Article

Improved Optimization for the Robust and Accurate Linear Registration and Motion Correction of Brain Images

Department of Engineering Science, University of Oxford, Oxford, England, United Kingdom
NeuroImage (Impact Factor: 6.36). 11/2002; 17(2):825-41. DOI: 10.1016/S1053-8119(02)91132-8
Source: PubMed

ABSTRACT

Linear registration and motion correction are important components of structural and functional brain image analysis. Most modern methods optimize some intensity-based cost function to determine the best registration. To date, little attention has been focused on the optimization method itself, even though the success of most registration methods hinges on the quality of this optimization. This paper examines the optimization process in detail and demonstrates that the commonly used multiresolution local optimization methods can, and do, get trapped in local minima. To address this problem, two approaches are taken: (1) to apodize the cost function and (2) to employ a novel hybrid global-local optimization method. This new optimization method is specifically designed for registering whole brain images. It substantially reduces the likelihood of producing misregistrations due to being trapped by local minima. The increased robustness of the method, compared to other commonly used methods, is demonstrated by a consistency test. In addition, the accuracy of the registration is demonstrated by a series of experiments with motion correction. These motion correction experiments also investigate how the results are affected by different cost functions and interpolation methods.

Download full-text

Full-text

Available from: Peter Bannister
  • Source
    • "The functional data for each subject was motion corrected (MCFLIRT [35]) and field maps were processed using PRELUDE + FUGUE [36, 37] to correct for field distortions in the functional data. Registration to each subject's high resolution structural image was performed using FLIRT [35, 38] and registration to standard space was then performed using FNIRT nonlinear registration [39] . Data was smoothed spatially with a Gaussian kernel with a FWHM of 5 mm and filtered with a highpass temporal filter (cut off of 100 s) and the data was demeaned on a voxel-by-voxel basis across the time course. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Chronic musculoskeletal pain (CMSKP) is attentionally demanding, complex and multi-factorial; neuroimaging research in the population seen in pain clinics is sparse. A better understanding of the neural activity underlying attentional processes to pain related information compared to healthy controls may help inform diagnosis and management in the future. Blood oxygenation level dependent functional magnetic resonance imaging (BOLD fMRI) compared brain responses in patients with CMSKP (n = 15) and healthy controls (n = 14) while completing a modified Stroop task using pain-related, positive-emotional, and neutral control words. Response times in the Stroop task were no different for CMSKP patients compared with controls, but patients were less accurate in their responses to all word types. BOLD fMRI responses during presentation of pain-related words suggested increases in neural activation in patients compared to controls in regions previously reported as being involved in pain perception and emotion: the anterior cingulate cortex, insula and primary and secondary somatosensory cortex. No fMRI differences were seen between groups in response to positive or control words. Using this modified Stroop tasks, specific differences were identified in brain activity between CMSKP patients and controls in response to pain-related information using fMRI. This provided evidence of differences in the way that pain-related information is processed in those with chronic complex musculoskeletal pain that were not detectable using the behavioural measures of speed and accuracy. The study may be helpful in gaining new insights into the impact of attention in those living with chronic pain.
    Full-text · Article · Dec 2016 · BMC Psychology
  • Source
    • "An overview of the pre and post-processing steps for resting-state and task-based protocols can be found in Figure 1. Similar pipelines were used for pre-processing resting-state and task-based fMRI data: motion correction was performed using MCFLIRT (Jenkinson et al., 2002); functional data was transformed into subjects' structural space and resampled to 2 mm × 2 mm × 2 mm voxel size using FLIRT (Jenkinson et al., 2002). Due to brain deformations caused by lesions, registrations between the structural and functional images were performed using rigid body transformation with 6 degrees-of-freedom (DOF) and later manually improved if necessary. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Functional magnetic resonance imaging, fMRI, is a well-known non-invasive technique for the study of brain function. One of its most common clinical applications is preoperative language mapping, essential for the preservation of function in neurosurgical patients. Typically, fMRI is used to track task-related activity, but poor task performance and movement artefacts can be critical limitations in clinical settings. Recent advances in resting-state protocols open new possibilities for pre-surgical mapping of language potentially overcoming these limitations. To test the feasibility of using resting-state fMRI instead of conventional active task-based protocols, we compared results from fifteen patients with brain lesions while performing a verb-to-noun generation task and while at rest. Task-activity was measured using a general linear model analysis and independent component analysis (ICA). Resting-state networks were extracted using ICA and further classified in two ways: manually by an expert and by using an automated template matching procedure. The results revealed that the automated classification procedure correctly identified language networks as compared to the expert manual classification. We found a good overlay between task-related activity and resting state language maps, particularly within the language regions of interest. Furthermore, resting-state language maps were as sensitive as task-related maps, and had higher specificity. Our findings suggest that resting-state protocols may be suitable to map language networks in a quick and clinically efficient way.
    Full-text · Article · Feb 2016 · Frontiers in Human Neuroscience
    • "All MRI images were converted to Analyze format, scaled up by a factor of 10 and spatially pre-processed with a custom-developed MATLAB functions (version R2014a). Briefly, T1-weighted and brainextracted images were spatially normalized using a rat brain template co-registered to a rat brain atlas (Schwarz et al., 2006) by a 12-parameter affine transformation using the FSL/ FLIRT tool (Jenkinson et al., 2002). This template was co-registered to the digitized Paxinos and Watson atlas (4th ed., 1998), enabling atlasbased generation of region-of-interest (ROI) masks for detailed anatomical analysis. "
    [Show abstract] [Hide abstract]
    ABSTRACT: The nonselective opioid receptor antagonist naltrexone is now used for the treatment of alcoholism, yet naltrexone’s central mechanism of action remains poorly understood. One line of evidence suggests that opioid antagonists regulate alcohol drinking through interaction with the mesolimbic dopamine system. Hence, our goal here was to examine the role of the nucleus accumbens connectivity in alcohol reinforcement and naltrexone’s actions using manganese-enhanced magnetic resonance imaging (MEMRI). Following long-term free-choice drinking of alcohol and water, AA (Alko Alcohol) rats received injections of MnCl2 into the nucleus accumbens for activity-dependent tracing of accumbal connections. Immediately after the accumbal injections, rats were imaged using MEMRI, and then allowed to drink either alcohol or water for the next 24 hours. Naltrexone was administered prior to the active dark period, and the second MEMRI was performed 24 hours after the first scan. Comparison of signal intensity at 1 and 24 hours after accumbal MnCl2 injections revealed an ipsilateral continuum through the ventral pallidum, bed nucleus of the stria terminalis, globus pallidus, and lateral hypothalamus to the substantia nigra and ventral tegmental area. Activation was also seen in the rostral part of the insular cortex and regions of the prefrontal cortex. Alcohol drinking resulted in enhanced activation of these connections, whereas naltrexone suppressed alcohol-induced activity. These data support the involvement of the accumbal connections in alcohol reinforcement and mediation of naltrexone’s suppressive effects on alcohol drinking through their deactivation.
    No preview · Article · Jan 2016 · European neuropsychopharmacology: the journal of the European College of Neuropsychopharmacology
Show more

Questions & Answers about this publication

  • Emil Ljungberg added an answer in FSL:
    Is there a resource for FSL(analysis tools for FMRI) that explains methods used in this software?

    Dear all,

    I want to know what method has been used for example for reistration in FSL. Could you please recommend a good source that helps me?

    most of the papers I have found are about how to use FSL to get good result. not the methods underlying them.

    Thank you in advance.

    Emil Ljungberg

    I think the best resources you can get is to look at the FSL website and browse the various tools. For most of them you will find the articles to cite on the tools. Example FLIRT: http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FLIRT

    Looking at the bottom of that page you find two articles:

    • Jenkinson, M., Bannister, P., Brady, J. M. and Smith, S. M. Improved Optimisation for the Robust and Accurate Linear Registration and Motion Correction of Brain Images. NeuroImage, 17(2), 825-841, 2002.
    • Jenkinson, M. and Smith, S. M. A Global Optimisation Method for Robust Affine Registration of Brain Images. Medical Image Analysis, 5(2), 143-156, 2001.
    • Greve, D.N. and Fischl, B. Accurate and robust brain image alignment using boundary-based registration. NeuroImage, 48(1):63-72, 2009.

    If you browse these articles I think you will find a lot of the details. The same goes for most tools developed by the FSL team. If there are other more composite tools, like SIENA etc, you should have a look at the scripts and try to follow what is going on. You will find all scripts under $FSLDIR/bin/

    You could also have a look at the presentation slides from the FSL course they give every year. In there they provide some background as well on the tools and the actual mathematics/programming that is going on under the hood.

    Hope that helps a bit!

    • Source
      [Show abstract] [Hide abstract]
      ABSTRACT: Linear registration and motion correction are important components of structural and functional brain image analysis. Most modern methods optimize some intensity-based cost function to determine the best registration. To date, little attention has been focused on the optimization method itself, even though the success of most registration methods hinges on the quality of this optimization. This paper examines the optimization process in detail and demonstrates that the commonly used multiresolution local optimization methods can, and do, get trapped in local minima. To address this problem, two approaches are taken: (1) to apodize the cost function and (2) to employ a novel hybrid global-local optimization method. This new optimization method is specifically designed for registering whole brain images. It substantially reduces the likelihood of producing misregistrations due to being trapped by local minima. The increased robustness of the method, compared to other commonly used methods, is demonstrated by a consistency test. In addition, the accuracy of the registration is demonstrated by a series of experiments with motion correction. These motion correction experiments also investigate how the results are affected by different cost functions and interpolation methods.
      Full-text · Article · Nov 2002 · NeuroImage

    + 2 more attachments