#### DMCA

## SPM12 Manual The FIL Methods Group (and honorary members) (2014)

### Citations

445 |
An overlap invariant entropy measure of 3D medical image alignment
- Studholme, DLG, et al.
(Show Context)
Citation Context ...unction Registration involves finding parameters that either maximise or minimise some objective function. For inter-modal registration, use Mutual Information [19, 91], Normalised Mutual Information =-=[86]-=-, or Entropy Correlation Coefficient [65].For within modality, you could also use Normalised Cross Correlation. Separation The average distance between sampled points (in mm). Can be a vector to allow... |

255 | Valid conjunction inference with the minimum statistic.
- Nichols, Brett, et al.
- 2005
(Show Context)
Citation Context ...the Conjunction Null hypothesis. This can be thought of as enabling an inference that subject 1 activated AND subject 2 activated AND subject 3... etc. For more discussion on this issue, see [38] and =-=[74]-=-. Gaussian field theory results are available for SPMs of minimum T- (or F-) statistics and therefore corrected p-values can be computed. Note that the minimum T-values do not have the usual Student’s... |

198 | Interpolation revisited,”
- Thevenaz, Blu, et al.
- 2000
(Show Context)
Citation Context ... the images are sampled when estimating the optimum transformation. Higher degree interpolation methods provide the better interpolation, but they are slower because they use more neighbouring voxels =-=[87, 88, 89]-=-. Wrapping This indicates which directions in the volumes the values should wrap around in. For example, in MRI scans, the images wrap around in the phase encode direction, so (e.g.) the subject’s nos... |

147 |
B-spline signal processing: Part II–Efficiency design and applications
- Unser, Aldroubi, et al.
- 1993
(Show Context)
Citation Context ... the images are sampled when estimating the optimum transformation. Higher degree interpolation methods provide the better interpolation, but they are slower because they use more neighbouring voxels =-=[87, 88, 89]-=-. Wrapping This indicates which directions in the volumes the values should wrap around in. For example, in MRI scans, the images wrap around in the phase encode direction, so (e.g.) the subject’s nos... |

114 | Comparing dynamic causal models.
- Penny, Stephan, et al.
- 2004
(Show Context)
Citation Context ... analytically, therefore an approximation to the model evidence is needed. One such approximation used by DCM, and many other models in SPM, is to make use of the Laplace approximation 1. As shown in =-=[81]-=-, this yields the following expression for the natural logarithm (ln) of the model evidence ( ηθ|y denotes the posterior mean, Cθ|y is the posterior covariance of the parameters, Ce is the error covar... |

66 | Bayesian fMRI time series analysis with spatial priors,”
- Penny, Trujillo-Barreto, et al.
- 2005
(Show Context)
Citation Context ...lly smoothed. Estimation will take about 5 times longer than with the classical approach. This is why VB is not the default estimation option. The VB approach has been described in a number of papers =-=[78, 82, 75, 76]-=-. After estimation, contrasts are used to find regions with effects larger than a user-specified size eg. 1 per cent of the global mean signal. These effects are assessed statistically using a Posteri... |

66 |
B-Spline signal processing: Part I - Theory
- Unser, Aldroubi, et al.
- 1993
(Show Context)
Citation Context ... is fastest, but not recommended for image realignment. Trilinear Interpolation is probably OK for PET, but not so suitable for fMRI because higher degree interpolation generally gives better results =-=[87, 88, 89]-=-. Although higher degree methods provide better interpolation, but they are slower because they use more neighbouring voxels. 3.4.3 Wrapping These are typically: * No wrapping - for images that have a... |

65 |
Bayesian model selection for group studies.
- Stephan, Penny, et al.
- 2009
(Show Context)
Citation Context ...dels. It will open the SPM batch tool for model selection. Specify a directory to write the output file to. For the “Inference method” you can choose between “Fixed effects” and “Random effects” (see =-=[85]-=- for additional explanations). Choose “Fixed effects” if you are not sure. Then click on “Data” and in the box below click on “New: Subject”. Click on “Subject” and in the box below on “New: Session”.... |

55 | Variational Bayesian inference for fMRI time series,”
- Penny, Kiebel, et al.
- 2003
(Show Context)
Citation Context ...assical (ReML) approach. If you choose Bayesian estimation these options will be ignored. For Bayesian estimation, the choice of noise model (AR model order) is made under the estimation options. See =-=[39, 78]-=- for further discussion of these issues. 8.10 Reviewing your design After you have completed the SPM “job” file for specifying your fMRI design, and have run it, you will then be able to review your d... |

55 |
A voxel-based method for the statistical analysis of gray and white matter density applied to schizophrenia.
- Wright
- 1995
(Show Context)
Citation Context ...e volumetric comparisons among populations of subjects. It requires the images to be spatially normalised, segmented into different tissue classes, and smoothed, prior to performing statistical tests =-=[92, 68, 6, 7]-=-. The ”optimised” pre-processing strategy involved spatially normalising subjects’ brain images to a standard space, by matching grey matter in these images, to a grey matter reference. The historical... |

48 | Dynamic causal modelling of evoked responses in EEG/MEG with lead field parameterization.
- SJ, David, et al.
- 2006
(Show Context)
Citation Context ...bout the scientific background, the algorithms used, or how one would typically use DCM in applications, we recommend the following reading. The two key methods contributions can be found in [23] and =-=[59]-=-. Two other contributions using the model for testing interesting hypotheses about neuronal dynamics are described in [60] and [25]. At the time of writing, there were also three application papers pu... |

47 | An empirical Bayesian solution to the source reconstruction problem in EEG.
- Phillips, Mattout, et al.
- 2005
(Show Context)
Citation Context ...o be generic in the sense it can incorporate and estimate the relevance of multiple constraints of varied nature; datadriven relevance estimation being made possible through Bayesian model comparison =-=[39, 83, 67, 35]-=-. • The subject’s specific anatomy is incorporated in the generative model of the data, in a fashion that eschews individual cortical surface extraction. The individual cortical mesh is obtained autom... |

43 |
Multimodality image registration by maximisation of mutual information
- Maes, Collignon, et al.
- 1997
(Show Context)
Citation Context ...ameters that either maximise or minimise some objective function. For inter-modal registration, use Mutual Information [19, 91], Normalised Mutual Information [86], or Entropy Correlation Coefficient =-=[65]-=-.For within modality, you could also use Normalised Cross Correlation. Separation The average distance between sampled points (in mm). Can be a vector to allow a coarse registration followed by increa... |

43 | Voxel-based morphometry of the human brain: Methods and applications. Curr Med Imaging Rev 1:105–113.
- Mechelli, CJ, et al.
- 2005
(Show Context)
Citation Context ...e volumetric comparisons among populations of subjects. It requires the images to be spatially normalised, segmented into different tissue classes, and smoothed, prior to performing statistical tests =-=[92, 68, 6, 7]-=-. The ”optimised” pre-processing strategy involved spatially normalising subjects’ brain images to a standard space, by matching grey matter in these images, to a grey matter reference. The historical... |

41 | MEG source localization under multiple constraints: an extended Bayesian framework.
- Mattout, Phillips, et al.
- 2006
(Show Context)
Citation Context ...o be generic in the sense it can incorporate and estimate the relevance of multiple constraints of varied nature; datadriven relevance estimation being made possible through Bayesian model comparison =-=[39, 83, 67, 35]-=-. • The subject’s specific anatomy is incorporated in the generative model of the data, in a fashion that eschews individual cortical surface extraction. The individual cortical mesh is obtained autom... |

31 |
A neural mass model of spectral responses in electrophysiology,”
- Moran, Kiebel, et al.
- 2007
(Show Context)
Citation Context ... recent SPM book [28], where Parts 6 and 7 cover not only DCM for M/EEG but also related research from our group. The DCMs for induced responses and steady-state responses are covered in [18, 17] and =-=[73, 69, 70]-=-. Also note that there is a DCM example file, which we put onto the webpage http://www.fil.ion.ucl.ac.uk/spm/data/eeg mmn/. After downloading DCMexample.mat, you can load (see below) this file using t... |

27 | Comparing families of dynamic causal models.
- Penny, Stephan, et al.
- 2010
(Show Context)
Citation Context ...3.16: Graph demonstrating PPI interaction. Chapter 34 Bayesian Model Inference This chapter describes the use of SPM’s Bayesian Model Inference capabilities. For a fuller background on this topic see =-=[80]-=-. We illustrate the methods using a DCM for fMRI study of the language system. 34.1 Background The neuroimaging data derive from an fMRI study on the cortical dynamics of intelligible speech [63]. Thi... |

25 |
Bayesian estimation of synaptic physiology from the spectral responses of neural-masses
- Moran, Kiebel, et al.
- 2008
(Show Context)
Citation Context ... recent SPM book [28], where Parts 6 and 7 cover not only DCM for M/EEG but also related research from our group. The DCMs for induced responses and steady-state responses are covered in [18, 17] and =-=[73, 69, 70]-=-. Also note that there is a DCM example file, which we put onto the webpage http://www.fil.ion.ucl.ac.uk/spm/data/eeg mmn/. After downloading DCMexample.mat, you can load (see below) this file using t... |

24 |
The General Linear Model
- Kiebel, Holmes
(Show Context)
Citation Context ...ve enough sensitivity to find activations in single subject PET data. This is why we scan multiple subjects. 31.3 Multiple subjects The data set can be analysed in several ways which are discussed in =-=[57]-=-. 2Normalisation using ANCOVA is advised for multi-subject studies unless differences in global flow are large eg. due to variability in injected tracer dose. Because ANCOVA uses one degree of freedom... |

24 |
The cortical dynamics of intelligible speech.
- Leff, Schofield, et al.
- 2008
(Show Context)
Citation Context ... see [80]. We illustrate the methods using a DCM for fMRI study of the language system. 34.1 Background The neuroimaging data derive from an fMRI study on the cortical dynamics of intelligible speech =-=[63]-=-. This study applied dynamic causal modelling of fMRI responses to investigate activity among three key multimodal regions: the left posterior and anterior superior temporal sulcus (subsequently refer... |

24 | Bayesian Comparison of Spatially Regularised General Linear Models. Human Brain Mapping, 2006. Accepted for publication
- Penny, Flandin, et al.
(Show Context)
Citation Context ...lly smoothed. Estimation will take about 5 times longer than with the classical approach. This is why VB is not the default estimation option. The VB approach has been described in a number of papers =-=[78, 82, 75, 76]-=-. After estimation, contrasts are used to find regions with effects larger than a user-specified size eg. 1 per cent of the global mean signal. These effects are assessed statistically using a Posteri... |

21 |
Dynamic causal modelling of evoked responses: the role of intrinsic connections.
- Kiebel, Garrido, et al.
- 2007
(Show Context)
Citation Context ...following reading. The two key methods contributions can be found in [23] and [59]. Two other contributions using the model for testing interesting hypotheses about neuronal dynamics are described in =-=[60]-=- and [25]. At the time of writing, there were also three application papers published which demonstrate what kind of hypotheses can be tested with DCM [43, 42, 41]. Another good source of background i... |

19 |
Variational Bayesian inversion of the equivalent current dipole model in EEG/MEG.
- Kiebel, Daunizeau, et al.
- 2008
(Show Context)
Citation Context ...“Variational Bayes Equivalent Current Dipoles” (VB-ECDs). For more details about the implementation, please refer to the help and comments in the routines themselves, as well as the original paper by =-=[58]-=-. 15.1 Introduction 3D imaging (or distributed) reconstruction methods consider all possible source location simultaneously, allowing for large and widely spread clusters of activity. This is to be co... |

18 |
Electromagnetic source reconstruction for group studies.
- Litvak, Friston
- 2008
(Show Context)
Citation Context ...ises the spatial resolution and thus subverts the main advantage of using an inversion method that can produce focal solutions. To circumvent this problem we proposed a modification of the MSP method =-=[64]-=- that effectively restricts the activated sources to be the same in all subjects with only the degree of activation allowed to vary. We showed that this modification makes it possible to obtain signif... |

16 |
Dynamic causal models of steady-state responses
- Moran, Stephan, et al.
- 2009
(Show Context)
Citation Context ... recent SPM book [28], where Parts 6 and 7 cover not only DCM for M/EEG but also related research from our group. The DCMs for induced responses and steady-state responses are covered in [18, 17] and =-=[73, 69, 70]-=-. Also note that there is a DCM example file, which we put onto the webpage http://www.fil.ion.ucl.ac.uk/spm/data/eeg mmn/. After downloading DCMexample.mat, you can load (see below) this file using t... |

16 |
Multi-modal volume registration by maximisation of mutual information. Medical Image Analysis
- Viola, Atsumi, et al.
(Show Context)
Citation Context ...optimisation algorithm [84]. Objective Function Registration involves finding parameters that either maximise or minimise some objective function. For inter-modal registration, use Mutual Information =-=[19, 91]-=-, Normalised Mutual Information [86], or Entropy Correlation Coefficient [65].For within modality, you could also use Normalised Cross Correlation. Separation The average distance between sampled poin... |

11 |
Dynamic causal models for phase coupling,”
- Penny, Litvak, et al.
- 2009
(Show Context)
Citation Context ...anges in a network of oscillators. The influence that the phase of one oscillator has on the change of phase of another is characterised in terms of a Phase Interaction Function (PIF) as described in =-=[79]-=-. SPM supports PIFs specified using arbitrary order Fourier series. However, to simplify the interface, one is restricted to simple sinusoidal PIFs with the GUI. 39.1 Data We will use the merged epoch... |

10 |
Dynamic causal modeling: a generative model of slice timing in fMRI.
- Kiebel, Klöppel, et al.
- 2007
(Show Context)
Citation Context ...data, which was the default in the original DCM version. For sequential (as opposed to interleaved) data, this modelling option allows to use DCM in combination with any TR (slice timing differences) =-=[62]-=-. Here, we proceed with the default values. 10. Enter 0.04 for “Echo Time, TE[s]”. 11. Modulatory effects: bilinear 12. States per region: one 13. Stochastic effects: no 14. Define the following extri... |

9 | Population dynamics: Variance and the sigmoid activation function.
- AC, Daunizeau, et al.
- 2008
(Show Context)
Citation Context ...hand side lets you choose the neuronal model. ’ERP’ is the standard model described in most of our older papers, e.g. [23]. ’SEP’ uses a variant of this model, however, the dynamics tend to be faster =-=[66]-=-. ’NMM’ is a nonlinear neural mass model based on a first-order approximation, and ’MFM’, is also nonlinear and is based on a second-order approximation. ’NMDA’ is a variant of the ’NMM’ model which a... |

8 | Dynamic causal modeling for EEG and MEG
- Kiebel, Garrido, et al.
- 2009
(Show Context)
Citation Context ...del inversion is implemented using a Bayesian approach, one can also compute Bayesian model evidences. These can be used to compare alternative, equally plausible, models and decide which is the best =-=[61]-=-. DCM for evoked responses takes the spatial forward model into account. This makes DCM a spatiotemporal model of the full data set (over channels and peri-stimulus time). Alternatively, one can descr... |

8 |
Bayesian analysis of single-subject fMRI: SPM implementation
- Penny, Flandin
- 2005
(Show Context)
Citation Context ...lly smoothed. Estimation will take about 5 times longer than with the classical approach. This is why VB is not the default estimation option. The VB approach has been described in a number of papers =-=[78, 82, 75, 76]-=-. After estimation, contrasts are used to find regions with effects larger than a user-specified size eg. 1 per cent of the global mean signal. These effects are assessed statistically using a Posteri... |

2 |
Dcm for steady state responses: a case study of anaesthesia dept
- Moran, Stephan, et al.
- 2009
(Show Context)
Citation Context ...view This chapter describes the analysis of a 2-channel Local Field Potential (LFPs) data set using dynamic causal modelling. The LFPs were recorded from a single rodent using intracranial electrodes =-=[72]-=-. We thank Marc Tittgemeyer for providing us with this data. The theory behind DCM for cross spectral densities (DCM-CSD) is described in [29]. This DCM is a generalization of DCM for Steady State Res... |

1 |
Neural fields, spectral responses and lateral connections
- Moran, Stephan, et al.
(Show Context)
Citation Context ...tion of a neuronal ensemble. In this model, cells possess AMPA, GABAA, and NMDA-like receptor dynamics, with appropriate ion-channel time constants and a voltage dependent switch for the NMDA channel =-=[71]-=-. From the graphical user interface trial specific effects can be selected for extrinsic connections or intrinsic connections, for the NMDA case the intrinsic connection that is modulated is an excita... |

1 |
Canonical microcircuits for predictive coding
- Usrey, Adams, et al.
(Show Context)
Citation Context ... is also nonlinear and is based on a second-order approximation. ’NMDA’ is a variant of the ’NMM’ model which also includes a model of NMDA receptor. ’CMC’ and ’CMM’ are canonical microcircuit models =-=[90]-=- used in the more recent paper to link models of neurophysiological phenomena with canonical models of cortical processing based on the idea of predictive coding. 16.5 Data and design In this part, yo... |