LOng-Term anatomical fluid dynamics for new Univentricular heartS palliation (LOTUS)

Children with functionally univentricular hearts (UVH) have poor life prognosis, despite surgical treatment. This project aims to build an open-source combined Artificial Intelligence and Computational Fluid Dynamics (CFD) tool to determine 'optimal' designs for the Fontan procedure to be used by surgeons, to enhance patient life expectancy (currently 20-30 years).
Detailed imaging from UVH patients will be provided by USM to produce 3D geometries on which UK numerical models can be applied. USM currently treats 40-50 new UVH patients/year and 150 in follow-up, therefore they have a large databank covering a wide range of anatomical morphologies changing with patient growth and various palliative stages. This databank enables the construction of a novel mathematical model of how morphology/flow in later life depends on surgical procedure, with the prediction of the 'optimal' surgical design minimizing the negative impact of the Fontan circulation. Combining high-fidelity curvilinear discretization, multiscale modelling, and machine learning, the CFD tool will reliably inform on long-term impact of surgical procedures, including a novel physiological Fontan procedure by the UK team.
Expected outcomes.
1. Study of the long-term prognosis of subjects with complex congenital heart defects, specifically following all patients with UVH.
2. Consolidate partnership between Malaysia and UK teams, provide training to USM personnel introducing CFD tools for surgery optimisation.
3. Develop a patient-specific open-source computational tool to optimize surgical procedures in children with functionally UVH, for better long-term quality of life.
4. Combine artificial intelligence techniques with efficient and reliable multiscale CFD modelling based on recent polyhedral discretisations to predict long-term outcome of surgery.
5. We envisage that the project's tool will provide a new insight on the reason for the failure of UVH palliation.
6. The tool will also permit the validation of new Fontan procedures based on more physiological types of surgical connections, such as the one recently proposed and evaluated with CFD studies by the UK team.

This project will take advantage of a unique partnership between USM and Hospital Raja Perempuam Zainab II, in Malaysia, and a world-leading team in the UK, blending together extensive clinical experience from both countries with engineering and mathematical modelling and High-Performance Computing (HPC) research. USM has a well-established research-active Unit of Paediatric and Congenital Cardiac Surgery, with state-of-the-art data acquisition capabilities and a unique and well-maintained data set of patients of all ages. This will be used for the calibration of the mathematical/computational model aimed at explaining the long-term impact of Univentricular Heart (UVH) palliation.
The prediction of the morphological changes in the vascular system, as consequences of growth and modifications induced by the surgical procedures, will be achieved by a state-of-the-art artificial intelligence technology based on the generative adversarial neural networks.
The project will make use of freely available specialised software, such as CRIMSON, SimVascular, VMTK, for image segmentation and geometry meshing, enhanced by advanced image segmentation software and novel geometric rendering of vascular alterations from surgery based on advanced parametric surface classes. New Computational Fluid Dynamics (CFD) approaches based on polyhedral discretisations will be developed for computational turnaround speedup and reduction of highly skilled time-consuming human intervention at the meshing stage. In particular, three incremental CFD solver will be released, of increasing fidelity, based on novel polytopic 3D meshes of the vessels, pipe-like curvilinear elements designed to reduce the computational costs inherent to standard 3D meshing, inclusion of fluid-structure interaction phenomena modelling compliant vessels.

Impact on children's health: LOTUS will directly help about one in every thousand children, who is born with a functionally univentricular heart (UVH) condition. Even with palliation, these children have poor life expectancy. Co-morbidities, economic hardship, and limited health-care resources give scope for the greatest improvement in DAC countries.
Through its unique combination of Malaysian medical data & clinical practice, with UK UVH modelling & long-term appraisal techniques, LOTUS is set to improve the long-term prognosis of children with UVH both in Malaysia and worldwide. This will deliver a step change in medical practice. Specifically, it will create a computer-assist aid for determining the best individualised treatment based on the impact of paediatric surgery in the long-term.
Impact on application of machine learning to UVH conditions: The impact of decisions made during surgery over a patient's life is poorly understood. Combining the available data from CT and MRI scans from USM (throughout a patient's life) with machine learning will allow the ageing of a new patient's pulmonary vasculature over time. This scientific advance pushes the boundaries of the state-of-the-art use of machine learning and artificial intelligence in UVH condition monitoring and prognosis. In fact, once this ageing can be predicted, multi-objective optimisation will deliver a surgical design that will improve the patient's prospects in both the short and long term.
Impact on CFD modelling of UVH vascular system: This project will make use of and extend novel CFD techniques on polyhedral meshes that will significantly reduce the time taken to carry out the simulations and hence make the multi-objective optimisation tractable over a time-frame suitable for surgeons' decision making. Disseminated through journal publications, these state-of-the-art advances will benefit the bioengineering modelling community.
Impact on healthcare professionals: A specific workpackage is dedicated to delivering impact in DAC healthcare by providing training to USM personnel for the introduction of CFD tools for surgery optimisation. The software will be made freely avilable and documentation and training materials will be provided so that surgeons and, hence, patients throughout the world will benefit from its use.
Impact on early stage researcher career: The PDRA employed in this project will acquire skills in line with the MRC skill priorities list. Importantly they will gain experience in advanced CFD (AptoFEM), modelling of the pulmonary vasculature (Crimson), machine learning and artificial intelligence, as well as understanding of the medical aspects of the project. They will also benefit from the interdisciplinary nature of the project, which combines medical imaging, surgery, computational mathematics and engineering. The PDRA will improve their management and project organisation skills, will work independently and develop presentation and communication skills, by presenting at both clinical and mathematical conferences. All these, combined with the wide range of supplementary courses on offer within the University, and elsewhere (British Science Association, Royal Society), will positively impact on the future career prospects of the PDRA and ensure they become a future research leader. The PI and Co-Is will benefit by establishing fruitful collaborations between disciplines in the life and physical sciences.
Impact on economy and on welfare: Each child with a UVH condition requires through-life support and follow-up. The condition limits the young adult's productivity due to death typically occurring in their 20's. LOTUS' through-life personalized optimization of treatment stands to decrease the later-life morbidity, support needs, increase life expectancy, and therefore adult life productivity. Bettering the quality of life conditions will reduce the welfare dependency of UVH treated children in adulthood.