Purpose: Magnetic resonance imaging developments for optimal targeting of stereotactic surgical approaches for tremor have yielded white matter attenuating sequences (FGATIR/FLAWS) directly showing a targetable hyperintensity in the subthalamic region (rubral wing, RW). The RW has been reported to coincide with the dentato-rubro-thalamic tract (DRT) without determining its exact portion (crossed, DRTx or uncrossed, DRTu). RW discernibility on the single subject level might be hampered due to low signal-to-noise ratio (SNR), potentially interfering with surgical outcomes. Methods: We performed manual delineations of RW on FLAWS sequences in native space (n=77, 3 raters) and warped results into MNI 152. From this tractographic analyses of DRT (human connectome project body, n=1000) were carried out, using the red nuclei and RW as waypoints. In another approach, we investigated fiber tightness peaks of DRTX vs DRTu portions along their z-axis. Results: Identification of RW was possible in all subjects. DICE coefficient for volumetric comparison was rather low with 0.54. Euclidean distance of the RW center of gravity (COG) between raters was <2mm. Tractographically, RW represents an optimal waypoint to define DRTx with optimal constraining fibers to the ipsilateral precentral gyrus (PCG). A peak for fibers descending from PCG (DRTx only) was found to coincide with RW. Conclusions: COG can be determined with good accuracy, but full volumetric appreciation is difficult on the single subject level (SNR), potentially necessitating additional DRT tractography for surgical targeting. In the light of this connectomic study, the RW‘s role as a valid and individually visualizable tremor target for DBS and SLS is strengthened. In a way, the application of the FLAWS/FGATIR delineation of RW would potentially allow surgeons to rely on a more generic personalized MRI grey sequence based and geometrically accurate targeting for tremor without a cumbersome use of the more demanding DWI technology. However, the volumetric interpretation of the structure might not be unequivocal (especially in its anterior-posterior extensions) and care must be taken to readily identify the structure on a single case basis. This potentially necessitates the further use of adjunct imaging modalities (DWI) in cases of a low quality RW depiction with FLAWS/FGATIR to avoid detrimental side effects.

Millimetric precision is fundamental to functional stereotactic neurosurgery and research. This relies on accurate identification of deep brain targets, which in turn often requires time-intensive manual segmentation, machine learning generalisability, and atlas limitations. We introduce RegiNet, our novel deep learning framework, addressing three critical issues. First, rapidly and robustly generating patient-specific deep brain nuclei segmentations across varied magnetic field strengths and MRI protocols. RegiNet, integrating subject-specific anatomical priors into a transformer-based model, demonstrates high fidelity for the subthalamic nucleus (STN) (Dice Similarity Coefficient (DSC) 0.94, 95th percentile Hausdorff Distance (HD95) 1.64 mm). This performance significantly outperformed three leading academic and industry segmentation algorithms by >28% on exclusively clinical STN-DBS scans (N=120). For thalamic nuclei—Ventral intermediate (Vim), Ventro-oral (Vo), and Ventro-posterior (Vp)—using multimodal inputs like fractional anisotropy maps, RegiNet achieved a mean DSC of 0.87. Crucially, HD95 values were consistently below the 2.0 mm surgical accuracy threshold (mean HD95: Vim ~1.36 mm, Vo ~1.60 mm, Vp ~1.62 mm). Second, domain shift, which degrades segmentation model performance across imaging centers, slowing adoption and necessitating burdensome, site-specific retraining. RegiNet’s architecture demonstrates inherent cross-scanner/protocol robustness. In leave-one-out experiments testing adaptability to unseen field strengths (1.5T, 3T, 7T), RegiNet retained an average DSC performance of 82% and never produced null predictions, unlike baseline models which showed significant degradation and failures (e.g., SwinUNETR 72% retention with 10 nulls, DINTS 49% with 16 nulls). On completely unseen external data (N=476), RegiNet achieved 86% performance retention, starkly contrasting with SwinUNETR (48% retention, 26% failure rate). This was evidenced by successful segmentation from T1w/T2w and fractional anisotropy maps, maintaining performance across diverse scanner vendors. Our methods may offer rapid adaptation to site-specific protocols without extensive manual annotation, reducing retraining needs. Third, the inadequacy of standard atlases (e.g., HCP-1065 from young healthy adults) for patients with disparate neuroanatomy, like older individuals with thalamic atrophy. Comparing RegiNet-derived Vim segmentations in our Parkinson's disease tremor (PDT) and Essential Tremor (ET) patients (N=45) against HCP-1065 templates (coordinates relative to AC-PC, Asymmetric MNI 2009b space) revealed substantial differences. RegiNet left Vim centroid (X=-13.62, Y=-18.43, Z=-2.38 mm) versus HCP-1065 (X=-14.36, Y=-18.20, Z=-1.58 mm) showed 1.11 mm displacement and 2.12 mm boundary disagreement (HD95). Right Vim showed 1.28 mm centroid displacement (RegiNet: X=13.12, Y=-17.56, Z=-2.56 mm; HCP-1065: X=14.14, Y=-17.40, Z=-1.80 mm) and 2.87 mm boundary disagreement. Our population-specific Vim locations were consistently more posterior/inferior, with >2 mm boundary disagreements, quantifying standard atlas mismatch. These results quantify potential systematic errors in normalised diffusion/quantitative MRI where patient-specific scans can be scarce. RegiNet offers robust, precise, subject-specific analysis, pivotal for advancing large cohort studies where accurate individual segmentations are often infeasible.

Objectives: To compare the performance of Brainlab’s commercially approved planning software with the newly developed CranioPass software in stereotactic neurosurgical planning for tremor treatment in patients unresponsive to medication and eligible for Deep Brain Stimulation (DBS). Methods: Twelve patients (8 male, 4 female; mean age 62.8 ± 11.9 years) with a mean tremor duration of 7.5 ± 5.5 years (range: 4–22) were included. Diagnoses included secondary Parkinson’s syndrome (6), essential tremor (2), dystonic tremor (2), Parkinson’s disease (1), and post-thalamic hemorrhage (1). Nine patients underwent bilateral DBS lead implantation; three had right-sided implantation. Electrodes were implanted in GPi (1 patient), STN (1), and Vim-PSA (10). Six patients received the Infinity 7 system (Abbott, USA) and five received the Vercise Genus system (Boston Scientific, USA). Postoperative CT scans were performed one day after surgery and fused with preoperative stereotactic CT for target verification. Target coordinates were calculated using both Brainlab and CranioPass software and compared using the Riechert-Mundinger (RM) target point simulator. Lead tip positions on CT were compared with planned coordinates. Tremor was assessed using the Fahn-Tolosa-Marin Tremor Rating Scale (FTMTRS) at baseline, 4 weeks, and 16 weeks postoperatively. Results: Among 21 implantations (12 left, 9 right), coordinate differences between the two software platforms were minimal. A 0.1 mm difference in the Z-axis was observed in one case. Arc settings were identical in 13 cases; A 0.1-degree difference has been observed in one parameter in 3 settings on the left and in 3 settings on the right side. In two values (0.2 and 0.1 degree) in 1 setting, and 0.1 and 0.1 degree in 1 setting. The average deviation between planned and actual lead tip positions was 0.7 ± 0.6 mm (left: 0.6 ± 0.7 mm; right: 0.8 ± 0.6 mm), which falls within the voxel size of the imaging datasets. FTMTRS scores showed an 87.5% improvement at 16 weeks, confirming clinical effectiveness. Conclusions: CranioPass demonstrated high accuracy and consistency compared to Brainlab software in stereotactic planning for DBS. Surgeries based on CranioPass planning were successfully completed with no adverse events. These results suggest CranioPass is a reliable alternative for stereotactic DBS planning, supporting its potential for broader clinical use.

Intact cerebral microcirculation is a key function for vital neural tissue. The cerebral microcirculation can be measured with laser Doppler flowmetry (LDF). Our research group has adapted an LDF system for use in relation to neurosurgical intervention that can investigate the microcirculation and act as a “vessel alarm” during stereotactic and functional neurosurgery. The aim is to give an overview of the system and give examples of applications already implemented and to suggest potential future applications. The system comprises a LDF module (PF5010/PF6010, Perimed AB, Sweden) with an inhouse developed software (Labview/MatLab) for data collection, storage and real-time visualization of tissue Perfusion (microcirculation) and TLI (total light intensity, grey-whiteness). A set of optical probes have been designed. Probes are available for stereotactic guidance and creation of a trajectory in relation to deep brain stimulation (DBS) implantations, and stereotactic and neuronavigated brain tumor biopsies. In the latter set up the probes are adjusted to the inner cannula of a biopsy kit modified with an opening at the tip. The opening allows forward looking measurements in tissue not yet touched by the probe. The system has been used in more than 130 DBS implantations and 50 brain tumor biopsies. As all probes are “forward looking” the microcirculation is recorded about 1 mm beyond the probe tip. Data is presented in real-time on a monitor during surgery. If necessary, e.g. due to very high Perfusion, the trajectory can be changed to avoid a hemorrhage. As the probe is securely fastened to the stereotactic device or neuronavigational system potential movement artifacts are reduced to a minimum. In addition, the TLI signal acts as an in-situ tracer of tissue grey-whiteness and can together with the Perfusion signal support in intraoperative guidance through brain structures during insertion of the probe. The system has a resolution of 0.5 mm, which makes it possible to detect changes in grey-whiteness between e.g. thin laminae in the pallidum and thalamus. Additional applications not yet explored are cell line implantations, drug administration, RF-lesioning and guidance during LITT. In principle all types of stereotactic interventions are candidates for use of the LDF-system. In conclusion, micocirculation measurements can together with grey-white matter identification help the surgeon avoid vessels along a trajectory. It can also be used to study cerebral microcirculation in healthy and diseased tissue in relation to stereotactic and functional neurosurgery.

Background In Parkinson's disease (PD) patients, modulation of the fibre tracts of the cortico-basal ganglia-thalamo-cortical loop is the presumed mechanism of action of deep brain stimulation (DBS) of the subthalamic nucleus (STN). Therefore, we explored patient-individual cortical structural connectivity of the volume of tissue activated (VTA), as well as DBS-induced modulation of fibre tracts connecting the STN with cortical and sub-cortical nodes and their correlation with therapeutic effects. Patients and Methods A retrospective cohort of n = 69 PD patients treated with bilateral DBS of the STN was analysed. Clinical response was assessed from the DBS-induced change in the UPDRS-III motor scores (total and symptom-specific sub-scores for tremor, rigidity, and bradykinesia) under regular medication after a median follow-up of 9.0 (range 2.6 – 20.2) months. Tractography based on patient-individual diffusion-weighted MRI was employed in two ways. Firstly, whole brain tractography was used to identify the cortical connections of fibres passing the VTAs. Then, reconstruction of specific white matter pathways of the motor loop connecting the STN with the basal ganglia and cortex (informed by the regions obtained by the first analysis) were used to identify the proportion of fibres within these pathways which was modulated by STN-DBS. This proportion of pathway modulation was used in a correlative analysis with clinical outcomes. Results Streamlines traversing the VTAs were primarily connected to the supplementary motor area (SMA), pre-SMA and the premotor cortex. Streamlines from both primary motor (M1) and sensory (S1) cortices were also identified, but to a much lesser degree. Within the pathways connecting the STN with the cortical and subcortical nodes, on average 30-40% (range 10-80%) of the fibres were modulated by STN-DBS. This proportion correlated significantly with the percentage change in UPDRS motor score for fibres connecting the STN to cortical regions like the SMA (ρ=0.28), pre-SMA (ρ=0.26), and ventral and dorsal pre-motor cortices (ρ=0.26 and ρ=0.29, respectively). Interestingly, modulation of the pathways between the STN and the globus pallidus externus (GPe, ρ=0.26) and internus (GPi, ρ=0.29) also showed significant correlation to UPDRS motor improvement. Finally, good clinical responses for both tremor and rigidity were associated with a significantly (p < 0.05) higher proportion of modulated fibres for the same cortico- and sub-cortico-STN connections. Conclusions Patient-individual tractography reveals that, in PD, most of the cortical fibres traversing the VTA are connected to the SMA. In addition, clinical efficacy is related to the proportion of DBS-affected fibres connecting the STN with nodes of both the hyperdirect (cortex-STN) and the indirect pathways (STN-basal ganglia). As such, patient-specific tractography, in particular of the basal ganglia, could be used in a clinical context as a tool to guide therapy. Funding: Funds have been provided by the European Joint Programme Neurodegenerative Disease Research (JPND) 2020 call “Novel imaging and brain stimulation methods and technologies related to Neurodegenerative Diseases” for the Neuripides project ‘Neurofeedback for self-stImulation of the brain as therapy for ParkInson Disease’.

Background: Convection-enhanced drug delivery (CED) for Parkinson's disease (PD) is costly, and current methods lack precision, often targeting the entire putamen, leading to the potentially inefficient use of resources. Our study addresses this by exploring a more targeted approach for drug delivery, which could reduce treatment costs by focusing therapy on specific regions of putamen. By optimizing drug delivery, we aim to make treatments more cost-effective without compromising efficacy. Methods: Twenty PD patients underwent diffusion-weighted imaging (DWI) to visualize the structural connectivity within the brain. A commercial subcortical auto-segmentation tool was used to define the putamen as well as the amygdala, the STN, and the cerebellum. Utilizing the Julich Brain Atlas, nine cortical regions (Brodmann areas 44, 45, 3a/b, 4a/p, pre-SMA, SMA, and insula) were semi-automatically segmented. Structural connectomes were analyzed through tractography, allowing for the parcellation of the putamen into four segments in relation to the anterior commissure. Two trajectories, occipital and frontal, were tested for segment coverage using stepwise injection of the therapeutic agent. A genetic algorithm was employed to simulate these injections and to compare the coverage of the target region. Results: Tractography revealed a significant projection of motor areas to the superior posterior segment of the putamen, suggesting this region as a more specific target for treating motor symptoms in PD via CED. Non-motor connections were most common in the inferior posterior segment for the amygdala and in the superior anterior segment for the insula. Both occipital and frontal trajectories were found to be equally feasible for targeting the putamen segments, with surgical feasibility varying by individual patient anatomy, and achieved comparable coverage, with no significant difference between them, highlighting the need for personalized surgical approaches. Conclusions: The application of DWI and tractography for the segmentation of the putamen by its cortical connections offers a pathway towards targeted gene therapy in PD. Identifying the posterior superior segment as the primary recipient of motor area projections and confirming the feasibility of using either the occipital or frontal trajectory underscores the flexibility in optimizing therapeutic efficacy while conserving resources. This precise targeting could potentially allow for reduced dosages and more focused treatment, minimizing exposure to non-motor segments of the putamen and associated risks.

【Introduction】 In stereotactic thalamic surgery, accurate identification of the target subnuclei significantly affects therapeutic outcomes. However, the boundaries of these subnuclei are often ambiguous and challenging to visualize with conventional MRI. For tremor treatment targeting the ventral intermediate nucleus (Vim), tractography is frequently used to delineate the dentatorubrothalamic tract (DRTT) as a reference. Nonetheless, distinguishing the DRTT from the posteriorly located medial lemniscus (ML) remains difficult and reproducibility is limited. In this study, we investigated the utility of FGATIR MRI imaging as a tool to clearly differentiate the DRTT from the ML and compared the results with Klüver-Barrera-stained human brain sections. 【Methods】 We compared preoperative MRI images used for planning MRI-guided focused ultrasound (FUS) thalamotomy with previously prepared human brain specimens. MRI scans were acquired using a Siemens MAGNETOM Lumina scanner, including 3D T2 Cube, 3D SPGR, DTI, and FGATIR sequences. Human brain specimens were sectioned into continuous frozen slices at 50 μm thickness, with every 500 μm slice stained using the Klüver-Barrera method and digitized. FUS thalamotomy was then performed, and the validity of the preoperative MRI images was evaluated by postoperative MRI and clinical outcomes. 【Results】 Among the preoperative imaging modalities, FGATIR provided excellent visualization of intrathalamic fiber tracts. Postoperatively, tremor was completely resolved without any adverse events such as sensory disturbances. MRI confirmed the lesion localized to the Vim including the DRTT, while the ML remained identifiable and unaffected. 【Discussion】 In tremor treatments targeting the Vim, the therapeutic "sweet spot" often lies near the boundary with the posterior ML, increasing the risk of sensory side effects when seeking maximal efficacy. For FUS, where microelectrode recording cannot be performed, visualizing the ML boundary is particularly important. FGATIR imaging may provide more accurate visualization of the ML within the thalamus compared to tractography.

Introduction: Deep brain stimulation (DBS) is a therapy for neurological disorders, with accurate electrode placement being key to its success. Poor electrode placement accounts for up to 46% of treatment failures.[1] Most DBS workflows use co-registration (‘fusion’) of preoperative MRI with perioperative CT for planning and evaluating electrode placement. However, inherent differences between CT and MRI present challenges to image registration algorithms. While CT provides excellent bone contrast, MRI renders bone tissues as nonspecific low signal structures in most sequences.[2] Moreover, while intraoperative CT has increased surgical efficiency, its lack of soft tissue contrast results in co-registration based exclusively on bone signal, compounding this problem.[3] Zero echo-time (zTE) imaging, a novel MR technique, enables precise reconstruction of cortical bone.[4] zTE achieves this via efficient capture of short-lived bone signal and uniform soft-tissue response. Here, we present an optimised zTE sequence (with preserved brain tissue contrast), applied to our DBS surgical workflow. Methods: We optimised and implemented a brain zTE sequence in our preoperative DBS protocol (Figure 1). 25 consecutive patients who underwent DBS following the novel imaging protocol were included (targets: STN, GPi, VIM, PPN). For initial validation, CT to T1-MPRAGE co-registration was performed in FSL-FLIRT[5] (6 degrees-of-freedom (DOF), normalised mutual information (nMI) cost function) with- and without the zTE image as an intermediate step i.e. CT-T1 and CT-(zTE)-T1, Figure 2. To directly assess impact on our clinical workflow, two separate stereotactic plans were created per patient using Renishaw NeuroInspire™: with either T1-MPRAGE or zTE as the base sequence. This generated similar CT-T1 and CT-(zTE)-T1 co-registrations (6DOF, nMI).[6] Registration accuracy was evaluated by: a) Extraction and quantification of registration and cost function matrices (CT-T1 vs. CT-(zTE)-T1) b) Calculation of Dice similarity coefficients of overlap between segmented skull structures (CT-T1 vs. CT-(zTE)-T1) c) Blinded anatomical landmark-based assessment by an Attending Neurosurgeon Results: a) The root mean square difference between CT-T1 and CT-(zTE)-T1 registration matrices across the cohort was (i) 1.10mm (95% CI 0.88 – 1.32, p<0.0001, two-tailed t-test) in the clinical workflow (Renishaw NeuroInspire) and (ii) 0.77 mm (95% CI 0.64 – 0.91, p<0.0001, two-tailed t-test) in the FSL-FLIRT pipeline. Euclidean errors for electrode contacts tips on postoperative imaging were similar in magnitude (p<0.05). b) Dice similarity coefficients for segmented skull structure overlap were greater for CT-(zTE)-T1 vs CT-T1 across the cohort (p<0.05). c) In the blinded qualitative expert evaluation, CT-(zTE)-T1 outperformed CT-T1 in 15/23 patients, with no discernible difference in the remainder. There were no instances of CT-T1 outperforming CT-(zTE)-T1. Discussion and Conclusions: We optimized a novel MRI sequence with clear delineation of bony anatomy and neuraxial structures (Figure 1). The addition of zTE to our DBS surgical workflow has resulted in a statistically significant difference in CT-MR co-registration, with improvements in registration accuracy. The clinical significance of this difference is currently under prospective investigation. While <2mm accuracy is considered acceptable in the movement disorders literature,[7] optimising image co-registration is an important component of reducing planning error and more accurately evaluating DBS targeting postoperatively. A limitation of this study is the lack of ground-truth structural information, leading to the use of Dice overlap coefficients and qualitative expert evaluations. An ongoing study with MR/CT fiducials and phantoms seeks to address this (to be discussed). Finally, we will discuss current development work on postoperative zTE, which can better localize DBS electrodes than existing MR methods (Figure 3). This could be utilised in postoperative MRI protocols. References: 1. Okun MS et al. (2005) Arch Neurol. PMID: 15956104 2. Florkow MC et al. (2022) J Magn Reson Imaging. PMID: 35044717 3. Kremer NI et al. (2019) Neuromodulation. PMID: 30629330 4. Wiesinger F, Ho ML (2022) Br J Radiol. PMID: 35616709 5. Jenkinson M et al. (2012) Neuroimage. PMID: 21979382 6. Geevarghese R et al. (2016) Stereotact Funct Neurosurg. PMID: 27318464 7. Kremer NI et al. (2023) J Neurol Neurosurg Psychiatry. PMID: 36207065

Background: Tractography is a non-invasive method for reconstructing and visualizing white matter tracts based on MRI imaging. The use of tractography as a clinical and research tool is increasingly growing, presenting promising results for targeting and improving neuromodulation procedures outcomes. Moreover, white matter characterization plays a key role in exploring and understanding normal populations and brain disorders, while multiple macroscopic and microscopic features of the tracts can be extracted. However, to date, while different tracts may contain thousands of fibers, the common approaches to tract parameters analysis are summarizing the tract with a single mean value or performing 1D along-tract analysis, which may result in loss-of-data and minimize its clinical value. Methods: 10 MRI scans of healthy subjects from the Human Connectome Project database, and 4 from Rambam Health Care Campus were acquired. Utilizing the DTI sequence, a tractography analysis of six tracts was performed for each subject bilaterally. An algorithm written in Matlab™ was built for tract characterization and inter-subject comparison of the tracts. The algorithm performs a 3D analysis of a tract while using its spatial and geometrical information to gather and integrate data. It utilizes the anatomical position along the tract, its morphology, and the angle around it to calculate and highlight the tract characteristics in a 3D fashion. The algorithm was applied to various tracts (motor and limbic) that highly differ in morphology, and statistical analysis was performed. Results: A method for 3D or 2D visualization of white matter tracts geometrical and microstructural features, generating 616 different maps. Utilizing the method for fractional anisotropy analysis yielded maps with a unique pattern for each of the six tracts, while preserving high inter-patient reproducibility. This method provides a comprehensive observation of a tract and its surrounding structures, an inter-subject comparison of different tracts through various parameters, and a tool for abnormalities detection, while projecting the results on the patient or MNI space anatomy. Conclusions: We present an innovative tool for analyzing white matter tracts. This tool may contribute to the research of white matter tracts, the understanding of various disorders, the detection of biomarkers, targeting, and even diagnosis.

Deep brain stimulation (DBS) has become essential to treat movement disorders. As technology evolves for an ever-finer modulation of pathological networks, increasing numbers of lead contacts and directionality complexify manual programming. Therefore, automated algorithms are being evaluated to predict DBS parameters based on probabilistic sweet spots. As a preamble, such software requires accurate co-registration between pre- and postoperative images, reliable brain volume normalization, brain shift correction and precise electrode detection. This computational workflow is already embedded in tools dedicated to DBS research – such as Lead-DBS. However, its robustness between distinct postoperative CT scans at the individual level has not yet been assessed systematically. To test the robustness of the image-processing workflow applied in Lead-DBS, we identified 34 retrospective patients (68 hemispheres) with Parkinson’s Disease (PD) or Essential Tremor who were implanted with DBS electrodes and received two distinct postoperative CT scans. Each of these two scans was processed independently in Lead-DBS (v3.1) for electrode reconstruction, using the same preoperative MRI as reference. At the individual level, the computed coordinates of the lead tip were compared between image sets, as well as the resulting volumes of tissue activated (VTA) based on patients’ effective stimulation parameters. At the group level for PD patients, we computed and compared probabilistic maps of clinical improvement, based on the first or second postoperative CT scan respectively, using the same clinical data. Between image sets, lead tip coordinates in the normalized space showed a mean translation of 0.78mm (min 0.21mm, max 1.70mm). In our dataset, the robustness of lead reconstruction was not significantly influenced by pneumocephalus, as we compared subgroups with or without intracranial air on the first postoperative CT. No significant correlation could be drawn between the extent of lead translation and the pneumocephalus volume. To assess the relevance of up to 1.70mm lead tip translation, VTAs were computed and compared between postoperative image sets. The mean Dice index was 0.73 (min 0.33, max 0.94). In the twenty hemispheres with lead tip translation of 1mm or more, the Dice index was systematically below 0.75. At the group level of PD patients, the two computed probabilistic maps were overlapping without significant discrepancy. From this robustness study, we conclude that brain shift correction algorithms are reliable, as no significant lead translation could be imputed to pneumocephalus. As VTA computation varies in a mean range of 27% at the individual level, the current co-registration, normalization and lead detection workflow still requires some improvement to serve as a robust base for automated DBS parameter prediction. However, the variability of lead reconstruction disappears at the group level and the current image processing workflow seems sufficient to compute probabilistic maps of clinical improvement – and therefore sweet spots. Future studies with postoperative MRI instead of CT may show better robustness of lead reconstruction at the individual level.

Objective: Although brain shifts can occur during deep brain stimulation (DBS) electrode implantation in the subthalamic nucleus (STN), the underlying causes and predictive factors for these shifts remain unclear. In this study, we utilized an index derived from X-ray films and software analysis to quantify the degree and direction of changes in the implanted electrodes relative to the STN and investigated potential clinical factors associated with these shifts. Methods: We analyzed 42 DBS electrodes implanted in the STN of 21 individuals with Parkinson’s disease. Electrode tip displacement was evaluated using Elements (BRAINLAB®️) software based on computed tomography (CT) scans obtained immediately after implantation (1st CT) and after complete dissipation of intracranial air (2nd CT), comparing the two images (2nd CT – 1st CT). We used “Angle A,” measured from the anterior and posterior commissure line and horizontal plane on X-ray films, along with intracranial air volume obtained from Elements software, as predictive factors for brain shifts. Results: The x-coordinate of the electrode tip shifted significantly in the lateral direction (0.4 ± 1.12 mm, p = 0.025), while the y- and z-coordinates shifted significantly in the anterior (1.35 ± 0.84 mm, p < 0.0001) and caudal (0.54 ± 0.63 mm, p < 0.0001) directions, respectively. The displacement in the y-coordinate significantly correlated with both intracranial air volume (p = 0.001, R = 0.476) and Angle A (p = 0.047, R = -0.308). The z-coordinate displacement correlated with intracranial air volume (p = 0.049, R = 0.305), although this correlation had limited statistical power. The x-coordinate displacement did not correlate with any assessed factors. Conclusion: Electrodes implanted in the STN shifted laterally, anteriorly, and caudally following the resolution of intracranial air. Minimizing intracranial air entry through burr holes and maintaining the head position closer to the horizontal plane may help mitigate brain shifts during STN DBS implantation.

INTRODUCTION Preoperative and intraoperative identification of eloquent areas near supratentorial lesions is standard practice in pediatric patients, particularly for motor and language mapping. Non-invasive tools such as transcranial magnetic stimulation (TMS) and DTI-based fiber tracking enhance surgical planning, while intraoperative techniques like neuromonitoring, neuronavigation, and intraoperative MRI support safe resections. This abstract presents the role of motor mapping with navigated TMS (Nexstim®) at Hospital Sant Joan de Déu (Barcelona), with results integrated into the surgical planning platform and used intraoperatively (Fig. 1). MATERIALS & METHODS Patients with supratentorial lesions were prospectively recruited from the surgical program. All underwent a diagnostic 3T MRI, with additional studies (fMRI, DTI, and nTMS) ordered based on lesion location and clinical presentation. nTMS is a non-invasive technique using magnetic stimulation with image-guided navigation to map the motor cortex. Muscle responses or MEPs (Motor Evoked Potentials) were recorded via electromyography and saved as spatial coordinates. These data were imported into the Brainlab Node® platform for co-registration, segmentation, and tractography using anatomical and/or functional ROIs. The resulting 3D model was used for planning and transferred to the Brainlab Dual Curve® system for intraoperative navigation (Fig. 2). Intraoperative motor mapping was performed using direct cortical stimulation (DCS, Inomed®) over suspected motor areas. Preoperative and intraoperative data were correlated by registering intraoperative coordinates using the Brainlab® navigated pointer (Fig. 3). RESULTS Between March 2022 and January 2025, 29 pediatric patients (ages 3–18, mean age 9.87) with supratentorial lesions were prospectively enrolled. All underwent preoperative motor mapping with nTMS. 38% (n = 11) were male and 62% (n = 18) female. Most (90%) had drug-resistant epilepsy associated with structural or tumor-related lesions; 10% (n = 3) had tumor pathology without epilepsy. Lesions were in the left hemisphere in 72% (n = 21) and right in 28% (n = 8). In 25 patients (86%), the motor cortex in the lesioned hemisphere was successfully localized. In 4 cases (13.7%), mapping failed due to poor cooperation—an inherent limitation in pediatric populations. 86% (n = 25) underwent surgery: 2 hemispherotomies, 10 tumor or epilepsy resections, and 11 laser ablations. Only hemispherotomies and resections were neuromonitored, so tractography and functional data were crucial for laser procedure planning. Of the successful mappings, 76% (n = 19) were imported into Brainlab for functional CST tractography. Customized tracts were derived from 3D functional ROIs; anatomical CST tracts were also generated using anatomical motor ROIs. In 12 cases, both methods were combined. All cases with tractography were navigated intraoperatively. Postoperative outcomes were consistent with planning: 64% had no motor deficits, and 36% had temporary, anticipated deficits (e.g., hemiparesis). In 3 cases, preoperative mapping was confirmed intraoperatively via DCS (Inomed®), validating functional localization. DISCUSSION Preoperative mapping is essential in pediatric epilepsy surgery, enabling precise localization of the motor cortex to avoid damaging eloquent areas. This reduces the risk of postoperative deficits such as paralysis or weakness. It also supports individualized surgical planning, accounting for age-dependent anatomical and functional variability. Functional maps guide tailored resections or ablations, minimizing the risk to non-epileptogenic motor areas. Correlating preoperative nTMS with intraoperative DCS provides an added layer of confidence, reinforcing a function-preserving, personalized surgical approach.

Topic: Subthalamic Nucleus Deep Brain Stimulation Communication preference: Poster or oral presentation Key words: Deep brain stimulation – Trajectory planning – Automation Introduction: Accurate electrode placement is essential to achieve good clinical outcomes in subthalamic nucleus deep brain stimulation (STN-DBS) for people with Parkinson’s disease. Electrode trajectory planning is based on the visual assessment of manually planned electrode trajectories on magnetic resonance imaging (MRI) scans. To our knowledge, there is currently no algorithm available that can present an automatically planned and analysed trajectory for patients undergoing STN-DBS surgery. Our aim is to develop an algorithm that can plan a safe trajectory for a given target point faster than it would take to manually plan the trajectory. Method: The algorithm contains two parts: the selection of possible trajectories and trajectory analysis. The selection of possible trajectories is based on historical entry points (EPs) (n=231) corresponding to the T2-weighted MRI coordinate space. Historical EPs are transformed to patient-specific MRI space by a pipeline of image registration and point transformation using the ICBM 2009c nonlinear symmetric brain template. Historical EPs were first transformed to template space and after transformed to patient-specific MRI space. A cortical region of interest is defined based on the historical EPs. Samples are taken in the cortical region, and candidate EPs are adjusted to the patient’s anatomy. Four MRI features define an optimal trajectory (Fig.1): (1) T2-MRI voxel intensities, (2) contrast enhanced T1-MRI voxel intensities, (3) average distance to CSF, and (4) average T2-MRI voxel intensity along the first 7 mm from TP to EP. We hypothesize that an optimal trajectory maintains a safe margin relative to ventricles, sulci, blood vessels, and white matter hyperintensities by combining the first three features, while the fourth ensures maximal STN passage. Results: The total time to output a top 5 semi-automatically planned trajectories is about 8 minutes, including all MRI pre-processing and computational steps. Preliminary results showed that a DBS experienced neurosurgeon chose the semi-automatically planned trajectory over their manual planned one in 50% of the cases. Further improvements were integrated in the algorithm. These improvements will be validated by two neurosurgeons and results will be available at the ESSFN congress. Conclusion: We developed a novel semi-automatic algorithm by using historical surgical data, MRI-preprocessing steps and MRI analysis to obtain a set of potential trajectories for a new patient undergoing STN-DBS surgery. The algorithms output is obtained in a short timeframe which is suitable for clinical implementation. The algorithm offers a more standardized way of trajectory planning to guarantee safe patient-specific planning.

Introduction: Deep brain stimulation (DBS) is an effective therapy for the treatment of Parkinson's disease (PD). It improves the symptoms by stimulating the motor part of the subthalamic nucleus (STN). Traditionally, DBS surgery is performed using a stereotaxic frame, which allows high precision. However, it is often uncomfortable for the patients and makes the several hours surgery a big strain for them. The navigation-based, frameless NexFrame targeting system provides more tolerable surgical conditions for the patients. The strain can be further reduced if the operating time could be significantly shortened and only the minimum requiered number of test electrodes are used for the electrophysiological registration. Methods: Between October 2023 and February 2025, 6 patients underwent DBS implantation using a new surgical technique. During the procedure, bilateral electrodes were implanted simultaneously under real-time anatomical and electrophysiological control using a double NexFrame system and intraoperative 3D X-Ray (Siemens Pheno Artis). To assess the efficacy of the new technique, we determined the number of test electrodes used, duration of surgery, number of days in hospital and the reduction in levo-dopa dose and improvement in UPDRS III score six months after surgery. Results: An average of 3 test electrodes (1.5 per side) were used per patient (5 in the first case, 3 in three patients and 2 in two cases). The average duration of surgery was 3 hours 29 minutes, which included the time for two 3D X-Ray scans (10 minutes per scan) and the time for anaesthesia and draping between the two phases of surgery. For comparison, the conventional single-sided technique uses 3 test electrodes per side and the surgery takes on an average of 4 hours 58 minutes. The final electrodes were implanted in an average of 2 hours and 12 minutes. Patients were discharged from the department on the third ostoperative day. Levodopa requirement decreased from 960.16 mg to 553 mg, while UPDRS III improved from 40.16 to 19.5 points (51.44%). Dicussion: Based on our results, the simultaneous use of the double NexFrame system on both sides significantly reduces the operating time and the number of test electrodes required. Therefore it creates a safer, more comfortable and tolerable surgical situation for the patients and significantly improves the patients' symptoms.

Objectives: Responsive neurostimulation (RNS) targeting the centromedian (CM) thalamic nucleus has shown promising results for refractory epilepsy in both adult and pediatric populations. These treatment populations may have substantial differences in clinical presentation, but the relationship between these and variability in treatment response or programming strategies has not yet been addressed. Methods: We retrospectively reviewed 14 pediatric and 21 adult patients with refractory epilepsy, who underwent bilateral CM RNS implantation at Massachusetts General Hospital (MGH). Seizure burden was quantified using long episode (LE) counts, which represents presumed electrographic epileptiform activity lasting longer than 30 seconds, as detected by the device. Programming parameters were analyzed across programming epochs for up to 3 years post-stimulation initiation, using last observation carried forward (mean follow-up adults=1.7 years; peds=3.7 years). LE counts were Z-scored within periods of static detection settings in each patient. Baseline seizure burden was calculated as the average LE count over the first 21 days post-implantation. Relative reduction in seizure burden from baseline was calculated at 6, 12, 18, 24 months. Differences in relative reduction of Z-scored LE counts at each timepoint between pediatric and adult cohorts were assessed, using Welch's t-tests. Results: There was no significant difference in epilepsy syndrome distribution between pediatric and adult cohorts (FBTCS, GTC, JAE, JME, LGS/Dravet; p=0.84). However, developmental delay was more prevalent among pediatric patients compared to adults (86% vs 43%; p=0.03). Pediatric patients received higher charge density at 12 months post-stimulation initiation (2.0 uC/cm2 vs 1.2 uC/cm2; p<0.01) and had a greater relative reduction in LE counts in the first 6 months (46% vs 13%; p=0.13). Conclusion: Bilateral closed-loop stimulation of the CM region is an effective treatment option for both adult and pediatric patients with refractory epilepsy. Baseline clinical differences may influence treatment trajectories, underscoring the need for age-specific programming strategies to optimize clinical response. The significance of reduction in LE count may correlate with clinical improvement, but requires further investigation.

Over multiple studies, our team has expanded and refined our understanding of the basal temporal language area (BTLA) in patients with drug-resistant temporal lobe epilepsy. By integrating stereoelectroencephalography (SEEG) language mapping, lesion-symptom correlation techniques (e.g., voxel-based lesion-symptom mapping), and postoperative neuropsychological assessments, we have characterized how specific ventral temporal regions contribute to naming and memory functions. Our findings indicate that BTLA resection leads to a specific and early decline in naming, which, although partially transient, often remains below baseline levels over time. Similarly, postoperative verbal memory performance declines when the BTLA is involved in the surgical resection. Further, using cortico-cortical evoked potentials (CCEP), we revealed a robust and distributed basal temporal language network in the ventral temporal cortex supporting visual naming. Within this network, posterior ventral temporal regions behaved as “projectors,” whereas the fusiform gyrus functioned predominantly as an “integrator.” In contrast, for non-eloquent sites, the anterior fusiform gyrus demonstrated an opposite pattern, acting as a strong projector rather than an integrator. We also explored the surgical limits and localization of the BTLA in light of recent VLSM (voxel lesion-symptom mapping) studies, noting that maintaining a resection cavity anterior to the VLSM-defined critical region does not necessarily prevent postoperative naming decline. Lastly, our investigations extended to functionnectomes, enhancing our knowledge of BTLA connectivity and further emphasizing the importance of precise identification and preservation of language-critical regions to optimize postoperative outcomes.

Background: Deep brain stimulation of the anterior nucleus of thalamus (ANT-DBS) is a treatment option in refractory epilepsy, with a proven short and long -term efficacy. European registry -based data suggests significant differences in treatment outcomes. Different subregions of the ANT have been suggested as the most optimal targets, potentially explaining variable treatment outcomes. In addition to the existing histological data regarding the anatomical connections of ANT, a large number of white matter pathways have been described in humans using diffusion weighted imaging and tractography methods. Objective: Here we have studied the white matter pathways of different anterior thalamic nucleus subregions using probabilistic multi shell multi tissue (MSMT) constrained spherical deconvolution (CSD) tractography. Materials and methods: Thirteen healthy volunteers underwent MRI with T1 and DW images which were further processed for tractography using Mrtrix3 software. The centre, anterior, posterior and inferior aspects of the ANT were used as 2 mm spherical regions of interest (Roi) for fiber tracking. The location of responding (n=25; ≥ 50% seizure reduction) and non-responding contacts (n=37, <50% seizure reduction) in the stereotactic space from a group of 15 patients with ANT-DBS was compared to calculated fiber tracts. Results: Three main white matter pathways were identified in accordance with previous data: 1) inferior thalamic peduncle with fiber connections to amygdaloid complex; 2) anterior thalamic radiation with orbitofrontal, prefrontal and occasionally cingulate connections; and 3) posterior fiber pathway with terminations in the hippocampus, occipital cortex and parietal cortex. The connectivity pattern was similar in the anterior, centre and posterior ANT Rois, but the inferior part of ANT differed markedly with stronger connections to frontal cortical areas but with lesser connections to the inferior of posterior main fiber streams. The location of responding contacts matched with the fiber streams calculated from the anterior, centre and posterior aspects of the ANT while non-responding contacts located along the fiber stream between the inferior part of ANT and frontal cortical areas. Discussion: The differences in connectivity patterns of different ANT subregions together with existing outcome data suggest that the stimulation of all three main white matter pathways, namely inferior, anterior and posterior pathways is essential for optimal treatment outcomes.

Background: Recent findings highlight the critical role of subcortical white matter tracts (WMTs) in supporting language processing. This report focuses on WMT monitoring using invasive stereo-electroencephalography (SEEG). We present two case studies: MN, a 17-year-old adolescent, and CO, a 10-year-old child, both with non-lesional, drug-resistant focal epilepsy originating from the left frontal lobe with rapid frontal propagation. Both patients underwent invasive intracranial monitoring to identify the epileptogenic zone and preserve eloquent functional areas. Electrodes were implanted along presumed dorsal language WMTs, as identified by preoperative MRI-DTI imaging. Method: A comprehensive language assessment battery was administered before, during, and after the monitoring procedure, evaluating naming, sentence repetition, comprehension and production of syntactically complex sentences, and reading of words and pseudowords. Results: MN demonstrated intact preoperative language functions. However, during SEEG monitoring, stimulation of electrodes located along the arcuate fasciculus (AF, figure 1 in the attached appendix) produced language interference resembling a phonological loop deficit, primarily manifested as sentence repetition failures. Stimulation of electrodes along the frontal aslant tract (FAT, figure 2 in the attached appendix) resulted in speech initiation difficulties (e.g., speech arrest or prolonged hesitations) and impaired production of syntactically complex sentences—particularly object-relative clauses. In contrast, CO exhibited preoperative language impairment. Nevertheless, monitoring of the AF using a word span repetition task was feasible. Importantly, in both cases, stimulation of electrode contacts placed outside the targeted tracts resulted in negative language mapping. Similarly, stimulation of electrodes not traversing known language pathways also yielded no interference. Conclusions: These cases underscore the value of subcortical language mapping using targeted language paradigms. Addressing tracts such as the AF and FAT enhances the precision of functional mapping, helping preserve critical cognitive functions—even in children with premorbid language difficulties—while optimizing surgical outcomes. Monitoring language in pediatric patients, particularly those with premorbid intellectual impairments, presents unique challenges. These are further complicated by the nature of SEEG, which does not fully cover the cortical surface but instead samples subcortical structures. Nevertheless, our findings demonstrate that stimulation along subcortical language tracts enables effective monitoring of both the tracts and their cortical terminations. This approach allows for indirect yet reliable identification of eloquent cortical language areas, offering a valuable tool for surgical planning when conventional surface-based mapping is limited. The full presentation will include video clips and imaging data illustrating the stimulated subcortical contacts and their spatial relationship to functional areas, providing a detailed visualization of the mapping process.

Sleep spindles are oscillatory events specific to Non-Rapid Eye Movement (NREM) sleep, characterized by waxing and waning waveforms in the 10–16 Hz frequency range. They are generated in the thalamic reticular nucleus and propagated to other brain regions via thalamo-cortical circuits. Sleep spindles have been associated with cognitive functions and general intellectual ability. In addition to their role in physiological neural plasticity, spindles have also been linked to pathological off-line plasticity in both epileptic models and patients with epilepsy. Several studies suggest a connection between sleep spindles and interictal epileptic discharges (IEDs), which are considered markers of pathological neural plasticity in the central nervous system. We investigated the relationship between thalamic sleep spindles and thalamic IEDs in the anterior (ANT) and mediodorsal (MD) thalamic nuclei of epilepsy patients. Whole-night local field potentials (LFPs) from the ANT and MD were co-registered with scalp EEG/polysomnography using externalized leads in 15 pharmacoresistant, surgically non-treatable epilepsy patients undergoing deep brain stimulation protocols. DBS electrodes were localized in the ANT for 15 patients, and in the MD for 10 patients. Sleep spindles and IEDs were detected during all-night non-artifactual NREM sleep (stage 2 and 3). Both slow (~12 Hz) and fast (~14 Hz) sleep spindles were detected in the thalamic nuclei. Approximately 10% of the detected sleep spindles co-occurred with IEDs. These IED-associated spindles exhibited significantly longer durations and a broadband increase in thalamic and cortical activity, both below and above the spindle frequency range, compared to typical spindles. Additionally, the density of IEDs in the MD showed a positive correlation with the number of years since epilepsy onset. These findings suggest a role for the human ANT and MD in both physiological and pathological sleep spindle-related neural plasticity.

Objective: Drug-resistant epilepsy (DRE) remains a significant clinical challenge, with approximately 30% of patients failing to achieve seizure control with anti-seizure medications (ASMs). Deep brain stimulation of the anterior nucleus of the thalamus (ANT-DBS) has emerged as an effective neuromodulatory treatment for DRE, yet a subset of patients remains refractory, with only a small portion achieving seizure freedom. Cenobamate (CNB), a novel ASM with GABA-A receptor modulation and persistent sodium current inhibition, has demonstrated promising efficacy in seizure control even in ultra-resistant epilepsy. This study evaluates the potential synergy between cenobamate and ANT-DBS in patients with ultra-refractory focal epilepsy. Methods: A retrospective analysis was conducted on 44 patients who underwent ANT-DBS for drug-resistant focal epilepsy at Tampere University Hospital, Finland. Among these, 29 patients were initiated on cenobamate therapy, with 27 continuing treatment for a median follow-up time of 18 months. Treatment responses (≥50%-74% and 75%-99% seizure reduction and seizure freedom rates) were analyzed in the total cohort, as well as in the CNB and non-CNB subgroups. Safety and tolerability of cenobamate in ANT-DBS patients were also assessed. Results: Before cenobamate initiation, 61% of all included patients were responders to ANT-DBS, with only 7% achieving seizure freedom. Following cenobamate addition, seizure freedom rates significantly increased to 27% in the total cohort. The overall responder rate rose from 61% to 75%. In the CNB subgroup, the responder rate increased from 63% to 85%, with seizure freedom rates rising from 0% to 33%. Patients who were initial ANT-DBS responders exhibited the highest benefit, with 94% achieving seizure reduction and 47% reaching seizure freedom. In contrast, initial ANT-DBS non-responders had a lower response rate of 70%, with 10% achieving seizure freedom. In the non-CNB subgroup, both the responder and seizure freedom rates remained unchanged during the observation period. Discussion: The combination of cenobamate and ANT-DBS has the potential to significantly enhance seizure control in patients with ultra-refractory focal epilepsy. The observed synergy may be attributed to overlapping mechanisms involving GABAergic enhancement, sodium channel modulation, and network-level reorganization. These findings suggest that cenobamate is a promising adjunctive therapy in ANT-DBS patients, potentially increasing the likelihood of seizure freedom. Future prospective studies are warranted to validate these results and further explore the mechanistic interactions between cenobamate and ANT-DBS.