Diffusion Tensor Imaging - Science method
The use of diffusion ANISOTROPY data from diffusion magnetic resonance imaging results to construct images based on the direction of the faster diffusing molecules.
Questions related to Diffusion Tensor Imaging
Dear DTI Experts,
I have run the following TBSS pipeline in order to get a whole-brain FA skeleton for a cohort of 4-8 year old children:
#Run TBSS Preprocessing (must navigate to the folder with all subjects' FA maps)
#TBSS Nonlinear registration: Register to standard space
#Get study-specific FA skeleton. Run from TBSS directory (directory before FA folder)
#TBSS Threshold for FA values <.2
As part of my resultant output, I receive a whole-brain standard space mean FA skeleton mask. How should i go about getting my whole-brain scalars (FA, MD, AD, etc.) using this output? Is it acceptable to warp the standard space mean FA skeleton mask to each subjects' diffusion space, and then overlay the resultant skeleton on each scalar to get whole-brain metrics for each participant? Please advise.
Hi FSL users,
any hint on how to use FSL topup on two phase encoded DTI scan each with 63 directions and 1 b0, the data is actually from PPMI dataset (PD patients) and am trying to correct for susceptibility distortion using topup. I know it can be used for one DTI scan and 2 b0 but have no experience with two scans!
Thanks in advance
Hi, I am conducting a Diffusion Tensor Imaging study on three groups of subjects, first-episode psychosis (FEP) group (Group 1), their siblings (S) (Group 2), and control group (C) (Group 3). I would like to infer the differences of their diffusivity values. I'm using TBSS to do this. I will also use F-Test and then T-Tests as post-hoc. However, I do not know how I can design my matrix in GLM because I have different sample sizes in my 3 group such as 19 FEP, 20 S, and 21 C. Could you help me on this issue, please?
Thank you so much,
I am currently working on a project where I am analysing narcissism features and white matter
I did multiple comparsions (multiple regression) because I was looking for correlations for three tracts, a few sub scales o a questionnaire and positive as well as negative contrasts.
Now I have to correct for multiple comparisons. Is there any other method than bonferroni which is more robust?
#diffusion tensor imaging
I have one DTI dataset, and I've applied two different reconstruction algorithms to it. Thus there were two datasets with the same anatomical structure but varying in sth like the artifacts and SNR. I want to generate their FA maps. I use EDDY and DITFIT from FSL. I find that the results of eddy correction are different. That is, the transformations of two datasets are not the same, and the shape of the corrected images are also different.
eddy_parameters is one of the output files of EDDY. It contains the parameters of the transformation. How can I apply the eddy_parameters of one dataset to the other to ensure the same correction results?
The eddy_parameters file contains a matrix with a size of 21*16. There are 21 volumes in the DTI dataset, which means there are 16 parameters of transformation for each volume. But I don't know what each parameter exactly means? Does anyone have experience in this?
Thank you very much!
Hi I am currently working on my masters thesis and could really use some help. The project includes comparing FA and MD values to sleep EEG values. I'm currently stuck on creating the script for extracting the FA and MD. E.g. if I want to extract the FA for all participant from the genu corpus callosum tract (we're using the JHU atlas and the index number is given is 3). what kind of script would that be?
We're only looking at a small portion of the tracts so I would need to have a command that I can modify to specify this and to specify that the index labels are based on the JHU tracts. And of course I would be able to use the same line of script but change it for MD.
the JHU files are available in a file path that is /usr/local/fsl5/data/atlases/JHU
and I have the data for each participant in my directory and with a FA and MD scan for each participant. They each also have an image where the JHU tracts have been registered into dti space.
Hopefully someone is able to help it would be greatly appreciated!
I am analyzing DTI images (DICOM files) of mice to compare myelination in the corpus callosum between wild type and heterozygous ANKS1B KO models. I have tried visualization via different software however I am having trouble distinguishing from noise and "true" white matter tracts. Attached image shows a whole brain example from diffusion toolkit (Trackvis) which has tracts running wild throughout the entirety of the brain which makes no anatomical sense and does not align with the literature. I am wondering what is the best way to eliminate the nonsense tracts and optimize the real or at least presumably real ones via angle/mask thresholds, btables, angle thresholds, etc. I am very new at this so any help would be appreciated.
I am interested in using this DTI atlas ( http://www.bmap.ucla.edu/portfolio/atlases/ICBM_DTI-81_Atlas/ ) but came across a publication suggesting it is not oriented properly ( https://www.frontiersin.org/articles/10.3389/fnins.2013.00004/full ). however, this was 8 years ago. Have these issues been fixed?
Current research challenges in breast imaging using deep learning or ensemble algorithm, Tesoro Flow, Diffusion Tensor Imaging
The best approach to hierarchical scalability of deep learning to maximize prediction accuracy (1)in the absence of large data, (2) and in the presence of large data.
Hello, this is Jean who is new to use ADNI data + DTI data.
I downloaded Axial DTI data of AD & CN, and I drop those data in dcm2niigui, but it does not work to covert to 4d nil image.
So I checked the data and it has 2,714 of dcm, which is different from what I got in PPMI DTI files.
Is it a matter of axial dti files? or is there any other issue regarding this??
After checking raw Diffusion-Tensor Imaging (DTI) data from some pilots, I noticed how systematically the first two directions show a b = 2150 in place of 2200.
Has it ever happened to any of you? If yes, do you have advice on how to solve it?
Thank you very much!
In cases that mild TBI without no abnormalities in conventional MRI or CT, is there any methods evaluating brain damage in the real clinical field, other than diffuse tensor tractography or diffuse tensor imaging?
I have a research about DTI correlations of cognitive impairment in essential tremor. We have 40 patients, 20 brain regions (right and left, total 40). 8 cognitive domains. I try to adress brain microstructural correlates of cognitive impairment in essential tremor. But I could not figure out which statistical analysis should I use from SPSS?
DTI images of brain tumors with ground truth. It is noted that there is no DTI available in the BRATS dataset.
I've access to a dataset of Schizophrenia patients' Diffusion MRIs in which 3 imaging sites with different acquisition protocols are included (b-values=700,800,1000, directions=64,12,6). Considering that I'm looking into differences of graph metrics and fractional anisotropy of several white matter tracts between healthy and patient groups, is it reasonable that I combine the results of my analysis for each sites and if so, how should I do that?
Hopefully these datasets are collected at 1mm or better resolution and include the CT data down the neck to include the skull base. Typically this is not done without reason but ideally these would be from healthy subjects and include neurophys data (EEG, MEG, etc.) and DWI (HARDI, DSI, etc.) and, why not, MRA. Although any data, including data with metal artifacts from grid, strip or single electrodes, is fine and would be appreciated, as long as it can be shared publicly.
I am interested in extracting brain clusters showing significant correlation between cortical thickness and local gyrification. I guess this can be done using --pvr flag in GLMFIT (FreeSurfer). But I am not sure how to set up this command and create contrast and fsgd files, followed by glmfit-sim command after considering age and gender as covariates.
I would really appreciate any help.
I am interested in isolating the corticobulbar tract, which does not exist in the TBSS or TRACULA atlases. So far, it seems my best bet is to use Freesurfer's dt_recon to use my custom surface labels as seeds for tractography. Has anyone used any other software that they can recommend?
I am using DTI to evaluate neuroplasticity resulted from a new therapy. I want to know is there any interpretation for significantly increase in FA of the GRAY MATTER? Is it true to use FA value for evaluation of neuroplasticity in GRAY MATTER or it is specially just for white matter? Is there any reference regarding this topic?
I am doing the analysis of intracortical EEG data and I want to test my results by predicting my own signal based in Kuramoto models to be sure that my results are correct. I would like to use the structural connectivity data as a coupling strength between the nodes (brain regions based on an atlas), but I do not have DTI data from all this patients. I would like to know from where can I have access to public DTI data to get this parameters.
Thanks in advance!
I have acquired 100 volumes in resting fMRI data in my subjects. So I have to delete 2 or 3 or 5 volumes. If I will consider all the volumes, then is it any problem in analysis because I have 100 volumes only.
Analyzing DTI data, I am looking for a way to measure thalamic and cortical shift of a corticothalamic connection. My hypothesis is that, in patients a given fiber tract reaches the thalami (or cortex) rostrally, for example, compared to controls. However, I don’t know which toolbox I can use to do it. As far as I know, FSL doesn’t provide a good way to perform this analysis. Am I right?
Does anyone know a way to figure this?
I would like to estimate the direction information with CSD and then determine from that the two main direction. From these both I would like to build a two-tensor-method.
So my questions are: When I do CSD what kind of output files do I get exactly? Somebody knows a software who can do Multi-Tensor reconstruction?
thanks for your help!
I have a question regarding the use of bvecs files during DTI preprocessing. I've noticed that the values in the bvecs text file for each participant is slightly different from one another; but only marginally so. (Btw the bvecs files were converted from DICOM to NIFTI using MRIConvert, latest version).
My question is, when performing the DTIfit procedure in FSL, do I need to use each participants' own bvecs file; or can I use one bvecs file for all participants? I've seen several guides/tutorials regarding this, and it seems to me that some of these use one bvecs file for all participants.
All comments are highly appreciated!
As we know Tractography is the only non-invasively tool for determining and measuring the pathways in the brain through observations of local diffusivity. The uncertainty of tractography which arises for example from noise.
I'd like to know what is this noise with example if possible.
I'm fairly new to DKI tractography. I have used ExploreDTI for DTI tractography before and I know a few software packages that work well for DTI, but I wanted to use specific software packages for DKI tractography. I have used DKE to estimate the tensor and it works fine for that aspect, but unfortunately, the tractography module does not work well. I was wondering if anyone here has used any software packages that are specifically designed for DKI tractography (or are at least capable of doing that, especially resolving the crossing fibers issues, etc.) and they have got reasonable outputs. I'd appreciate if you can introduce me some of those packages.
Thanks a lot,
As far as I know, diffusion metrics like fractional anisotropy, and mean, radial, and axial diffusivity can give a direct clue about water diffusion and an indirect clue about myelination in a specific voxel, but to which extent this difference in myelination inferred from increased FA (e.g. between two groups) is related to increase in myelin layer thickness or it is related to myelination of unmyelinated fibers (i.e. increased number of oligodendrocytes) is unknown and I couldn't find any methods. I wonder whether there is a tool or method using DWI or combinational methods (e.g. DWI+MTR) to differentiate these two possible underlying mechanisms.
DTI studies show that a two week mindfulness training is accompanied by a decrease of axonal density in the ACC - as index by axial diffusivity. Which effects or symptoms are normally associated with such a reduction?
We are trying to perform in vivo diffusion tensor imaging of mouse brain on a Bruker 9.4T MRI. I have tried both EPI-based and Spin Echo-based sequences, but have been having difficulty with respiratory motion artifacts. In the few mice where we have near-perfect head immobilization, the spin echo images are very good. However, most of them have respiratory motion artifacts that seriously degrade the tensor calculations. For EPI, we are having a lot of EPI ghosting that I think may be related to respiratory motion.
We are currently using tape and gauze - what are your favorite tricks/apparatuses for fixing the heads very still? (Any other suggestions for improving in vivo DTI quality also appreciated.)
We recently concluded a study on the prevalence and characteristics of maxillofacial injuries in patients with mTBi, and analysed the possible influence of maxillofacial (MF) trauma over specific cognitive deficits post trauma (namely executive function, memory and attention). We also looked out for WM tracts that were affected both in the acute and follow up phase [controlling for both the presence of maxfac injuries and as well as the CT imaging findings (intracranial lesion vs. none)]. The results were quite interesting and seem to challenge the conventional understanding and management of patients with mTBI.
We found that patients with maxillofacial injuries without intracranial lesion doing significantly worse over time in the domains of executive function and memory. Miscrostructurally, these patients seem to have poorer WM integrity especially involving the projection and association fibers (mainly corona radiata, cingulum, superior longitudinal fasiculus, optic radiation and genu of the corpus callosum).
Would appreciate your thoughts on the biophysics and biomechanics of maxillofacial trauma in mTBI and how that could explain the findings.
In other words: how many of subject should best have a tract in the voxel so that the voxel is included in the group mask?
I have the following thresholds in my mind: 50, 60, 70, 80, 90, 100% but have no real experiences so far.
Thank you for answering it!!
My current data on neuropsychological performance in patients with mTBI across two time points (admission vs follow up) indicate that although the patients with the cortical injury do significantly worse in almost all domains of neurocognitve assessment during the acute phase, these patients, however, outperform the patients with "plausible" axonal injury (based on the DTI data) in the follow up phase (6 months post trauma)? What do you think is happening here?
Traumatic brain injury, even in its mildest form, is known to result in degenerative processes including demyelination and dysmyelination of the axons over time. The shearing and tearing of the axons (primary injury) due to the acceleration and deceleration force of high velocity impact would also normally trigger off the secondary injury cascades. This includes the synaptic deregulation, cell death and axonal degeneration.
But how quickly does these processes start (especially the demyelination of the axons) in patients with mild TBI? I am of the opinion that it will take at least a few days or weeks before such degenerative process starts. What are your thoughts?
For my project, we will acquire HARDI DTI data of brain tumor patients and healthy controls in order to construct structural connectivity matrices (based on tractography results) for graph theory analysis.
Can anyone recommend any program, preferably script-based?
I heard of the Diffusion Toolkit/TrackVis, MRtrix, and Camino, but I don't know which one would be best for my purpose.
In terms of 1) efficiency 2) ability to perform ROI analysis for DTI parameters (Eigen values, FI, CL, et.) 3) efficient help (easy to follow) 4) free (for scientific purpose)
Some of the missense mutations of BDNF have been theoretically implicated (rs6265 in particular) to influence various outcomes of brain injury. Can someone please elaborate on how certain genetic polymorphisms of BDNF may aggravate cerebral oedema in mild TBI. If you have come across any such articles, please feel free to share.
I am currently working on MR diffusion model to monitor defects in human heart. Meanwhile, I am able to assume that diffusion coefficient is related to the relaxation rates (R1=1/T1 and R2=1/T2), I need experts to advise on which substance whose T1 and T2, I will have to use.
I need these features because the substance which is diffusing is expected to carry molecular signatures of the heart. Thank you.
I have a 64 gradient plus 5 B0s DTI dataset and need to remove some intermediate volumes (ie. volumes number 10, 23, 35, 55). In fact I have a software that requires all the B0s to be at the beginning of the study, and a dataset where the B0s are distributed through the acquisition. How can I do this? Can I use Matlab? Could you suggest me the correct procedure? Alternatively I need to move the same volumes so that they all end up at the beginning of the volumes sequence.
Alteration of sexual functioning in patients with traumatic brain injury is a phenomenon that has been widely studied. However, these studies are predominantly limited to self reported symptoms and hypothalamic-pituitary dysfunction (assessed through hormonal workouts mostly. Detailed imaging studies of this phenomenon are extremely scarce, if not unheard of.
What are your thoughts on possible influence of deep white matter tracts on sexual functioning in TBI, and MTBI specifically?
I'm using the Quality Assessment tool from ExploreDTI based on residuals. There I get a diagram of absolute residual model errors per DW-Image, but I don't really understand what the measure of the scaling is. When do I know which DW-Images are corrupted, is there a specific threshold?
And more generally:
Can I exclude some because of obviously visual artifacts? Is there a percentage of how many images I can exclude per person? Are there any guidelines how to do this right?
Thanks a lot for your help in advance!
Given that someone works with noisy datasets, what would you recommend as a quality measure? There are no repeated B0s to estimate sigma.
I may be wrong, but I have understood that this is the only way to work with spherical deconvolution algorithms. Can you suggest any other solutions to obtain diffusion images obtainable with the same type of scanner that would allow working with spherical deconvolution algorithms?
Literatures are generally equivocal about negative correlation found in diffusion tensor imaging (DTI) of mTBI patients, especially when the inverse correlations are found at the initial admission DTI and neuropsychological testing. Some associate the negative correlation with cytotoxic edema (thus the increased FA vs poorer neurocognitive performance). How do you justify both positive correlation and negative correlation with poorer cognitive performance at admission?
I have some DWI (b=0 and 1000) scans and need the ADC maps, unfortunately not in all subjects the ADC map was computed at the console. I could write a python script that would do that, but I want to know if there is a downloadable software that would do the same.
I am currently working on MEG data from patients with GPi and STN DBS. All connectivity studies including ours can find prominent theta connectivity between the temporal lobe and the basal ganglia structures. Theta is the most prominent hippocampal rhythm.
I am currently working on simultaneous MEG-LFP recordings and would like to compare my results of oscillatory connectivity to those from imaging studies.
I am interested in starting a collaboration on processing cardiac DW-MRI sequences. Our research group has experience in brain DW-MRI processing.
I'm conducting a study on three groups of subjects, 1 control group (HC) and two different degrees of the same pathology (D1 and D2). I would like to infer the differences of their diffusivity values. I'm using TBSS to do this. I've just compared them, two by two, using unpaired t-tests. Do you suggest to use an ANOVA design before? If yes, which should be the contrast matrix?
I suppose it should be a relatively easy question but I did not find a general agreement about this.
The Economist magazine (Aug. 3, 2013) published the following statement in the article entitled ‘The Machine of a New Soul’:
“An important property of a real brain is that it is what is referred to as a small-world network. Each neuron within it has tens of thousands of synaptic connections with other neurons. This means that, even though a human brain contains about 86 billion neurons, each is within two or three connections of all the others via myriad potential routes.”
Does anyone have a reference for the statement ‘each [neuron] is within two or three connections of all the others via myriad potential routes’?
I'm currently comparing the DTI data of three groups of subjects using TBSS.
At first glance I found significant differences among them by performing unpaired t-tests.
After removing two outliers, due to their clinical condition, and using a multiple comparison correction strategy I didn’t find significant differences in two of the comparisons anymore. I therefore tried to compare them without corrections (using a standard threshold) and I found the expected differences again.
Does anyone has the same problem? Do you think it possible to publish uncorrected data if they are reasonable?
I recently acquired the DTI images of a big sample of children (age range 2-12 years) affected by a certain pathology.
The principal aim of my work is to study the evolution of the white matter in this kind of disease. The principal issue is that we have no controls of the same age range because it's generally difficult to enrol healthy children to be sedated.
Given that it isn't correct to compare data acquired from 2 different scanners, does anyone have any suggestions?
There has been some discussion regarding which approach – ROI vs Whole brain analysis – is better to utilise. My personal opinion is to use every possible mean to confirm your findings. However, the question I would like to ask you is the following: if you have performed a whole brain analysis and you find a correlation in an area that is not theoretically one of the typical areas you would expect, when you perform ROI analysis, would you put that area in there along with the ones theoretically associated with your task? If you have seen this in a publication, would you criticize that this area was used as a ROI?
The structural connectivity network can be derived from diffusion tensor imaging. However, we encounter a strange question. The MFC and PCC were defined according to the AAL template in MNI space and then were transformed to the individual DTI space for constructing the cingulum tracts. Unfortunately, there is nothing we got at last for about 40 subjects. Then, we dilated the ROI we made before and acquired some good results. My question is that is it necessary for us to dilate all the 90 ROIs to constructing the whole brain structural network for graph theory analysis?
Philips PAR files (ver. 4.1) contain the x,y,z components of each gradients direction. Whilst it is known that the order must be changed, and one sign must be flipped, my question is if the slice orientation (given in the header of the PAR) should also be taken into account. If true, this would imply a different set of bvecs per scan. Opinions?