ESR 10: Optimization-Based Fusion of Surgical Planning Data for Intraoperative Navigation

Early Stage Researcher: Patricia Alcañiz
Host institution: GMV Soluciones GLobales Internet Sau (GMV, ES)
Supervisor: Prof. Miguel Otaduy, Universidad Rey Juan Carlos (URJC, ES)
Clinical expertise: David Macías Verde, Medical Physicist, Hospital Universitario de Gran Canaria Dr. Negrín, ES,
Co-supervisor: Carlos Illana, Advanced Healthcare Technology Product Manager, GMV
Further institution involved: University of Copenhagen (UCPH, DK)

Objectives

This ESR will enable intelligent fusion of preoperative planning data with real patient anatomy in an intraoperative setting. The first objective will consist of a novel automated methodology to design a compact deformation model from a patient’s medical image, using numerical coarsening methods. This compact model will be used to deform preoperative simulation data, navigation markers, etc. in an efficient manner. The second objective will consist of a novel variational formulation to match the input anatomy to intraoperative tracking data. The task will be formulated as a constrained optimization problem, which will be efficiently solved thanks to the compact deformation model resulting from the first objective. The third and final objective will comprise novel metaphors for the intuitive specification of objective functions and constraints for the optimization problem. This will allow clinicians to work directly on the patient data, without interacting with technical details. The fusion methods will be tested on an intraoperative radiotherapy planning and navigation tool.

Expected Results

An optimization tool for fusion of preoperative planning data into intraoperative navigation, built on top of novel compact deformation models of medical images, variational solvers for medical image fusion, and interface metaphors for problem specification. Intraoperative radiology will be selected as clinical test case.

Planned research stays

  • GMV (3 months): Training on the radiology application and design of the preliminary medical image deformation methods.
  • URJC (2 months): Design of optimization-based fusion methods built on top of the compact medical image deformation model.
  • GMV (1 month): Addition of intuitive problem specification metaphors, integration on the intraoperative radiology planning and navigation tool
  • URJC (2 months): continued methodological training
  • GMV (5 months): Addition of intuitive problem specification metaphors, integration on the intraoperative radiology planning and navigation tool and validation on clinical benchmarks
  • URJC (3 months): continued methodological training
  • GMV (9 months): continued work on navigation tool and validation
  • UCPH (3 months): Training on advanced homogenization methods.
  • GMV (8 months): thesis writing

Further information of the project at URJC lab webdescription.