Deep Brain Stimulation 
Reducing the side effects
for treatment of Parkinson Disease

Bart ter Haar Romeny, Bram Platel, Ellen Brunenberg, Birgit Plantinga, Ralph Brecheisen,  Anna Vilanova
and many MSc students 

A visual account of a research project done in the Biomedical Image Analysis (BMIA) Group at Eindhoven University of Technology, the Netherlands, in close collaboration with Maastricht University Medical Center, Dept. Neurosurgery, 2007-2016, where our BMIA group had a special sub-section: Image-Guided Surgery, led by postdoc Bram Platel.

What is Deep Brain Stimulation?

Parkinson disease

Patients with Parkinson disease may finally develop severe tremor. It is known that the origin of this uncontrolled motor behaviour is in one of the basal ganglia: the subthalamic nucleus (STN), deep in the brain. The discovery of Deep Brain Stimulation (DBS) is attributed to Alim Benabid in the late 1980s, who found that electrical stimulation of the basal ganglia improved the symptoms of Parkinson's disease (but it has a complex history).

Electrical stimulation

Continuous stimulation at 80 or 150 Hz is accomplished by the operative placement of two electrodes in both STNs (left and right), driven by an implanted brain pacemaker under the clavicula.

The result of Deep Brain Stimulation can be pretty spectacular:

Without DBS.

With DBS.

The goal

Limbic (emotion / memory)

The goal of deep brain stimulation for Parkinson’s disease is to stimulate the motor part of the subthalamic nucleus (blue) and to avoid stimulation of the non-motor areas (green and red).

Side effects

However severe side effects can develop:
• In 50% of the cases a behavioral change is noticed in the patient;
• In 14% of the cases the change is severe:
• Depressions
• Mania (e.g. hypersexuality, euphoria)
• Agressiveness The reason is the inaccurate positioning of the stimulating electrode near the STN. The STN consists of three layers (see figure), and stimulation should be done at the somatomotor partition [Handbook on DBS, in Dutch].

The clinical question:
can we help in targeting the STN more accurately?

The project started with a feasibility study by TU/e BME MSc student Ellen Brunenberg (supervised by dr. Bram Platel and dr. Anna Vilanova). Ellen’s MSc thesis was published at MICCAI, and she was rewarded a Dutch Top-Talent PhD scholarship (2007-2011).

The DBS neuro-surgical operation

Before the operation: to image the small STN (lens-shaped, ~ 7 x 6 x 3 mm) the patient is scanned by 3T MRI and high-resolution CT, and the 3D volumes are co-registered.

A frame is mounted to the patient’s skull, to aid the registration. It is quite difficult to locate the STN.

We were surprised how primitive the location was determined (the classical method in literature): 11.5 mm lateral, 2.5 mm posterior and 4.1 mm inferior to the mid-point of the AC–PC line.

The DBS operation

The DBS operation typically takes 5-6 hours. The patient is awake, and his leg stiffness and pupil- and finger reflexes are tested at frequent intervals. 

The skull is penetrated and a needle with 5 electrodes is painstakingly slow driven towards the STN. 

The proximity of the needle to the STN is signaled by (listening to) the high spiking activity of the STN.

The cause of Parkinson disease

The Substantia Nigra, another brain basal ganglia structure, produces dopamine, a neurotransmitter.  In Parkinson Disease patients the cells of the Substantia Nigra degenerate.

In a healthy person there is a good balance between excitation and inhibition on the STN.

In Parkinson, the diminished dopamine releases the inhibition on the STN, which starts firing in an uncontrolled manner.

We formulated the following research questions: 


How well can you see the STN on 3T MRI and CT scans?


Can you subdivide the STN in its functional parts?

structural connectivity

How is the STN structurally (with fiber connections) connected to its surroundings?

functional connectivity

How is the STN functionally (correlated activity) connected to its surroundings?

Q2.Can you subdivide the STN in its functional parts?

Answer 2:

9.4T animal scanner

We hypothesized that this diffusion might be different in the different parts of the STN due to possible differences in cell shapes and local connections. We studied this in the rat (where the STN has two subdivisions) with an excised brain in a small-bore 9.4 Tesla Bruker-Biospec AVANCE-III MRI scanner.

It turned out that the new DTI technique with many more (132) orientations, called High Angular Resolution Diffusion Imaging, gave good results. With a new geometric separation technique (Sobolov norm) we could much better discriminate the subdivisions than with conventional separation methods.

Separation by glyph shape

Diffusion glyphs in the STN region with anatomical context. (a) Normalized Q-ball glyphs in the STN region (from Figure 5.3(b)), with atlas overlay based on [240] (ic = internal capsule, ns = nigrostriatal bundle, stn = subthalamic nucleus, zid = dorsal zona incerta, ziv = ventral zona incerta). (b) Supposed subdivision of rat STN into the large lateral motor part (motor) and the smaller medial associative and limbic part (a/l), indicated by the dashed line.

The dotted line in the lower figure indicates the subdivision.

Examples of the subdivisions of the left (A,C,E) and right (B,D,F) subthalamic nuclei of three subjects into a limbic (red), associative (green), and motor (blue) zone. Intermediate colors show overlap between the motor and associative zones (light blue) and between the associative and limbic zones (brown).

 Examples of the electrode position in five STNs as defined during the post-operative programming. The active contacts (red) lie within the pre-operatively computed motor areas.

Guest visit to the University of Minnesota

From February till August 2015 Birgit visited prof. Noam Harel, at the Center for Magnetic Resonance Research (CMRR) of the University of Minnesota. Prof. Harel is known for developing high-field fMRI capabilities for mapping columnar and laminar organization in cerebral cortex both in human and animal models. 

Fiber tracks of the SNc and SNr.
(A,B) Fiber direction within the SNc (A) and SNr (B) color coded for orientation. 

(C) Fibers tracked between the GPi and SNr (brown tracks) and SNc (green tracks). (D) Fibers tracked between the GPe and SNr (orange tracks) and SNc (blue tracks). A, anterior; L, lateral; M, medial; P,

Birgit Plantinga defended her TU/e BMIA PhD thesis on 17-11-2016 in Eindhoven.

Back to focus area
computer-aided diagnosis