Impact of reader experience on the time-saving potential of a novel computer-aided tool for Ann Arbor staging of pediatric hodgkin lymphoma.
AI interpretation is pending for this paper.
Open original publication →What the AI sees
Not AI summarized yet.
Research significance
Pending deeper interpretation.
Source abstract
BACKGROUND: Accurate staging of Hodgkin lymphoma in children is important for selecting appropriate therapies. However, evaluation of Ann Arbor staging criteria on whole-body 18F-FDG PET scans is time-consuming and subject to reader-dependent variability. OBJECTIVE: To determine if a novel computer-aided staging tool can improve the time-efficiency and accuracy of Ann Arbor staging. METHODS: 18F-FDG PET/MRI scans of fifty-six pediatric patients were divided into two groups with matched Ann-Arbor stages. Group 1 was staged via traditional inspection and Group 2 was staged with a newly developed web-based staging tool by one radiology resident, four radiologists and two nuclear medicine physicians. An additional radiologist and nuclear medicine physician jointly established the reference standard. The time to generate the Ann Arbor score was compared between group 1 and 2 with the Wilcoxon Rank-Sum test, and agreement with the reference standard was assessed with Cohen's kappa analysis. In addition, staging time was correlated with reader experience level, measured as the number of PET cases interpreted per year, using linear regression analysis. RESULT: Use of the tool significantly reduced Ann Arbor staging time (10.8 ± 6.33 min) compared to traditional staging without the tool (13.87 ± 8.16 min; p = 0.001). The greatest reduction in staging time was observed for the radiology resident (17.63 ± 2.3 min with the tool and 21.5 ± 3.3 without the tool), followed by radiologists (13.63 ± 2.5 min with the tool and 18.03 ± 2.6 without the tool), while no significant change was observed for nuclear medicine physicians (1.45 ± 0.82 min with the tool and 1.39 ± 1.12 without the tool). Agreement with the reference standard improved with tool assistance (κ = 0.616) compared to traditional staging without the tool (κ = 0.945 with the tool). Linear regression demonstrated a significant inverse association between reader experience and time saved using the tool (β = -0.812, p < 0.001; R² = 0.66), indicating greater benefit for less experienced readers. CONCLUSION: The computer-aided staging tool significantly improved time-efficiency and accuracy of Ann Arbor staging, especially for less experienced readers. We provide the tool as a freely accessible web resource.