OSCAR artificial intelligence algorithm for iliac stent sizing validated in a multicenter retrospective study

An artificial intelligence algorithm for iliac stent sizing detected 84.6% of stenoses and predicted stent diameter with a 0.13 mean absolute percentage error, similar to expert variability, in 149 patients.

• This retrospective multicenter study evaluated 149 patients who underwent iliac artery stent implantation; annotations came from four board-certified interventional radiologists and model validation used leave-one-out cross-validation.
• OSCAR detected 84.6% of stenoses, with recall 0.89 ± 0.21, precision 0.65 ± 0.28, and Dice similarity coefficient 0.77 ± 0.11.
• Stent diameter and length predictions had mean absolute percentage errors of 0.13 ± 0.18 and 0.33 ± 0.31, comparable to expert variability; the authors describe the work as proof-of-concept.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.