Thank you Silvia and all
Indeed what I want to do is just a visual QC to catch at least macroscopic errors (e.g.: motion artefacts, wrong MR sequence, segmentation algorithm that completely failed due to initial completely wrong alignment etc...). The things that happens when you have a batch running blindly on 300 images.
And... I was wondering if there's a method to go through them quite fast and not wasting 1 minute per patient in just setting up visualization parameters.
Thank you
Luca
________________________________________
Da: Ingala, S. <[log in to unmask]>
Inviato: martedě 23 giugno 2020 09:32
A: Mutsaerts, H.J.M.M. (Henk Jan); PRESOTTO LUCA; Lorenzini, L. (Luigi)
Cc: [log in to unmask]
Oggetto: Re: (Semi-)automatic QC T1 segmentation
Hi Luca,
Quality check depends largely on the dataset you are analysing (age, disease, possible comorbidities,..).
We found that IQR parameter from SPM pipeline gives quite reliable results. I would suggest, nevertheless, to have visual QC next to the quantitative outcomes to understand alternative sources of variance, such as anatomical variations, incidental findings, etc.
The QC toolbox that we are developing is currently being optimized and validated, and we will make sure keep you informed with developments.
Kind regards,
Silvia and Luigi
[cid:image001.png@01D3FF42.3DFC36A0]
Silvia Ingala, MD | PhD candidate
Dept. Radiology & Nuclear Medicine
Location VUmc | PK-1X100 | De Boelelaan 1118, 1081HV Amsterdam
T: +31 647632081 | I: +31 020 4442333 | E: [log in to unmask]<mailto:[log in to unmask]>
www.amsterdamumc.nl<http://www.amsterdamumc.nl/> | www.vumc.nl<http://www.vumc.nl/>
---------------------------------------------------
Disclaimer
This e-mail is confidential and may contain information that is not intended for you. If you are not the addressee or if this message was sent to you by mistake, you are requested to inform the sender and delete the message. VU University Medical Center accepts no liability for damage of any kind resulting from the risks inherent in the electronic transmission of messages.
From: "Mutsaerts, H.J.M.M. (Henk Jan)" <[log in to unmask]>
Date: Tuesday, 23 June 2020 at 09:01
To: "[log in to unmask]" <[log in to unmask]>, Luigi Lorenzini <[log in to unmask]>, "Ingala, S." <[log in to unmask]>
Cc: "[log in to unmask]" <[log in to unmask]>
Subject: (Semi-)automatic QC T1 segmentation
Ciao Luca,
Luigi (CC) is doing a large formal comparison between quantitative QC parameters and has developed a semi-automatic toolbox that would suit your needs.
Perhaps this would be a good option for you to have a go-to solution and for Luigi to get your feedback on his tool, and validate our results in your study.
Otherwise, he can tell you the outcome of the best parameters from literature you are able to use.
Probably best to check further with him. I also CC Silvia who has worked with this from the visual QC perspective (as neuroradiologist in training).
Best wishes,
Henk
BTW: not sure which software you used for segmentation but I can recommend the newest CAT12.
------------------------------
Henk(-Jan) Mutsaerts, MD PhD
Amsterdam UMC/ UZ Ghent/UCL
Phone: +31 6 4390 8284
----------------------------------------------------------------------
Date: Mon, 22 Jun 2020 08:51:54 +0000
From: PRESOTTO LUCA <[log in to unmask]<mailto:[log in to unmask]>>
Subject: How to quality check a massive amount of segmentations?
Dear experts,
I've segmented ~250 MR 3D T1 studies (standard batch) for further analysis. I'd like to quality check visually all of them now. Using spm_check_reg on every MR and c1 image, once for each patient, then select the zoom, then select the countour and number of contours seems like a huge waste of time.
Is there a quicker way to do this? Can I script spm_check_reg so that it opens with my preferred settings two images at once and when I click somewhere it automatically goes to the next patient with again my preferred settings?
Do you have other ideas to achieve the same objective? I'd just like to check for major artefacts in the segmentation and for major errors in the MR images (e.g.: motion).
Thanks in advance
Luca
Rispetta l’ambiente: non stampare questa mail se non č necessario.
Respect the environment: print this email only if necessary.
______________________________________________________
VUmc disclaimer : www.vumc.nl/disclaimer
AMC disclaimer : www.amc.nl/disclaimer
|