**3. Field Strength**

Pre-operative MRIs are obtained to both visualize the DBS target and to assess for potential comorbidity and identify venous architecture to ensure a safe entry route for surgery. The quality of MRI is dependent on a large number of factors. One of these factors is the signal-to-noise ratio (SNR), which is strongly influenced by field strength (Tesla or T for short) (see Figure 1) [24,49,50]. SNR can be defined as the di fference in signal intensity, e ffectively determining the amount of signal that represents the true anatomy compared to noise and random variation [51,52]. Low-field MRIs such as 1.5 or 3 T are routinely used for DBS targeting. However, recently, an ultra-high field (UHF) 7 T MRI system has been approved for medical neuroimaging [53]. Compared to 7 T, 1.5 and 3 T MRI tend to su ffer from both inherently lower SNR and low contrast-to-noise (CNR). CNR reflects the di fference in SNR between di fferent tissue types, which is therefore essential for specificity [54,55]. Moreover, the STN is an inherently di fficult structure to visualize as it is a small structure located within a very deep and dense portion of the basal ganglia and is surrounded by structures containing similar chemical compositions. This is exemplified by vast inconsistencies in observed volumetric measures, size, and location estimates of subcortical nuclei reported at low field strengths [44–46].

**Figure 1.** Visualizing deep brain stimulation (DBS) targets with different magnetic resonance imaging (MRI) field strengths (adapted from [24] illustrating DBS targets across field strengths, requiring different contrasts. We obtained 1.5 T images from a 52-year-old male Parkinson's disease patient at the Maastricht University Medical Center (MUMC). Clinical 3 T and 7 T images were obtained from a from 57-year-old male Parkinson's disease patient at the Maastricht University Medical Center (MUMC), and the optimized 3 T images were obtained from a healthy male age-matched subject at the Spinoza Center for Neuroimaging, Amsterdam. All images are shown in the axial plane and are present in their native space with no post-processing to replicate the visualization of each nucleus as performed on neurosurgical planning software. The T1 contrasts show the anterior thalamic nucleus and nucleus accumbens at all field strengths. The subthalamic nucleus and globus pallidus (GP) are shown with a T2 contrast at 1.5 T and clinical 3 T scan. Note that in the 7 T contrast, the medial medullary lamina is visible, allowing us to distinguish between the internal and external segmen<sup>t</sup> of the GP. For optimized 3 T and 7 T, the STN and GP are shown with a T2\* contrast. The acquisition times (TA) for each scan are included to highlight the fact that optimized 3 T can provide high-quality images similar to those at 7 T but take nearly twice as long to obtain. While the STN and GP are visible in both 3 T images, the contrast and sharpness of borders increases at 7 T.

The quality of the magnetic field is also determined by magnetic field gradients. MRI gradients are characterized by the change in the magnetic field as a function of distance. The MRI gradient arises from gradient coils, which are a set of electromagnetic components within the scanner that are used to control the magnetic field [56,57]. Weaker gradients arising from lower magnetic fields cause g-factor penalties, whereby an inhomogeneous B1 field causes artificial signal differences and noise amplification in tissues further from the coil in the subcortex at 3 T compared with 7 T MRI [58,59]. SNR is therefore lower in subcortical structures relative to the cortex due to the larger distance between the center of the brain and receiver coil elements. These differences are amplified at low field compared to UHF [60–62].

However, SNR scales supra-linearly with the static magnetic field, with up to a sixfold increase at 7 T compared to 3 T MRI [54,55]. This means that UHF-MRI can provide better quality images at a higher spatial resolution, increased contrast, and shorter acquisition times [51,63,64]. Reduced acquisition is essential, as clinical radiologists are often under strict time pressures that are intrinsically linked to value-based healthcare systems and cost-e ffectiveness rather than scientific value [65]. Numerous empirical studies and reviews have noted the advantages of utilizing UHF-MRI in clinical settings, performing direct comparisons between low- and high-field strengths for visualizing finer details of smaller nuclei, which are common targets for DBS [34,52,66–70].

Developments in array coil designs and parallel imaging techniques have resulted in the possibility to measure specific portions of tissue simultaneously. The simultaneous measurement increases SNR by a factor of 3 to 10 when compared to standard volume coils used at clinical field strengths, which are unable to selectively excite separate portions of tissue [60,63]. This is discussed in more detail later in the paper.

Importantly, there are caveats with regards to the implementation of UHF-MRI. Firstly, the Siemens 7 T MAGNETOM Terra is the only UHF-MR system to have obtained Food and Drug Administration (FDA) 510(k) clearance for clinical neuroradiology, and other applications of 7 T MRI are therefore considered experimental. Expense and accessibility is among the most important and most time-limiting factors in implementing UHF-MRI into clinical settings; less than one hundred 7 T systems exist worldwide, making up about 0.2% of all MRI systems [24,71]. Moreover, increased specific absorption rates (SAR), field inhomogeneities, local signal intensity variations, and signal dropout are factors that can reduce the benefits of 7 T MRI when not properly accounted for [72]. These can be countered with optimized shimming and pre-processing techniques such as bias field correction. However, these techniques require expertise that is not typically available within clinical settings [73–75].

### **4. Current Procedures for Intra- and Post-Operative Verification with Microelectrode Recordings**

Current standard practices within The Netherlands includes both pre-operative planning with neuroimaging methods and intra-operative verification with microelectrode recordings (MER). In this case, once the target has been decided, the DBS system will be implanted in two steps. First, the surgeon will create a burr hole in the skull on both hemispheres. If microelectrode recordings (MER) are used, the MER leads will be inserted into predefined coordinates. In 0.5 to 2 mm intervals from around 10 mm above the target coordinate, MER will start recording activity through macrostimulation. Multiple MERs may be placed into the STN at around 2 mm apart within the anterior, posterior, central, medial, and lateral portions. The MER lead that outputs consistent oscillations of beta bursts that are indicative of STN activity will be selected for test stimulation and subsequent implantation. If the patient is awake, additional intraoperative behavioral testing may be performed to assess the therapeutic e ffect of specific stimulation programs. Once the target has been verified via intra-operative neuroimaging (CT or ultra-low field MRI), the leads will be permanently implanted and then connected to a cortical grid and a stimulator will be inserted under the chest [76–79].

Not all centers use pre-operative CT or MRI and instead rely on standard coordinates with MER verification (and vice versa). There are reports that sugges<sup>t</sup> MER significantly improves DBS outcomes [80], and that MER fails to show any significant benefit compared to direct targeting [81]. Moreover, there remains a mismatch of around 20% in the planned target coordinate based on MRI, compared to the actual optimal location identified with MER when using 1.5 and3T[82,83]. Further, the use of intra-operative ultra-low field MRI for identification of the test leads during surgery has shown to be as e ffective as MER in improving post-operative motor symptoms [84]. Moreover, while not a strictly scientific issue, the application of MER more than doubles the cost of a bilateral STN surgery [85]. See [86] for an extensive overview on comparisons between MER and MRI for STN identification in PD.

Lastly, post-operative managemen<sup>t</sup> requires the identification of optimal stimulation parameters. These parameters can vary per patient, and some patients may require DBS in combination with medication. Microlesioning e ffects and acute foreign body reactions can impact the homeostasis of STN function and lead to a misinterpretation of DBS efficacy. Therefore the patient should ideally be assessed several times at different stages after the surgery [87]. Baseline motor function is initially obtained after total withdrawal of dopaminergic medication [88]. Axial motor symptoms such as bradykinesia, rigidity, stability, gait, posture, and dysarthria are assessed with rating scales such as the Unified Parkinson's Disease Rating Scale Part 3 (UPDRS, III) or Movement Disorders Society (MDS)-UPDRS [79,89]. As the DBS lead consists of multiple contact points, each point is tested separately through monopolar stimulation, beginning with a standard frequency of 130 Hz and pulse width of 60 μs [90]. Amplitudes are varied in a step-wise manner and the lowest amplitude that results in the highest suppression of clinical symptoms with the absence of sustained adverse effects will be chosen as the optimal stimulation parameters [27]. More in-depth literature on practices for post-operative verification, stimulation programming, and care can be found in [91–93] and the references therein.
