Femoral CT Foundation Model Accurately Estimates Age and Sex

A 3D vision transformer foundation model achieved high sex classification (AUC 0.979-0.988) and age estimation mean absolute error of ~8 years on femoral CT from 1,489 Japanese individuals, with attention maps highlighting distal femur for sex and shaft for age.

• Retrospective study of 1,489 Japanese postmortem CT femora (age 20-90) used a 3D vision transformer foundation model with multitask learning for simultaneous sex and age estimation.
• Sex classification achieved balanced accuracies of 0.897 (left) and 0.929 (right), with AUCs of 0.979 and 0.988; age estimation had mean absolute error ~8 years (lower error in females).
• Attention maps showed the distal femur was most important for sex estimation, while the femoral shaft contributed most to age estimation.

Automated summary

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