A Patient Digital Twin to Enhance Orthopedic Care

The emergence of patient digital twins has revolutionized orthopedic care, driven by advancements in 3D modeling and motion capture (MoCap) technology. This PhD thesis aims to develop a comprehensive patient digital twin framework to enhance the prevention, diagnosis, treatment, and management of orthopedic conditions. The integrated approach combines insights from medical professionals, extensive literature reviews, and innovative data from advanced technologies. Central to this research is the use of 3D modeling to the development of patient digital twin. Deep learning (DL) techniques for automated image segmentation permits to create precise 3D models from medical imaging data. These models facilitate detailed analysis of patient anatomy and morphology, essential for identifying pathologies and customizing treatments. Through statistical shape modeling (SSM), it is possible to study anatomical variations and pathologies. They are even employed to study further automates landmarking and measurements, streamlining the creation of personalized prostheses. Furthermore, Finite Element Analysis (FEA) simulates biomechanical changes due to pathologies or treatments, providing insights into their effects compared to healthy conditions. Motion capture technology is crucial for developing patient digital twins, offering critical tools for analyzing human movement. This analysis is vital for identifying locomotor system pathologies, evaluating rehabilitation protocols and surgeries, and monitoring disease progression. The patient digital twin enables comprehensive patient assessments, including risk factor identification, pathology diagnosis, impact analysis, personalized treatment planning, and evaluation of treatment efficacy. The methodology presented in this thesis addresses the need for a versatile approach applicable to various orthopedic cases, accommodating diverse pathologies while ensuring robustness and comprehensibility. Selected orthopedic pathologies, chosen for their prevalence, clinical significance, and representation of different anatomical regions, demonstrate the adaptability and effectiveness of the methodology. Key contributions include the automatic segmentation of knee bones, particularly the femur, using DL techniques. A statistical shape model of ACL-injured femurs was developed to study anatomical variations under pathological conditions. Comprehensive patient care assessments were illustrated through clinical cases, investigating morphological risk factors for ACL injury in the knee and instability in the shoulder. The methodology also supported the diagnosis of shoulder disorders by examining patient morphology and identifying Trendelenburg gait using MoCap systems. The FE simulations were used to explore the impact of meniscal tears on knee biomechanics. Personalized treatment approaches for the knee, including the design of customized knee implants, were investigated. Additionally, the impact of total hip arthroplasty and various surgical approaches was evaluated using gait analysis. This research advances the development of patient digital twins in orthopedics, providing a robust and adaptable framework to improve patient outcomes through personalized and precise medical care.