Intra- and inter-operator variability in MRI-based manual segmentation of HCC lesions and its impact on dosimetry

Abstract

Purpose The aim was to quantify inter- and intra-observer variability in manually delineated hepatocellular carcinoma (HCC) lesion contours and the resulting impact on radioembolization (RE) dosimetry. Methods Ten patients with HCC lesions treated with Y-90 RE and imaged with post-therapy Y-90 PET/CT were selected for retrospective analysis. Three radiologists contoured 20 lesions manually on baseline multiphase contrast-enhanced MRIs, and two of the radiologists re-contoured at two additional sessions. Contours were transferred to co-registered PET/CT-based Y-90 dose maps. Volume-dependent recovery coefficients were applied for partial volume correction (PVC) when reporting mean absorbed dose. To understand how uncertainty varies with tumor size, we fit power models regressing relative uncertainty in volume and in mean absorbed dose on contour volume. Finally, we determined effects of segmentation uncertainty on tumor control probability (TCP), as calculated using logistic models developed in a previous RE study. Results The average lesion volume ranged from 1.8 to 194.5 mL, and the mean absorbed dose ranged from 23.4 to 1629.0 Gy. The mean inter-observer Dice coefficient for lesion contours was significantly less than the mean intra-observer Dice coefficient (0.79 vs. 0.85, p < 0.001). Uncertainty in segmented volume, as measured by the Coefficient of Variation (CV), ranged from 4.2 to 34.7% with an average of 17.2%. The CV in mean absorbed dose had an average value of 5.4% (range 1.2–13.1%) without PVC while it was 15.1% (range 1.5–55.2%) with PVC. Using the fitted models for uncertainty as a function of volume on our prior data, the mean change in TCP due to segmentation uncertainty alone was estimated as 16.2% (maximum 48.5%). Conclusions Though we find relatively high inter- and intra-observer reliability overall, uncertainty in tumor contouring propagates into non-negligible uncertainty in dose metrics and outcome prediction for individual cases that should be considered in dosimetry-guided treatment.

Direct comparison and reproducibility of two segmentation methods for multicompartment dosimetry: round robin study on radioembolization treatment planning in hepatocellular carcinoma

“…Similar to prior studies summarized in Table 4 , inter-observer investigation demonstrated better reproducibility for larger volumes, i.e., whole liver and normal tissue [ 19 – 21 ]. This finding is consistent with factors contributing to increased variability, primarily driven by the partial volume effect [ 18 , 20 ]. Despite the similarities in RDC values for whole liver volumes, the variability was higher for whole liver and whole liver NTAD for 99m Tc-MAA SPECT segmentation versus anatomic segmentation.…”

“…The DOSISPHERE-01 study, as well as other studies, utilized such a 99m Tc-MAA SPECT count threshold-based segmentation method to guide volumes of interest, which assumes that 99m Tc-MAA preferentially accumulates in the tumour [ 9 , 17 ]. Other studies used anatomic images (i.e., CT/MRI) to define volumes of interest [ 4 , 11 , 16 , 18 – 20 ]. The TARGET study evaluated both anatomic and 99m Tc-MAA SPECT segmentation methods to compare the real-world utility for 90 Y treatment planning [ 7 , 9 ].…”

Physics-informed deep generative learning for quantitative assessment of the retina

“…In the assessment of vasculature, manual image labelling constitutes a considerable bottleneck in terms of time, expense and labelling accuracy [49], as manual segmentation of a single 2D retinal image can take multiple hours [50]. Inter and intra grader variability can also be significant within the segmentation process [51,52]. Most segmentation studies have been conducted in 2D retinal fundus photographs using public datasets [12,53,54].…”

Improved automated tumor segmentation in whole-body 3D scans using multi-directional 2D projection-based priors

“…in whole-body examinations, is generally a much more challenging task as they can vary significantly in terms of size, shape, texture, etc [ [17] , [18] , [19] ]. In addition, the intra-and inter-rater variability is often very high which makes it difficult to obtain standardized ground truth labels [ 21 ]. Unlike organs and tissues, always present in roughly the same location, tumors can appear in different body parts and spread throughout the body.…”