maps

[kbockhorst]

I calculated CBV, MTT, .. maps. How do I read the values? Pixel scan? These values are completely off the expected values

Kurt

[wullemom]

You can read the values e.g. with the pixel scan function or draw ROIs und use the Roi/Statistics function.

If your values are off from the expected range, you should check the AIF. The AIF is defined either by the ROI-method or cluster analysis. The ROI method is very stable also for pure image quality.

See also these manual pages:

Absolute quantification of blood volume, blood flow and MTT is obtained in two steps. First the parameters of the GVF for pure vessel signal are selected. This can be achieved either manually or by using cluster analysis. The method can be chosen in FlowQuan parameter area.
? ROI: Use the pixels defined by ROI n. The ROI must be drawn manually in the window, which contains the raw data set. It is advised to draw very small ROIs, which contain only pure blood signal. However, it is possible to define ROIs for more than one vessel (e.g. right and left carotid artery and both vertebral arteries) and to merge them with ROI/Merge.
? CA: Perform a cluster analysis on the GpDF-maps and select the cluster with the largest signal amplitude. This function will perform a cluster analysis on the three parameters (height, width, delay). Cluster analysis can be employed for multivariate statistical analysis. It attempts to group similar samples of a population each characterized by a set of common variables. During computation of a cluster analysis, cluster centers are formed and each sample is assigned to one cluster. Clusters and cluster centers are arranged in order to minimize the distance between samples within one cluster and maximize the distance between different clusters. The distance is defined by a metric, which was Euclidean here. The maximum number of clusters that should be created can be defined in the dialog window. Moreover, limits for each parameter can be defined. Voxels, which are outside these limits are not taken into account for the computation. Usually three non-empty clusters were defined from the pixel parameters of the AIF-slice. The one containing the pixels from feeding arteries, was characterized by high amplitudes a, high values for s and early peak times p. The second cluster contained pixels from draining veins, characterized by a late peak time. The third cluster contained pixels from tissue or small vessels.
? Actual: Use the values of the parameter field
[/quote]

[kbockhorst]

Thanks,

what units do the maps have? MTT in seconds, CBV in ml/100g, CBF in ml/100g/min? Are these the units for the other maps, too (ttp ins)?

Kurt

[wullemom]

The unit of the TTP map is sec after steady state. The units for the other parameter maps are as expected:
MTT: sec
rCBV ml/100g
rCBF ml/100g/min

[kbockhorst]

The ctc map is – I guess- the most simple map to generate. I do not need any AIF. Still, this is the only window, which remains blank, better, homogeneous grey.

Can I read the Bruker files directly into the perfusion analysis or do I have to import them first into Analyze format?

On page 52 of the manual, last line, it says the files are stored on disk as GpDF_a, GpDF_t, GpDF_d and GpDF_c. The files on my Tiger Mac (it finally works, thanks again!), my Linux Redhat and WindowsXP end atd_a, atd_t, .. That’s just the same thing, right?

[wullemom]

The CTC does not need the AIF. It simply transform signal values to concentration values by calculating somthing like exp(-Signal/Signal0).

Perfusion analysis can just be done for data, which was already stored in the database.

The maps are indeed the same

[kbockhorst]

‘Perfusion analysis can just be done for data, which was already stored in the database’ was your response to my question. Does this mean, one can not directly use Bruker’s 2dseq files but has to transform them into another format, i.e. Analyze format. Do you understand this as ‘storing it in the database’? Or how would I do this?

[kbockhorst]

My results are still far off the expected values. here is what I do:

transform 2dseq into analyze format
load scan
start DSC-PWI
adjust the intensity threshold from 200 to 5 (values range between 0 and 128)
I set TE, TR, cutoff time 40 s, baseline end 22, amplitude cutoff 0.3, … and create maps with procedures/atd model/map. I get the four files ending
…c, …a, ….t, .. d.
I use tools/probe AIF and get the response cluster analysis not possible, number of valid clusters only 0.

Any suggestions what went wrong? Is this a common error message?

[wullemom]

There might be two explanations:
1) the atd-values for a potential AIF-voxel are out of the expected range. In that case adjust min/max-values in the table below.
2) the data and therefore the atd-parameter maps are of low quality: Movement of the head, to high concentration of CA, low resolution, strong paramagnetic shift effect. If you like, you can send one dataset to us for inspection.

Please note: absolute quantification of perfusion parameters depends critically on the AIF. If it is only a little bit distorted you will obtain values which are far off from what you expected. For this reason we have designed a robust sequence with a very short TE just for the AIF slice (Rausch et al. MRI, 200x).

[kbockhorst]

I would appreciate a lot if you could have a look at my data. It is a 15 MB tarred and zipped file, containing everything beside the FID file. Please tell me, where to send it to.

[wullemom]

Hi,

We’ll email you details on how to upload the file to our ftp server.

[kbockhorst]

Hi, I emailed the files in 5 pieces and also ftp-ed a tarred and gzipped file to your server a week ago. Did you have time to look at it? Are your results of our data close to that what is expected?

[kbockhorst]

You mentioned a paper from Rausch et al about a special sequence for recording perfusion data. Could you please give me the year, when it was published (you wrote MRI 200X)? Thanks.

[kbockhorst]

Hi,

did you have the chance to look at my data I sent you about 2 weeks ago? I am a little bit under pressure to evaluate my data. I would like to use BioMap but if it is not suited for my data I need to find something else. Would you please be so kind and give me a note whether your software is good for my data or not?

Kurt

[Author]

Posts