Mathias no longer works for Örebro University. These pages are kept for archive purposes only.

Medical Image Processing

I am currently together with P. Thunberg managing a small collaboration project together with the Centre for Biomedical Engineering Research (MTFC), Örebro University Hospital. The collaboration project involves staff from both Örebro University as well as from Örebro University Hospital and includes several specific research projects with the common goal of using advanced image processing, facilitatet by the emergence of GPGPU based computations, to improve medical procedures and diagnosis.

Finansing

This project is financed by faculty money covering 40% of the salary of M. Broxvall and by MTFC, Örebro University Hospital covering 40% of the salary of M. Daotis, and contributing the time and resources of the hospital staff.

Staff and collaborators

The following staff is directly involved in this collaboration project.
  • M. Broxvall, Docent, Örebro University, 40%
  • P. Thunberg, Head of MTFC, Örebro University Hospital
  • M. Daoutis, Post Doc, Örebro Universityl, 40%
Furthermore the projects collaborate with a number of other researchers affiliated with MTFC and other departments at Örebro University Hospital. This includes Dr. Kent Emilsson (Clinical Physicology), Dr. M. Liden (Dept. of Radiology), S. Jorstig (MTFC) and associated staff members at the respective departments.

Visualization of the Aortic Cusps using timesampled 3D echocardiography

Many commonly occuring forms of heart diseases, such as insufficience, involve the aorta cusp and constitute acute medial emergencies. In many cases where the cusp is repaired or replaced, the surgeon needs to make accurate medical decisions based on the shape and form of the cusps. Due to limitations in visualisation of 3D medical imaging, these decisions currently have be performed during surgery which gives only a very limited timeframe for deliberation. In this project, we investigate a method to pre-compute a quantitatively correct visualisation of the aorta cusps as they will appear during surgey, based on non-invasive 3D ultrasound imaging. The core of the method is to create a 3D model of the cusps as they are working, and to unfold this model to reflect the way the cusps will be opened during the surgery. This will allow the diagnosing and preparation of patients well before surgery. Additionally, it will offer the physician the possibility to investigate cusps as they are operating under normal cirumstances. This gives additional information that can be used for medical assessments. The main research challenges of this goal is due to the computational complexity of c

Simulated view looking down through the aorta
Preliminary results in this project have led to the creation of a new GPU based method for fast adaptive filtering of echocardiography. This research have been published in IEEE Transactions of Medical Imaging (see publications) and we are currently in addition to developing the above 3D model of the heart also further quantifying the gains of this filtering method for diagnosis of heart diseases.
Last modified 03/13/12