Birte U. Forstmann

YNIMG Journal 2022 Journal Article

A unified model for reconstruction and R 2 * mapping of accelerated 7T data using the quantitative recurrent inference machine

• Chaoping Zhang
• Dimitrios Karkalousos
• Pierre-Louis Bazin
• Bram F. Coolen
• Hugo Vrenken
• Jan-Jakob Sonke
• Birte U. Forstmann
• Dirk H.J. Poot

Quantitative MRI (qMRI) acquired at the ultra-high field of 7 Tesla has been used in visualizing and analyzing subcortical structures. qMRI relies on the acquisition of multiple images with different scan settings, leading to extended scanning times. Data redundancy and prior information from the relaxometry model can be exploited by deep learning to accelerate the imaging process. We propose the quantitative Recurrent Inference Machine (qRIM), with a unified forward model for joint reconstruction and R 2 * -mapping from sparse data, embedded in a Recurrent Inference Machine (RIM), an iterative inverse problem-solving network. To study the dependency of the proposed extension of the unified forward model to network architecture, we implemented and compared a quantitative End-to-End Variational Network (qE2EVN). Experiments were performed with high-resolution multi-echo gradient echo data of the brain at 7T of a cohort study covering the entire adult life span. The error in reconstructed R 2 * from undersampled data relative to reference data significantly decreased for the unified model compared to sequential image reconstruction and parameter fitting using the RIM. With increasing acceleration factor, an increasing reduction in the reconstruction error was observed, pointing to a larger benefit for sparser data. Qualitatively, this was following an observed reduction of image blurriness in R 2 * -maps. In contrast, when using the U-Net as network architecture, a negative bias in R 2 * in selected regions of interest was observed. Compressed Sensing rendered accurate, but less precise estimates of R 2 *. The qE2EVN showed slightly inferior reconstruction quality compared to the qRIM but better quality than the U-Net and Compressed Sensing. Subcortical maturation over age measured by a linearly increasing interquartile range of R 2 * in the striatum was preserved up to an acceleration factor of 9. With the integrated prior of the unified forward model, the proposed qRIM can exploit the redundancy among repeated measurements and shared information between tasks, facilitating relaxometry in accelerated MRI.

YNIMG Journal 2022 Journal Article

Charting human subcortical maturation across the adult lifespan with in vivo 7 T MRI

• Steven Miletić
• Pierre-Louis Bazin
• Scott J.S. Isherwood
• Max C. Keuken
• Anneke Alkemade
• Birte U. Forstmann

The human subcortex comprises hundreds of unique structures. Subcortical functioning is crucial for behavior, and disrupted function is observed in common neurodegenerative diseases. Despite their importance, human subcortical structures continue to be difficult to study in vivo. Here we provide a detailed account of 17 prominent subcortical structures and ventricles, describing their approximate iron and myelin contents, morphometry, and their age-related changes across the normal adult lifespan. The results provide compelling insights into the heterogeneity and intricate age-related alterations of these structures. They also show that the locations of many structures shift across the lifespan, which is of direct relevance for the use of standard magnetic resonance imaging atlases. The results further our understanding of subcortical morphometry and neuroimaging properties, and of normal aging processes which ultimately can improve our understanding of neurodegeneration.

YNICL Journal 2021 Journal Article

Variability in subthalamic nucleus targeting for deep brain stimulation with 3 and 7 Tesla magnetic resonance imaging

• Bethany R. Isaacs
• Margot Heijmans
• Mark L. Kuijf
• Pieter L. Kubben
• Linda Ackermans
• Yasin Temel
• Max C. Keuken
• Birte U. Forstmann

Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is an effective surgical treatment for Parkinson's disease (PD). Side-effects may, however, be induced when the DBS lead is placed suboptimally. Currently, lower field magnetic resonance imaging (MRI) at 1.5 or 3 Tesla (T) is used for targeting. Ultra-high-field MRI (7 T and above) can obtain superior anatomical information and might therefore be better suited for targeting. This study aims to test whether optimized 7 T imaging protocols result in less variable targeting of the STN for DBS compared to clinically utilized 3 T images. Three DBS-experienced neurosurgeons determined the optimal STN DBS target site on three repetitions of 3 T-T2, 7 T-T2*, 7 T-R2* and 7 T-QSM images for five PD patients. The distance in millimetres between the three repetitive coordinates was used as an index of targeting variability and was compared between field strength, MRI contrast and repetition with a Bayesian ANOVA. Further, the target coordinates were registered to MNI space, and anatomical coordinates were compared between field strength, MRI contrast and repetition using a Bayesian ANOVA. The results indicate that the neurosurgeons are stable in selecting the DBS target site across MRI field strength, MRI contrast and repetitions. The analysis of the coordinates in MNI space however revealed that the actual selected location of the electrode is seemingly more ventral when using the 3 T scan compared to the 7 T scans.

YNIMG Journal 2020 Journal Article

fMRI protocol optimization for simultaneously studying small subcortical and cortical areas at 7 ​T

• Steven Miletić
• Pierre-Louis Bazin
• Nikolaus Weiskopf
• Wietske van der Zwaag
• Birte U. Forstmann
• Robert Trampel

Most fundamental cognitive processes rely on brain networks that include both cortical and subcortical structures. Studying such networks using functional magnetic resonance imaging (fMRI) requires a data acquisition protocol that provides blood-oxygenation-level dependent (BOLD) sensitivity across the entire brain. However, when using standard single echo, echo planar imaging protocols, researchers face a tradeoff between BOLD-sensitivity in cortex and in subcortical areas. Multi echo protocols avoid this tradeoff and can be used to optimize BOLD-sensitivity across the entire brain, at the cost of an increased repetition time. Here, we empirically compare the BOLD-sensitivity of a single echo protocol to a multi echo protocol. Both protocols were designed to meet the specific requirements for studying small, iron rich subcortical structures (including a relatively high spatial resolution and short echo times), while retaining coverage and BOLD-sensitivity in cortical areas. The results indicate that both sequences lead to similar BOLD-sensitivity across the brain at 7 ​T.

YNIMG Journal 2020 Journal Article

Sharpness in motion corrected quantitative imaging at 7T

• Pierre-Louis Bazin
• Hannah E. Nijsse
• Wietske van der Zwaag
• Daniel Gallichan
• Anneke Alkemade
• Frans M. Vos
• Birte U. Forstmann
• Matthan W.A. Caan

Sub-millimeter imaging at 7T has opened new possibilities for qualitatively and quantitatively studying brain structure as it evolves throughout the life span. However, subject motion introduces image blurring on the order of magnitude of the spatial resolution and is thus detrimental to image quality. Such motion can be corrected for, but widespread application has not yet been achieved and quantitative evaluation is lacking. This raises a need to quantitatively measure image sharpness throughout the brain. We propose a method to quantify sharpness of brain structures at sub-voxel resolution, and use it to assess to what extent limited motion is related to image sharpness. The method was evaluated in a cohort of 24 healthy volunteers with a wide and uniform age range, aiming to arrive at results that largely generalize to larger populations. Using 3D fat-excited motion navigators, quantitative R 1, R 2 * and Quantitative Susceptibility Maps and T 1-weighted images were retrospectively corrected for motion. Sharpness was quantified in all modalities for selected regions of interest (ROI) by fitting the sigmoidally shaped error function to data within locally homogeneous clusters. A strong, almost linear correlation between motion and sharpness improvement was observed, and motion correction significantly improved sharpness. Overall, the Full Width at Half Maximum reduced from 0. 88 mm to 0. 70 mm after motion correction, equivalent to a 2. 0 times smaller voxel volume. Motion and sharpness were not found to correlate with the age of study participants. We conclude that in our data, motion correction using fat navigators is overall able to restore the measured sharpness to the imaging resolution, irrespective of the amount of motion observed during scanning.

YNIMG Journal 2019 Journal Article

Functional neuroanatomical review of the ventral tegmental area

• Anne C. Trutti
• Martijn J. Mulder
• Bernhard Hommel
• Birte U. Forstmann

The ventral tegmental area (VTA) and substantia nigra pars compacta (SNc) are assumed to play a key role in dopamine-related functions such as reward-related behaviour, motivation, addiction and motor functioning. Although dopamine-producing midbrain structures are bordering, they show significant differences in structure and function that argue for a distinction when studying the functions of the dopaminergic midbrain, especially by means of neuroimaging. First, unlike the SNc, the VTA is not a nucleus, which makes it difficult to delineate the structure due to lack of clear anatomical borders. Second, there is no consensus in the literature about the anatomical nomenclature to describe the VTA. Third, these factors in combination with limitations in magnetic resonance imaging (MRI) complicate VTA visualization. We suggest that developing an MRI-compatible probabilistic atlas of the VTA will help to overcome these issues. Such an atlas can be used to identify the individual VTA and serve as region-of-interest for functional MRI.

YNIMG Journal 2018 Journal Article

Outlook on deep neural networks in computational cognitive neuroscience

• Brandon M. Turner
• Steven Miletić
• Birte U. Forstmann

YNIMG Journal 2016 Journal Article

Automated segmentation of the substantia nigra, subthalamic nucleus and red nucleus in 7 T data at young and old age

• Eelke Visser
• Max C. Keuken
• Birte U. Forstmann
• Mark Jenkinson

With recent developments in MR acquisition at 7T, smaller brainstem structures such as the red nuclei, substantia nigra and subthalamic nuclei can be imaged with good contrast and resolution. These structures have important roles both in the study of the healthy brain and in diseases such as Parkinson's disease, but few methods have been described to automatically segment them. In this paper, we extend a method that we have previously proposed for segmentation of the striatum and globus pallidus to segment these deeper and smaller structures. We modify the method to allow more direct control over segmentation smoothness by using a Markov random field prior. We investigate segmentation performance in three age groups and show that the method produces consistent results that correspond well with manual segmentations. We perform a vertex-based analysis to identify changes with age in the shape of the structures and present results suggesting that the method may be at least as effective as manual delineation in capturing differences between subjects.

YNIMG Journal 2016 Journal Article

Automatic segmentation of the striatum and globus pallidus using MIST: Multimodal Image Segmentation Tool

• Eelke Visser
• Max C. Keuken
• Gwenaëlle Douaud
• Veronique Gaura
• Anne-Catherine Bachoud-Levi
• Philippe Remy
• Birte U. Forstmann
• Mark Jenkinson

Accurate segmentation of the subcortical structures is frequently required in neuroimaging studies. Most existing methods use only a T 1-weighted MRI volume to segment all supported structures and usually rely on a database of training data. We propose a new method that can use multiple image modalities simultaneously and a single reference segmentation for initialisation, without the need for a manually labelled training set. The method models intensity profiles in multiple images around the boundaries of the structure after nonlinear registration. It is trained using a set of unlabelled training data, which may be the same images that are to be segmented, and it can automatically infer the location of the physical boundary using user-specified priors. We show that the method produces high-quality segmentations of the striatum, which is clearly visible on T 1-weighted scans, and the globus pallidus, which has poor contrast on such scans. The method compares favourably to existing methods, showing greater overlap with manual segmentations and better consistency.

YNIMG Journal 2016 Journal Article

Striatal activation reflects urgency in perceptual decision making

• Leendert van Maanen
• Laura Fontanesi
• Guy E. Hawkins
• Birte U. Forstmann

Deciding between multiple courses of action often entails an increasing need to do something as time passes - a sense of urgency. This notion of urgency is not incorporated in standard theories of speeded decision making that assume information is accumulated until a critical fixed threshold is reached. Yet, it is hypothesized in novel theoretical models of decision making. In two experiments, we investigated the behavioral and neural evidence for an “urgency signal” in human perceptual decision making. Experiment 1 found that as the duration of the decision making process increased, participants made a choice based on less evidence for the selected option. Experiment 2 replicated this finding, and additionally found that variability in this effect across participants covaried with activation in the striatum. We conclude that individual differences in susceptibility to urgency are reflected by striatal activation. By dynamically updating a response threshold, the striatum is involved in signaling urgency in humans.

YNIMG Journal 2016 Journal Article

The Age-ility Project (Phase 1): Structural and functional imaging and electrophysiological data repository

• Frini Karayanidis
• Max C. Keuken
• Aaron Wong
• Jaime L. Rennie
• Gilles de Hollander
• Patrick S. Cooper
• W. Ross Fulham
• Rhoshel Lenroot

Our understanding of the complex interplay between structural and functional organisation of brain networks is being advanced by the development of novel multi-modal analyses approaches. The Age-ility Project (Phase 1) data repository offers open access to structural MRI, diffusion MRI, and resting-state fMRI scans, as well as resting-state EEG recorded from the same community participants (n=131, 15–35y, 66 male). Raw imaging and electrophysiological data as well as essential demographics are made available via the NITRC website. All data have been reviewed for artifacts using a rigorous quality control protocol and detailed case notes are provided.

YNIMG Journal 2014 Journal Article

Do we need to revise the tripartite subdivision hypothesis of the human subthalamic nucleus (STN)?

• Anneke Alkemade
• Birte U. Forstmann

The exciting development of ultra-high resolution 7Tesla (T) magnetic resonance imaging (MRI) has made it possible to clearly visualize and delineate the subthalamic nucleus (STN). Ultra-high resolution MRI provides a first step in the ongoing improvement of imaging techniques rendering it likely that in the near future specific subareas of small brain nuclei such as the STN can be visualized. These developments can contribute to improve clinical imaging, allowing even more accurate targeting of the STN. This is interesting in view of putative limbic, associative, and sensomotoric subdivisions within the STN. The concept of anatomically distinct subdivisions is attractive, both from an anatomical as well as a clinical perspective. However, we argue that the current leading hypothesis of three STN subdivisions is based on low numbers of clinical observations and primate tracing studies. 7T imaging provides us with markers that could potentially help us to distinguish subdivisions, but our preliminary findings do not indicate the existence of subdivisions. In our opinion additional research is needed. As a consequence the tripartite hypothesis should therefore still be a topic of debate. In view of the possible clinical implications, we would like to raise the question whether anatomical evidence on the topological organization within the STN points towards delineated subdivisions, or an organization without strict anatomical boundaries or septa. The latter would require a revision of the current tripartite hypothesis of the human STN.

YNIMG Journal 2013 Journal Article

A Bayesian framework for simultaneously modeling neural and behavioral data

• Brandon M. Turner
• Birte U. Forstmann
• Eric-Jan Wagenmakers
• Scott D. Brown
• Per B. Sederberg
• Mark Steyvers

Scientists who study cognition infer underlying processes either by observing behavior (e. g. , response times, percentage correct) or by observing neural activity (e. g. , the BOLD response). These two types of observations have traditionally supported two separate lines of study. The first is led by cognitive modelers, who rely on behavior alone to support their computational theories. The second is led by cognitive neuroimagers, who rely on statistical models to link patterns of neural activity to experimental manipulations, often without any attempt to make a direct connection to an explicit computational theory. Here we present a flexible Bayesian framework for combining neural and cognitive models. Joining neuroimaging and computational modeling in a single hierarchical framework allows the neural data to influence the parameters of the cognitive model and allows behavioral data, even in the absence of neural data, to constrain the neural model. Critically, our Bayesian approach can reveal interactions between behavioral and neural parameters, and hence between neural activity and cognitive mechanisms. We demonstrate the utility of our approach with applications to simulated fMRI data with a recognition model and to diffusion-weighted imaging data with a response time model of perceptual choice.

YNIMG Journal 2012 Journal Article

Cortico-subthalamic white matter tract strength predicts interindividual efficacy in stopping a motor response

• Birte U. Forstmann
• Max C. Keuken
• Sara Jahfari
• Pierre-Louis Bazin
• Jane Neumann
• Andreas Schäfer
• Alfred Anwander
• Robert Turner

The subthalamic nucleus (STN) is a small but vitally important structure in the basal ganglia. Because of its small volume, and its localization in the basal ganglia, the STN can best be visualized using ultra-high resolution 7 Tesla (T) magnetic resonance imaging (MRI). In the present study, first we individually segmented 7T MRI STN masks to generate atlas probability maps. Secondly, the individually segmented STN masks and the probability maps were used to derive cortico-subthalamic white matter tract strength. Tract strength measures were then taken to test two functional STN hypotheses which account for the efficiency in stopping a motor response: the right inferior fronto-subthalamic (rIFC-STN) hypothesis and the posterior medial frontal cortex-subthalamic (pMFC-STN) hypothesis. Results of two independent experiments show that increased white matter tract strength between the pMFC and STN results in better stopping behaviour.

YNIMG Journal 2006 Journal Article

The parcellation of cortical areas using replicator dynamics in fMRI

• Jane Neumann
• D. Yves von Cramon
• Birte U. Forstmann
• Stefan Zysset
• Gabriele Lohmann

In this paper, we show that replicator dynamics can be used as an exploratory analysis tool to detect subregions of cortical areas on the basis of the similarity between fMRI time series. As similarity measure, we propose to use canonical correlation, a multivariate extension to the typically employed Pearson's correlation coefficient. We applied the replicator process to data obtained from two different experimental paradigms in the search for subregions within the left lateral frontal cortex (LFC). In both cases, the replicator process resulted in a parcellation that corresponds to a recently suggested subdivision of the LFC in anterior–posterior direction. Most notably, these results were very consistent when compared across different measurements of a single subject and across a group of subjects.