Dual-domain MIM based contrastive learning for CAD of developmental dysplasia of the hip with ultrasound images

Ke Sun, Jing Shi, Ge Jin, Juncheng Li, Jun Wang, Jun Du, Jun Shi

Abstract: Highlights•The DDMCL integrates the spatial domain MIM and frequency domain MIM into a unified framework.•The pretext task simultaneously captures relevant information in dual domains to learn more comprehensive and effective feature representations.•A novel contrastive loss function is developed to promote information interaction between spatial and frequency domain MIMs during CL.