Periodic Reporting for period 5 - CAD4FACE (Computational modelling for personalised treatment of congenital craniofacial abnormalities)
Berichtszeitraum: 2024-03-01 bis 2025-02-28
Craniosynostosis is a group of congenital craniofacial abnormalities affecting 1 in 1,700 newborns and consisting in premature fusion of one or more cranial sutures during infancy (Fig. 1). This results in growth restriction perpendicular to the axis of the suture and promotes growth parallel to it, causing physical deformation of the craniofacial skeleton, as well as distortion of the underling brain, with detrimental effects on its function: visual loss, sleep apnoea, feeding and breathing difficulties, and neurodevelopment delay. Conventional management of craniosynostosis involves surgery delivered by excision of the prematurely fused sutures, multiple bone cuts and remodelling of the skull deformities, with the primary goal of improving patient function, while normalising their appearance. Craniofacial remodelling surgical procedures, aided by internal (stainless steel springs, Fig. 2) and external (rigid external distractor, RED frame, Fig. 3) devices at Great Ormond Street Hospital (GOSH), have proven functionally and aesthetically effective in correcting skull deformities, but final results remain unpredictable and often suboptimal because of an incomplete understanding of the biomechanical interaction between the device and the skull.
The overall aim of this grant is to create a validated and robust computer framework that integrates patient information and device design to deliver personalised care in paediatric craniofacial surgery to improve clinical outcomes. A virtual model of the infant skull with craniosynostosis including mechano-biology regulation will be developed to simulate device implantation and performance over time, and will be validated using clinical data from patient populations treated with current devices. Bespoke new devices will be designed allowing for pre-programmed 3D shapes to be delivered with continuous force during the implantation period. Patient specific skull models will be used to virtually test and optimise the personalised devices, and to tailor the surgical approach for each individual case.
The overall aim of this grant is to create a validated and robust computer framework that integrates patient information and device design to deliver personalised care in paediatric craniofacial surgery to improve clinical outcomes. A virtual model of the infant skull with craniosynostosis including mechano-biology regulation will be developed to simulate device implantation and performance over time, and will be validated using clinical data from patient populations treated with current devices. Bespoke new devices will be designed allowing for pre-programmed 3D shapes to be delivered with continuous force during the implantation period. Patient specific skull models will be used to virtually test and optimise the personalised devices, and to tailor the surgical approach for each individual case.
The CAD4FACE project has delivered a set of research and technological innovations that have advanced the field of personalised craniofacial surgery for children born with craniosynostosis, by integrating biomedical engineering methodologies with clinical expertise to develop patient-specific tools and novel devices for surgical planning and treatment.
A critical focus of CAD4FACE was understanding the biomechanical properties of the skull bones in children born with craniosynostosis. Over 300 bone samples from more than 250 children were collected during surgery. Part of these samples were imaged at high resolution, and part mechanically tested to study structural and material properties. These data informed computational models that now more accurately simulate skull reshaping after surgery and predict patient-specific surgical outcomes.
To support diagnosis and surgical planning, the team developed statistical shape models for different craniosynostosis conditions. These models describe anatomical variability across populations and serve as a reference for identifying abnormalities. They also support automatic classification, outcome prediction, and allow for machine learning (ML) development using compact, geometry-based descriptions.
The core of the project was a validated computational modelling pipeline based on finite element analyses that integrates patient pre-operative images and population-based computational modelling to predict surgical outcomes for each individual case. The finite element modelling allows simulation on how the child’s skull would respond to different surgical strategies and approaches, including different position/lengths of the osteotomies (bony cuts), and position/number/types of craniofacial distractors. These simulations allow surgeons to explore and compare treatment options in advance.
The population based computational model validated on clinical cases was used to design a new set of distractors made of nitinol, that deliver constant, lower forces when in situ. The design of these devices was optimised using the computational models and prototyped for in-vitro mechanical performance validation and sterilisation testing. Additionally, an instrumented version of the rigid external distractor (RED) system was developed to include a system of strain gauges and derive the distraction forces when the device is in use on the patient in real time, with data streamed via wireless technology.
To overcome the barriers of translating finite element simulations in everyday clinical tools, such as high computational demand and need for engineering expertise, we developed machine learning (ML) models trained on >3,000 synthetic finite element simulations. The ML models can instantly predict changes in skull shape and volume following surgery, providing surgeons with real-time decision support (<5% prediction error).
The project pioneered the integration of virtual and augmented reality (VR and AR) technologies, with applications developed in house, to allow the craniofacial surgeons to plan complex procedures remotely and interactively, to projects the ML predicted optimal surgical plan directly onto a patient’s head in the operating room, and to enhance the communication with the family.
CAD4FACE has enabled development and strengthening of several key collaborations between engineers and clinicians from centres not only in the UK and Europe, but also in the USA, building a strong and interdisciplinary network for advancing care for children with craniofacial conditions.
A critical focus of CAD4FACE was understanding the biomechanical properties of the skull bones in children born with craniosynostosis. Over 300 bone samples from more than 250 children were collected during surgery. Part of these samples were imaged at high resolution, and part mechanically tested to study structural and material properties. These data informed computational models that now more accurately simulate skull reshaping after surgery and predict patient-specific surgical outcomes.
To support diagnosis and surgical planning, the team developed statistical shape models for different craniosynostosis conditions. These models describe anatomical variability across populations and serve as a reference for identifying abnormalities. They also support automatic classification, outcome prediction, and allow for machine learning (ML) development using compact, geometry-based descriptions.
The core of the project was a validated computational modelling pipeline based on finite element analyses that integrates patient pre-operative images and population-based computational modelling to predict surgical outcomes for each individual case. The finite element modelling allows simulation on how the child’s skull would respond to different surgical strategies and approaches, including different position/lengths of the osteotomies (bony cuts), and position/number/types of craniofacial distractors. These simulations allow surgeons to explore and compare treatment options in advance.
The population based computational model validated on clinical cases was used to design a new set of distractors made of nitinol, that deliver constant, lower forces when in situ. The design of these devices was optimised using the computational models and prototyped for in-vitro mechanical performance validation and sterilisation testing. Additionally, an instrumented version of the rigid external distractor (RED) system was developed to include a system of strain gauges and derive the distraction forces when the device is in use on the patient in real time, with data streamed via wireless technology.
To overcome the barriers of translating finite element simulations in everyday clinical tools, such as high computational demand and need for engineering expertise, we developed machine learning (ML) models trained on >3,000 synthetic finite element simulations. The ML models can instantly predict changes in skull shape and volume following surgery, providing surgeons with real-time decision support (<5% prediction error).
The project pioneered the integration of virtual and augmented reality (VR and AR) technologies, with applications developed in house, to allow the craniofacial surgeons to plan complex procedures remotely and interactively, to projects the ML predicted optimal surgical plan directly onto a patient’s head in the operating room, and to enhance the communication with the family.
CAD4FACE has enabled development and strengthening of several key collaborations between engineers and clinicians from centres not only in the UK and Europe, but also in the USA, building a strong and interdisciplinary network for advancing care for children with craniofacial conditions.
Structural and mechanical properties of the skull bones in children born with craniosynostosis.
Validated computational models of spring assisted cranioplasty in different craniosynostosis conditions.
New nitinol springs for sagittal craniosynostosis and posterior vault expansion, optimised using computational modelling.
Instrumented RED frame prototype for in situ distraction force measurements.
Machine learning algorithms for fast simulation of spring assisted cranioplasty.
Extended realities (virtual and augmented) applications for surgical planning and enhanced informed consent for patients’ families.
Knowledge transfer for complex separation of conjoined twins.
Validated computational models of spring assisted cranioplasty in different craniosynostosis conditions.
New nitinol springs for sagittal craniosynostosis and posterior vault expansion, optimised using computational modelling.
Instrumented RED frame prototype for in situ distraction force measurements.
Machine learning algorithms for fast simulation of spring assisted cranioplasty.
Extended realities (virtual and augmented) applications for surgical planning and enhanced informed consent for patients’ families.
Knowledge transfer for complex separation of conjoined twins.