MRI for

A collaboration of the TU/e Biomedical Image Analysis group and the Neurosurgery Dept. of Maastricht University Hospital and Medtronic, 2005-2014.

Brain shift

Neurosurgical procedures are typically used for:

  • Treatment of movement disorders (e.g. deep brain stimulation);
  • Removal of brain tumours;
  • Vascular brain surgeries.

Accurate targeting accuracy is required to minimize the risk of brain damage. But immediately after opening the skull, the pressure changes, and a severe brain shift may occur (up to 5-15 mm, see movie).
The pre-operative MRI is then of less value.
A solution is imaging during the operation.

In 2004, the Dept. of Neurosurgey of Maastricht University Hospital decided to acquire an ultra-low field (0.12 Tesla) inter-operative MRI: Medtronic's  Odin Polestar N20.
The system came equipped with a SteathStation for accurate optical tracking of instruments during surgery.
As clinical physicist, I have been advising in the acquisition process, and we started a number of supportive BME projects (see below).

The installation of the new MRI scanner (02-05-2005).
Left dr. Geert Spincemaille. 

During the opration the magnets are down.

The static north and south pole magnets can be brought up, on both sides of the head of the patient.

During the scan a Faraday cage is moved over the system and the patient, preventing the pick-up of any unwanted external radio waves.

Neurosurgeon Pieter Kubben MD (author of the world's #1 neurosurgical app NeuroMind, which has been downloaded over 130.000 times) wrote his PhD thesis on the use of interoperative MRI for neurosurgery:
Ultra low-field strength intraoperative MRI for glioblastoma surgery (Maastricht University, 06-02-2014).

Pieter Kubben wrote a thanks in his thesis: B. ter Haar Romeny, dear Bart, thank you for the pleasant cooperation with the department Biomedical Image Analysis (BMIA) from TU/e. It was a new world for me, where you quickly made you feel at home. It was an enrichment to experience science through technical eyes, and sometimes a challenge to create the connection with clinical thinking. I look back on a fun and educational time and wish you and your wife all the best for your retirement, and above all good health!

Bram Platel, I really enjoyed our collaboration in the BMIA context. Wether the conversation now was about non-linear image transformations (on a Mac, of course) or unlocking the first iPhone: as Apple adepts, we understood each other perfectly! The UMCN can count itself lucky with you added to the crew … 

MSc and PhD students from the BMIA group (Ralph, Ellen, Joost, Annet), thanks for the nice collaboration. You have amazed me what you can do in such a short period of time to achieve, under the guidance of Bart, Bram, and Anna Vilanova, and I hope a similar one in the future to experience collaboration again.

learn more

The neurosurgeons now manage to remove a maximum amount of tumor tissue from the brain or pituitary gland while maximally spare surrounding unaffected tissues. That limits and even prevents additional brain damage as a result of the surgery itself. During the surgery, a scan is made, and checked whether everything has been removed. And if that is the case (or not the case) the operation is continued until the goal is reached.

Another advantage of the ultra-low magnetic field (0.12 T) is the increased safety, and no need for special MRI-compatible surgical and supporting equipment. 

The disadvantage is the lower image quality.
Therefore we carried out the following projects with our staff and BME students, to support the neurosurgeons with:

  • Brain shift quantification – image warping
  • Image registration 0.12T – 3T
  • Image enhancement – denoising
  • Warping a brain atlas to the iMRI images
  • Calibration phantom for field homogeneity

Field homogeneity
measurements for spatial accuracy

The field homogeneity determines the spatial accuracy. We designed a 3D spatial phantom with teflon bars and spheres.

A team of TU/e BME students performed all quantitative experiments with the phantom on-site.

Atlas matching

For precise navigation it is extremely helpful to have a good brain atlas. I knew prof. Wieslaw Nowinski from the European Congress of Radiology in Vienna (see here). He develops the world’s best and most accurate brain atlases (by Thieme).

BME students Edwin Benning and Joris Korbeeck developed a clever registration method, based on radial basis functions and nonlinear multi-scale Nelder-Mead optimization, to warp Nowinski’s Cerefy atlas to the interoperative MRI images, all in 3D.

Integration of registered reference data with the Cerefy Atlas of Brain Anatomy.

Manual marking of recognizable landmarks in both atlas and high resolution data.

Edwin and Joris were welcome at the A*STAR Institute of prof. Nowinskin in Singapore for a 5-months externship (Apr-Aug 2005).
The code is fully written in Mathematica.



The interoperative MRI is equipped with a Medtronic Stealthstation. This system tracks with 3 cameras the position of reflective balls on the surgical equipment, for precise 3D tracking and visualization of the instruments in the pre- and interoperative images.

In 2009 Joris Korbeek wrote a coupling between the StealtStation and an iPhone as his MSc project, supervised by Bram Platel.

Back to focus area
computer-aided diagnosis