Key Researchers, University of Aarhus, Denmark Thomas Sangild Sørensen – Research Assistant Professor – Master of Science in Computer Science (MSCS) –

Key Researchers, Key Researchers, University of Aarhus, DenmarkUniversity of Aarhus, Denmark

• Thomas Sangild Sørensen

– Research Assistant Professor

– Master of Science in Computer Science (MSCS)

– PhD in Medicine

• Jesper Mosegaard

– Lecturer, Institute of Information Studies

– Master of Science in Computer Science (MSCS)

– PhD candidate in Computer Science (thesis defence November 2006)

• Karsten Østergaard Noe

– Master of Science in Computer Science (MSCS)

– PhD fellow in Computer Science

Key clinical partnersKey clinical partners

• Aarhus University Hospital, Denmark

– Dept. Cardiothoracic Surgery, Vibeke Hjortdal, Professor of Surgery, MD

– Dept. Cardiology, Senior Consultant, Keld Sørensen, MD

– Dept. Oncology, Cai Grau, Research Professor, MD

• Guys Hospital, Kings College, London

– Reza Razavi, Professor of Cardiovascular Imaging, MD

• University Hospital Tubingen, Germany

– Ludger Sieverding, Professor of Cardiology, MD

• Clinic of Congenital Heart Disease, Ruhr University Bochum, Germany

– Philipp Beerbaum, Director of Pediatric Cardiac MRI, MD

Surgical Simulation on the GPUSurgical Simulation on the GPU

• Real-time simulation of surgical procedures on congenitally malformed hearts

• Can we reduce mortality rates in complex surgery?

• Scenarios

– Training of novice surgeons

– Preoperative surgical planning of patient-specific procedures

[Click for Movie]

ReconstructionReconstruction

• Morphological models reconstructed from 3D MRI

– Patient-specific morphology

– Possibly enhanced by post-processing if used for educational purposes

– On the model to the right we image a patient with several atrial and ventricular septal defects – i.e. holes between the main chambers of the heart

Technical requirementsTechnical requirements

• Real-time tissue deformation

– The CPU is simply not fast enough

– Use the GPU for general computations instead

• >20-fold acceleration

• The GPU can significantly accelerate computations in a range of medical simulation applications

Dose planning in radiotherapyDose planning in radiotherapy

• CT imaging of neck cancer

• The tumor (arrow) shrinks during treatment

• It will be advantageous to update the dose plan according to the morphological changes

• A deformable registration method (accelerated >30 times on the GPU) is used to calculated the accumulated dose