Deep Learning Enables Automated 3D Hip Morphology Assessment in 2,454 Adolescents

Automated nnU-Net segmentation of hip MRI achieved Dice scores of 0.97 for femoral bone and 0.76-0.77 for cartilage, providing reference ranges for 3D morphology in 2,454 young adults (mean age 18.4y). Significant sex and BMI differences found. Code open source.

• Reference ranges for 3D hip morphological biomarkers established from 2,454 participants.
• Significant differences in hip morphology between males and females and across BMI groups.
• All code publicly available, enabling reproducible large-scale assessments.

Automated summary

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