Clinical Evaluation of Deep Learning for Tumor Delineation on<sup>18</sup>F-FDG PET/CT of Head and Neck Cancer
Dávid Kovács, Claes Nøhr Ladefoged, Kim Francis Andersen, Jane Maestri Brittain, Charlotte Birk Christensen, Danijela Dejanović, Naja Liv Hansen, Annika Loft, Jørgen Holm Petersen, Michala Holm Reichkendler, Flemming Littrup Andersen, Barbara Malene Fischer
Abstract
Artificial intelligence (AI) may decrease <sup>18</sup>F<b>-</b>FDG PET/CT–based gross tumor volume (GTV) delineation variability and automate tumor-volume–derived image biomarker extraction. Hence, we aimed to identify and evaluate promising state-of-the-art deep learning methods for head and neck cancer (HNC) PET GTV delineation. <b>Methods:</b> We trained and evaluated deep learning methods using retrospectively included scans of HNC patients referred for radiotherapy between January 2014 and December 2019 (ISRCTN16907234). We used 3 test datasets: an internal set to compare methods, another internal set to compare AI-to-expert variability and expert interobserver variability (IOV), and an external set to compare internal and external AI-to-expert variability. Expert PET GTVs were used as the reference standard. Our benchmark IOV was measured using the PET GTV of 6 experts. The primary outcome was the Dice similarity coefficient (DSC). ANOVA was used to compare methods, a paired <i>t</i> test was used to compare AI-to-expert variability and expert IOV, an unpaired <i>t</i> test was used to compare internal and external AI-to-expert variability, and post hoc Bland–Altman analysis was used to evaluate biomarker agreement. <b>Results:</b> In total, 1,220 <sup>18</sup>F<b>-</b>FDG PET/CT scans of 1,190 patients (mean age ± SD, 63 ± 10 y; 858 men) were included, and 5 deep learning methods were trained using 5-fold cross-validation (<i>n</i> = 805). The nnU-Net method achieved the highest similarity (DSC, 0.80 [95% CI, 0.77–0.86]; <i>n</i> = 196). We found no evidence of a difference between expert IOV and AI-to-expert variability (DSC, 0.78 for AI vs. 0.82 for experts; mean difference of 0.04 [95% CI, −0.01 to 0.09]; <i>P</i> = 0.12; <i>n</i> = 64). We found no evidence of a difference between the internal and external AI-to-expert variability (DSC, 0.80 internally vs. 0.81 externally; mean difference of 0.004 [95% CI, −0.05 to 0.04]; <i>P</i> = 0.87; <i>n</i> = 125). PET GTV–derived biomarkers of AI were in good agreement with experts. <b>Conclusion:</b> Deep learning can be used to automate <sup>18</sup>F<b>-</b>FDG PET/CT tumor-volume–derived imaging biomarkers, and the deep-learning–based volumes have the potential to assist clinical tumor volume delineation in radiation oncology.