“Global” cardiac atherosclerotic burden assessed by artificial intelligence-based versus manual segmentation in 18F-sodium fluoride PET/CT scans: Head-to-head comparison

Abstract

Background: Artificial intelligence (AI) is known to provide effective means to accelerate and facilitate clinical and research processes. So in this study it was aimed to compare a AI-based method for cardiac segmentation in positron emission tomography/computed tomography (PET/CT) scans with manual segmentation to assess global cardiac atherosclerosis burden.Methods: A trained convolutional neural network (CNN) was used for cardiac segmentation in 18 Fsodium fluoride PET/CT scans of 29 healthy volunteers and 20 angina pectoris patients and compared with manual segmentation. Parameters for segmented volume (Vol) and mean, maximal, and total standardized uptake values (SUVmean, SUVmax, SUVtotal) were analysed by Bland-Altman Limits of Agreement. Repeatability with AI-based assessment of the same scans is 100%.Repeatability (same conditions, same operator) and reproducibility (same conditions, two different operators) of manual segmentation was examined by re-segmentation in 25 randomly selected scans.

Common carotid segmentation in 18F‐sodium fluoride PET/CT scans: Head‐to‐head comparison of artificial intelligence‐based and manual method

“…To our knowledge, they are the first reported models used for this purpose. Our model was able to segment much faster than what is possible with the manual approach even after training based on a very small number of scans, similar to what we have previously shown with regard to CNN‐based segmentation of the aorta and the heart (Piri, Edenbrandt, Larsson, Enqvist, Nøddeskou‐Fink et al, 2021 ; Piri, Edenbrandt, Larsson, Enqvist, Skovrup et al 2021 ).…”

NaF-PET Imaging of Atherosclerosis Burden

“…One such approach is fused assessment of peak systolic wall shear stress measured by 4D magnetic resonance imaging and NaF-PET to characterize disturbed aortic vascular function [ 68 ]. Another is the combination of measurement of average uptake in vessel whole sections, like major parts of or whole aorta [ 37 , 69 , 70 , 71 ], or in the entire heart [ 23 , 24 , 25 , 26 , 27 , 72 , 73 ]. The latter approach compensates for cardiac movements and shortcomings in analyzing only part of cardiac atherosclerosis burden.…”

“…These many efforts call for a second initiative, i.e., systems that can facilitate, improve, and speed up manual or semi-automated, time-consuming, and observer-dependent segmentation. An efficient answer to this appears to be artificial intelligence-based processing, which, besides minimizing observer dependence and being much faster ( Figure 5 ), is constantly being improved, while at the same time allowing consideration also of multiple other relevant data [ 72 , 73 , 74 , 75 , 76 ].…”

PET-Based Imaging with 18F-FDG and 18F-NaF to Assess Inflammation and Microcalcification in Atherosclerosis and Other Vascular and Thrombotic Disorders

“…A retrospective analysis of 86 healthy controls and 50 patients with persistent chest pain revealed using the ACCS approach of NaF uptake in the whole heart as measured by the mean standardized uptake value (SUVmean) was higher in patients compared to the control subjects and could be employed to retrospectively predict the patient status [ 62 ]. Overall, these studies demonstrate the suitability and potential of quantifying disease risk through global assessment, of which the latter has now become an attractive option that is quick and easy to perform, especially using artificial intelligence-based processing [ 63 , 64 , 97 , 101 ].…”