 |
 |
Dear Michael, thank you for your detailed response. I have just one more question. In the conference abstract of Chen, Wang and Detre (ISMRM 2011) the authors compared different M0 quantification approaches. The approach you mentioned is the M0t method with a mean lambda value between gray and white matter. In the M0tLambda method they used different lambda values for gray matter and white matter, but in this case you need the segmented images (they used the segemnted pve images with a threshold of 0.75). Do you think the effort is well worthwile or the M0t approach is quite sufficient. Best Ralf > Ralf, > Oxford_asl calculates a perfusion image in the same raw units as the > original data, if you provide it with a calibration image it will > also provide a perfusion image in absolute units (perfusion_calib), > but this is done > using the CSF as a reference in the same way that > the > GUI approaches it. If you want to use the voxelwise approach and > already have an M0 image then you can do the calibration directly > using the (uncalibrated) perfusion image provided by oxford_asl. > Something like this should work: > fslmaths perfusion -div M0 -mul 0.9 -mul <alpha> -mul 6000 perfusion_calib > Note that I have included the partition coefficient in the > calculation here to convert the M0 image from M0 of tissue into the > (approximately) equivalent arterial blood value (M0a). I also > include the inversion efficiency, you will need to choose the > correct value for your sequence (0.85 for pcASL, 0.98 for pASL). > The reason for point people to asl_calib is that in some cases it > might be necessary to do some calculations the calibration image > before using it as an M0 image - it sounds like that is not relevant > here. I will try to make that bit of the guide clearer in future. As > is so often the case this will be easier in a future release where > the voxelwise calculation is built directly into oxford_asl. > Michael > On 5 Aug 2015, at 11:37, Ralf Veit <[log in to unmask]> wrote: > Zitat von Ralf Veit <[log in to unmask]>: > Dear FSL list (and Michael), > I have a question regarding the calibration of ASL images using a > voxel wise approach (M0a). In the BASIL guide the authors write > “Voxelwise calculation of M0a cannot be carried out using Asl_gui, > use the asl_calib tool directly”, unfortunately I could not find the > parameters in ASL_calib for the voxel-wise approach. Thus before I > used (in asl_calib) the perfusion image as input, the CSF as tissue > reference with the corresponding values, the structural image, the > M0-image and a CSF mask for the M0 image (I have no calibration > image for coil sensitive correction). I found the formula CBF= > (perfusion /Moa)*6000. Thus it mean that each voxel will be divided > with its specific M0 value (not one global value). Or is it enough > to divide the M0 image with lambda 0.9 and use a global M0-value? > Why do I need ASL_calib, because perfusion image is already created > during oxford_asl. It would be nice if someone could give me the > exact approach for the voxel wise approach (which parameters I need). > > Best > Ralf > > Ralf Veit PhD > Institute of Medical Psychology and Behavioral Neurobiology > Otfried-Müller str. 47 > 72076 Tübingen > phone: +49-7071-2987703 > fax: +49-7071-295706 Ralf Veit PhD Institute of Medical Psychology and Behavioral Neurobiology Otfried-Müller str. 47 72076 Tübingen phone: +49-7071-2987703 fax: +49-7071-295706