Abstract
Background: The aim of this research was to develop a novel artificial intelligence (AI)-guided clinical decision support (CDS) system, to predict radiation doses to subsites of the mandibleusing diagnostic CT scans acquired before planning of head and neck radiotherapy (RT). Methods: A dose classifier was trained using RT plans from 86 oropharyngeal cancer patients; thetest set consisted of an additional 20 plans.The classifier was trained topredictwhether mandible subsites would receive a mean dose >50Gy.The AI predictionswere prospectively evaluated and compared to those of a specialist head and neck radiation oncologist for 9 patients.Positive predictive value (PPV), negative predictive value (NPV), Pearson’s correlation coefficient, and Lin's concordance correlation coefficient were calculated to compare the AIpredictions to those of the physician. Results: In the test dataset, the AIpredictions had a PPVof 0.95 and NPVof 0.88.For 9 patients evaluated prospectively, there was a strong correlation between the predictions of the AIalgorithm and physician (p = 0.72, p < 0.001).Comparing the AI algorithm versus the physician, the PPVs were 0.82 versus0.25, and the NPVs were 0.94versus 1.0,respectively.Concordance between physician estimatesand final planned doses was 0.62;this was 0.71 between AI-based estimates and final planned doses. Conclusions: An AI-guidedCDStool to predict dental dosimetry prior to head and neck RT was built, validated, and prospectively tested.AI-guided decision supportincreasedprecision and accuracy ofdental dose estimates and improved the quality of pre-RTdental assessment.
Prospective study of artificial intelligence-based decision support to improve head and neck radiotherapy plan quality
