<p>Knee lesions frequently result in morphological alterations or degenerative defects on the joint surface. This study aims to develop a reconstruction model that restores the pathological femoral condyle surface into a healthy morphology. A statistical shape model (SSM) was constructed using a healthy femoral condyle surface database as the base template for surface restoration. The restoration was achieved through differential coordinates (DC) transformation, constrained by statistical model parameters, template data, and the healthy portion of the pathological femoral condyle surface. Both the proposed method and the conventional SSM method were used for lesion restoration. The performance of both methods was compared using static geometric errors and dynamic joint motion errors, assessing surface accuracy and joint motion impact. For the proposed model, reconstruction errors remained low across both localized and extensive regions. Compared with the conventional SSM approach, the proposed method achieved better accuracy in both static geometric reconstruction and dynamic joint motion. It provides critical osteotomy positioning parameters and design references for patient-specific implants, and enhances the accuracy and efficiency of joint replacement surgeries.</p> Graphical Abstract <p></p>