Does consensus contours improve robustness and accuracy on $$^{18}$$F-FDG PET imaging tumor delineation?

Abstract

Purpose: The aim of this study is to explore the robustness and accuracy of consensus contours with 225 nasopharyngeal carcinoma (NPC) clinical cases and 13 extended cardio-torso simulated lung tumors (XCAT) based on 2-deoxy-2-[$$^{18}$$ 18 F]fluoro-D-glucose ($$^{18}$$ 18 F-FDG) PET imaging. Methods: Primary tumor segmentation was performed with two different initial masks on 225 NPC $$^{18}$$ 18 F-FDG PET datasets and 13 XCAT simulations using methods of automatic segmentation with active contour, affinity propagation (AP), contrast-oriented thresholding (ST), and 41% maximum tumor value (41MAX), respectively. Consensus contours (ConSeg) were subsequently generated based on the majority vote rule. The metabolically active tumor volume (MATV), relative volume error (RE), Dice similarity coefficient (DSC) and their respective test–retest (TRT) metrics between different masks were adopted to analyze the results quantitatively. The nonparametric Friedman and post hoc Wilcoxon tests with Bonferroni adjustment for multiple comparisons were performed with $$P<$$ P < 0.05 considered to be significant. Results: AP presented the highest variability for MATV in different masks, and ConSeg presented much better TRT performances in MATV compared with AP, and slightly poorer TRT in MATV compared with ST or 41MAXin most cases. Similar trends were also found in RE and DSC with the simulated data. The average of four segmentation results (AveSeg) showed better or comparable results in accuracy for most cases with respect to ConSeg. AP, AveSeg and ConSeg presented better RE and DSC in irregular masks as compared with rectangle masks. Additionally, all methods underestimated the tumour boundaries in relation to the ground truth for XCAT including respiratory motion. Conclusions: The consensus method could be a robust approach to alleviate segmentation variabilities, but did not seem to improve the accuracy of segmentation results on average. Irregular initial masks might be at least in some cases attributable to mitigate the segmentation variability as well.

Does consensus contour improve robustness and accuracy in 18F-FDG PET radiomic features?

“…The realistic anthropomorphic simulations were conducted using the XCAT phantom [ 16 ] and software for tomographic image reconstruction (STIR) [ 17 ] as built exactly the same as our previous study (Fig. 1 ) [ 15 ]. In this study the respiratory motion was not taken into consideration to avoid the influence of respiratory movement.…”

“…We reanalyzed PET/CT scans on 225 NPC patients who underwent 18 F-FDG PET/CT scans at Meizhou people’s hospital from 2018 to 2020. Patient characteristics were previously described [ 15 ]. All PET images have a matrix size of 200 200 voxels with a voxel size of 0.30 0.41 0.41 cm 3 .…”

“…To overcome the inconsistency in tumor delineation and analysis, the consensus methods have been employed in many recent studies [ 11 – 15 ]. McGurk et al [ 11 ] assessed consensus contour derived from different segmentation methods and found that consensus contour could improve accuracy and robustness compared with the varying performances of different segmentation methods.…”

“…In our previous companion study, we investigated the robustness and accuracy of consensus contour based on different individual segmentation results using clinical NPC cases and extended cardio-torso (XCAT) simulated tumors. Our results demonstrated that consensus contour could be a robust approach to mitigate segmentation variability, but did not appear to improve the segmentation accuracy on average [ 15 ]. Yet, on the basis of results published in the literature so far, it is still questioned to the impact of consensus contour from different PET segmentation methods on the robustness and accuracy for radiomic features.…”

Optimizing scan duration and tracer dose in radiomic analysis for 18F-FDG PET/MR brain cancer

“…However, the results of radiomic analysis varied considerably in many studies, creating challenges to quantitative PET image analysis [3][4][5][6][7][8][9][10][11][12][13][14][15][16]. To mitigate these considerable variabilities in tumor quantitative analysis, strategies such as consensus methods, and harmonization of imaging procedures, have been developed and evaluated in recent studies [17][18][19][20][21][22]. Besides, potential factors that may affect tumor quantitative analysis have been widely explored as well and thus enhance the harmonization in PET imaging [3,7,[22][23][24].…”