Clinical Validation of Deep Learning for Intracranial and Ventricular Volume Segmentation on CT

A deep learning model (DeepCTE3D) segmented intracranial and ventricular volumes on diverse CT scans with high similarity scores. Differences by patient sex and scanner type were not clinically significant.

• Validated on a real-world dataset including normal and pathological scans from diverse patient demographics and scanner models.
• Secondary analysis found volume differences by sex and scanner model, but these were not clinically significant.

Automated summary

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