Dear Greg,
probably some of the strong signal inhomogeneities are causing false detection of WMHs. CAT12 is only using T1 images to detect WMHs which is far from being perfect and mainly thought to obtain a more reliable registration in the presence of WMHs.
For young subjects, it's very unlikely that WMHs occur. However, an example image with the GM segmentation or label would be helpful to say more.
Best,
Christian
On Thu, 11 Mar 2021 00:52:48 +0000, Kronberg, Greg <[log in to unmask]> wrote:
>Thanks again Christian for the speedy response.
>Is there any other potential explanation for these WMH's. The ages in our study range from 21 to 60 (mean: 41, sd:8), but I am seeing the WMH just as prominently in the youngest subjects (at least based on the pdf reports). Either way I can set the WMH's to be corrected to WM to be safe
>
>I can't thank you enough for the help here
>
>-----Original Message-----
>From: Christian Gaser <[log in to unmask]>
>Sent: Wednesday, March 10, 2021 6:15 PM
>To: Kronberg, Greg <[log in to unmask]>
>Cc: [log in to unmask]
>Subject: Re: CAT12 poor bias grade/rating
>
>Dear Greg,
>
>On 10 Mar 2021, at 23:43, Kronberg, Greg wrote:
>
>> Thank you Christian for the quick and very helpful response!
>>
>> I'm also noticing in the reports that there are some clusters flagged
>> as "uncorrected WMHs=GM!"
>> Are there any best practices for how to handle these at the moment?
>Sounds like you have older subjects in your sample. The occurrence of WMHs increases with age. Internally these WMHs are corrected to obtain a more reliable registration. For the statistical analysis uncorrected WMHs are kept as GM and will increase local variance. Because the location is in WM this will usually not affect GM analysis. To be on the safe side you can try to set WMH correction to „2“ to correct WMHs to WM.
>
>Best,
>
>Christian
>>
>> Thanks again!
>>
>> -----Original Message-----
>> From: Christian Gaser <[log in to unmask]>
>> Sent: Wednesday, March 10, 2021 3:34 AM
>> To: [log in to unmask]; Kronberg, Greg <[log in to unmask]>
>> Subject: Re: CAT12 poor bias grade/rating
>>
>> USE CAUTION: External Message.
>>
>> Dear Greg,
>>
>> On Tue, 9 Mar 2021 15:55:47 +0000, Kronberg, Greg
>> <[log in to unmask]> wrote:
>>
>>> Dear CAT12 users,
>>>
>>> Im using CAT12 version 12.7 with SPM12 version 7771 for a VBM
>>> analysis. After the segmentation step for most images, the cat12
>>> automated report has poor ratings for the bias under "Image and
>>> Processing Quality." The ratings are typically around 65-70% with a
>>> grade of D+, which shows up in an ominous red color. I tried
>>> adjusting the "Power of SPM Inhomogeneity Correction" to strong and
>>> heavy, as well as affine preprocessing to full, but this didn't seem
>>> to help with the bias rating.
>>
>> The estimated intensity inhomogeneity effects (bias) can be corrected
>> very efficiently with CAT12 or SPM12 and this affects weighted average
>> image quality rating (IQR) only o a minor extent. The bias rating will
>> be unaffected by any settings in CAT12 because these parameters are
>> obtained for the image before any preprocessing:
>> https://urldefense.proofpoint.com/v2/url?u=http-3A__www.neuro.uni-2Djena.de_cat12-2Dhtml_cat-5Fmethods-5FQA.html&d=DwIFaQ&c=shNJtf5dKgNcPZ6Yh64b-A&r=XYZjF4kXjBrTJFN4r0Z3TRY3MRfL10a7IBqBj6s75mk&m=yjeR6EKR60-3JcMHUvWZaw1CX149DZ5bDQhk1CYQ_Ss&s=Mtnuxgtwc2p4hMs8FVtfN1ujYo0zyfqJRRoQCcVnPLk&e=
>>
>> Thus, I would keep the inhomogeneity corrections to the default value
>> or probably change it to "strong", but would not change affine
>> preprocessing parameters.
>>
>>>
>>> The scans were collected a skyra 3T, 32 channel coil, at 0.8x0.8x0.8
>>> mm resolution, without normalization filters.
>> Unfortunately, the default Siemens settings are not that appropriate.
>> "Prescan Normalization" has a huge effect on your image inhomogeneity
>> and it's damn efficient. However, it will be difficult to change that
>> parameter for future acquisitions if you intend to mix the data. This
>> might systematically affect your statistical analysis. If you don't
>> mix old and new data enabling prescan normalization might be a good
>> idea to largely improve image quality.
>>
>>>
>>> I'm looking for some guidance on 1) how this rating is determined 2)
>>> any specific steps I can take to improve it 3) whether the results
>>> are usable as they are.
>> 1)
>> https://urldefense.proofpoint.com/v2/url?u=http-3A__www.neuro.uni-2Djena.de_cat12-2Dhtml_cat-5Fmethods-5FQA.html&d=DwIFaQ&c=shNJtf5dKgNcPZ6Yh64b-A&r=XYZjF4kXjBrTJFN4r0Z3TRY3MRfL10a7IBqBj6s75mk&m=yjeR6EKR60-3JcMHUvWZaw1CX149DZ5bDQhk1CYQ_Ss&s=Mtnuxgtwc2p4hMs8FVtfN1ujYo0zyfqJRRoQCcVnPLk&e=
>> 2) see above
>> 3) don't care too much if the IQR is fine
>>
>> Best,
>>
>> Christian
>>>
>>> I'm new to neuroimaging in general, so I may be missing something
>>> very basic.
>>> Any help would be greatly appreciated!
>>>
>>> Greg
>>>
>>>
>>> Greg Kronberg, PhD
>>> Postdoctoral Fellow
>>> Neuropsychoimaging of Addiction and Related Conditions (NARC) Lab
>>> Department of Psychiatry, Icahn School of Medicine at Mount Sinai
>>>
>>>
>>>
>>>
|