A software system for measuring hip displacement in children with cerebral palsy frompelvic radiographs

Background Cerebral palsy (CP) is the most common physical disability of childhood in the UK affecting 1 in 400 children. Children affected by CP are at high risk of developing hip problems – with the ball of the hip moving out of the socket. This often leads to severe pain, problems sitting and difficulties with personal care. Hip dislocation can be prevented by monitoring the hip closely with radiographs and performing minor procedures early if there are signs of hip problems. The Cerebral Palsy Integrated Pathway (CPIP) is an NHS programme to closely monitor the hips of children with CP. In CPIP, children have regular radiographs, which are reviewed by experts to identify abnormalities. Whilst this approach is effective, the review is time-consuming for medical staff, has regional differences in uptake, and adds delays and costs into the CPIP programme. Aims and Objectives We will develop and register as a medical device a complete system for measuring key parameters (Reimer s Migration Percentage, Head Shaft Angle) from routinely acquired pelvic radiographs of children. The system should be as accurate as expert human annotators. The system will (i) access radiographs from the NHS Trust s image archives; (ii) automatically process images to calculate the relevant measurements; and (iii) send the measurements to the centralised national CPIP database. Methods We have developed software that can automatically locate the outline of bones in radiographs of children s hips. We have shown that this works amongst children whose hips are beginning to dislocate or are dislocated, and that we can make measurements with accuracy similar to that of human experts. We will use this prototype software as the basis for a complete medical device that can be integrated into the existing clinical workflow. We will expand our training/tests sets to optimise the performance of the system and evaluate it in representative conditions on images collected from four NHS trusts. We will establish infrastructure to develop the system to relevant standards and prepare for regulatory accreditation as a medical device. Patient Involvement The prototype has been developed in partnership with families affected by CP. We have held multiple family workshops to discuss the role of research, the use of their data, and the acceptability of the system. Families have suggested how the system could be optimised to support them when advocating for their child. We will continue to foster this partnership throughout the study, by involving families in the study progress and oversight to ensure that the benefit of the system is maximised for affected children. Anticipated impact and dissemination Results will be disseminated through traditional (i.e. conference/journal publications) and non-traditional (i.e. social media, blogs of CP user groups) routes, to ensure that outcomes penetrate through to the clinical community, commissioners and affected families. Once certified and integrated into the clinical workflow, the system will help identify children at risk of hip dislocation. This will save clinicians time, enable earlier interventions and ensure prompt, consistent diagnoses to facilitate equal access to CP services across the UK.