical Coherence Tomography for Identifying Plaque Rupture, Plaque Erosion, and Thrombosis in a Rabbit Model

“…Compared with conventional IVUS studies, IVUS of the coronary arteries might be preferably imaged at 60 MHz than at 40 MHz [ 32 ]. In a previous study comparing IVUS and OCT, 20 MHz IVUS could only identify 23.6% of OCT-defined PR [ 33 ].…”