CT-based deep learning model for ureteral length matches gross specimen measurements

A deep learning model for CT-based ureteral length measurement yielded a mean of 269 mm, not significantly different from the gross specimen reference (p>0.999).

• Trained on 3629 CT series (1502 patients) and prospectively tested on 70 patients undergoing nephroureterectomy.
• Regression showed a small systematic underestimation with longer ureters (coefficient -0.54 mm per mm, 95% CI -0.70 to -0.39; p<0.001); no other patient factors affected discrepancy.
• Single-center design with a modest prospective cohort limits generalizability; multicenter validation is needed.

Automated summary

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