Software release: "Karawun" - convert MRtrix tractography studies into DICOM for use in surgical navigation software

Hi all,

We are excited to announce our software release, called “Karawun” to the MRtrix community.

This software allows you to perform conversion of MRtrix tractography along with other Nifti files (e.g. brain and lesion mask, thresholded fMRI blobs…etc) to Brainlab navigation (iplan-cranial) compatible DICOM format. Thus enable intraoperative image guidance using tractography reconstructed from MRtrix.

We hope this tool can encourage research and clinical uptake of MRtrix and CSD-based tractography in the area of neursurgery/surgical image-guidance.

Please spread the words!

For those who are interested, “Karawun” is a grass used as a source of fibre by Aboriginal tribes in south-eastern Australia. We are adding an Aussie twist to the name…

Cheers

Dr. Joseph Yang and A/Prof. Richard Beare
Royal Children’s Hospital and Murdoch Children’s Research Institute, Melbourne, Australia

4 Likes

Silly me, forgot to attached the github link

Here you go:

2 Likes

Dear Joseph,

thank you very much for this great news.

I was wondering if there’s a chance that the resultant DICOM track file can be imported in navigation software from other vendors as well. May I kindly ask for your guess? :slight_smile:

Thank you in advance.

Best regards

Samuel Holly
Jessenius - diagnostic centre
Nitra, Slovakia

Hi, Samuel,

Many thanks for your post.
This version of the software release is specifically for Brainlab - which is the resident neuronavigation software used at our hospital.

It is possible our current DICOM converter may work for other navigation vendors, if they use the same DICOM format as Brainlab. We have not looked into these other navigation platforms just yet, but anticipate to work on them next if there is a demand, with Medtronic Stealth being the next obvious one to target.

Hope this answers your question

Cheers

Joseph

1 Like

Hi,

thank you for making available karawun. This is definitively needed.

I would like to know whether is works with Brainlab iPlan view/cranial version 3.0 and/or with Philips dicom data as template?

Thx,
Best regards,

Stefan.

Hello Joseph,

I know this is an old thread, but I am curious to know if tractography results obtained from MRTrix/Karawun can be exported to Medtronic’s Stealth navigation system for surgical planning purposes. Based on your previous post, it looks like this was a potential plan a couple years ago. Can you provide an update on this?

Thanks,

Travis

Hello Stefan

Apology that this reply comes way overdue. I had not been paying attention with this old MRtrix thread, and only discovered your post today when I am replying to another query posted yesterday! My sincere apology!

We have Siemens scanner in-house, so naturally all the testing were performed using Siemens data, so I am not sure about compatibility with Philips DICOM data. If you have access to a Philip scanner and Brainlab, I think it is something that would be very easy to test. I will be very interested to learn the outcome of your testing!

thanks again for your post,

Joseph

Hello, Trevis

Many thanks for your interest and query. Unfortunately, we had not pursued the direction of developing Karawun for other vendors yet, including medtronic. This was largely due to the focus of our development was catering to our institution’s clinical service, which uses Brainlab.
We don’t have the facilities and contact with Medtronic reps to perform the testings. We are very much open to talking about what is possible with others, if there are interests and demand.

As far as I know, medtronic stealth displays tracts as voxel-based tract masks. I am not sure if it has a streamline display option? Are you an experienced metronic stealth user? if you are, one thing you could test, would be to use Karawun to convert binarised tract mask as a labelled image in DICOM format, and try to upload it on medtronic to check if it works and displays properly?

Many thanks, Joseph

Thanks for your response. I am not an experienced metronic stealth user but in the process of learning. I have been successful at identifying tracts using DSI studio and “burning” the tracts into patients’ t1 dicoms, which can then be labeled with StealthViz and loaded into the main stealth station. I am currently trying to compare MRtrix with DSI studio. Both seem to me to be superior to Medtronic’s StealthDTI software for identifying major tracts.

I will try to convent binarised tracts as labelled image in DICOM format and report back (will probably be a couple weeks before I can get to this). I don’t see a reason why this would not work. Thanks for the tip.

Lastly, both MRtrix and DSI studio seem like great programs. Can you speak to any specific advantages MRtrix has over DSI studio? Based on my reading thus far, the ability to correct for edema (or at least interface with FSL to correct for edema) is one apparent advantage of MRtrix in my opinion. Are there more?

Thanks,

Travis

Hello, Travis

Unfortunately, I don’t have any experience using DSI studio, so it would be difficult for me to comment on the difference between the two software. As far as I know, both software are capable of processing HARDI type of data and higher order diffusion model. In theory, they should both be superior to StealthDTI software that performs FACT tracking and using DTI model, which has well recognised limitation of unable to model for and resolve crossing-fibers, which are presented in >90% of brain WM voxels (see Ben Jeurissen’s paper).

I have recently written and published a comprehensive review on tractography in neurosurgery (Diffusion MRI tractography for neurosurgery: the basics, current state, technical reliability and challenges - IOPscience); and we have also recently submitted our Karawun preprint (Karawun: assisting evaluation of advances in multimodal imaging for neurosurgical planning and intraoperative neuronavigation | medRxiv), you may want to check out both for more information slightly_smiling_face:

Regarding tracking through oedema, you are correct that the multi-tissue CSD model from MRtrix, in theory should provide more accurate FOD estimation over the oedematous WM region compared to conventional CSD approach and DTI, because the modeling itself allows for specific tissue responses, including a “CSF-like” response function. So in a voxel contained oedematous WM, it would have a response function mixture of both CSF-like and WM-like response function. However, we should all be reminded there are multitudes of factors influencing tractography results (see my review article for detailed infor), so it is only one factor that determines your tracking results.

If you want, we are happy to provide further assistance/advice for your project. Perhaps, after you have tried Karawun conversion on Stealth, we can continue the discussion over email. You can email both myself and A/Prof Richard Beare, my senior colleague who is the co-founder and co-developer of Karawun
My email is: joseph.yang4@rch.org.au
richard.beare@mcri.edu.au

Many thanks,

Joseph

Dear Joseph,

Your tool aims to solve a very important problem in which I personally spent some time in the past. Our lab preprocess DWIs and several fMRI protocols (verbal, motor, EEG-fMRI, and so on) but is dependent on the poor BrainLab possibilities in terms of paradigm design and preprocessing. I will try your tool for tractography and hoping it will be compatible with fMRI (already, or soon). Congrats!
Best regards,
Brunno.

Thanks, Brunno!

Many thanks for your post and interest in Karawun. Karawun is capable of converting more than just tck tractography streamlines from MRtrix, also a range of imaging files (fMRI, PET, SPECT…as long as they are in Nifti format) and binarised masks (such as any labelled segmentation fo anything, i.e. tumour , cortical parcel…etc, again in Nifti format). The fMRI activation maps can be converted using Karawun and imported into Brainlab either as an image file (i.e a thresholded activation map), or as a labelled image (i.e. binarised the thresholded activation map). We routinely convert our language fMRI together with our tractography for intraoperative guidance.

Feel free to drop me a line anytime. Any user feedback about Karawun would be much appreciated. We also recently put out a preprint for Karawun (https://www.medrxiv.org/content/10.1101/2021.09.09.21262253v1) in case you or your lab is interested.

have a nice day

Joseph

1 Like

Hello, Travis

Unfortunately, I don’t have any experience using DSI studio, so it would be difficult for me to comment on the difference between the two software. As far as I know, both software are capable of processing HARDI type of data and higher order diffusion model. In theory, they should both be superior to StealthDTI software that performs FACT tracking and using DTI model, which has well recognised limitation of unable to model for and resolve crossing-fibers, which are presented in >90% of brain WM voxels (see Ben Jeurissen’s paper).

I have recently written and published a comprehensive review on tractography in neurosurgery (Diffusion MRI tractography for neurosurgery: the basics, current state, technical reliability and challenges - IOPscience); and we have also recently submitted our Karawun preprint (Karawun: assisting evaluation of advances in multimodal imaging for neurosurgical planning and intraoperative neuronavigation | medRxiv), you may want to check out both for more information slightly_smiling_face:

Regarding tracking through oedema, you are correct that the multi-tissue CSD model from MRtrix, in theory should provide more accurate FOD estimation over the oedematous WM region compared to conventional CSD approach and DTI, because the modeling itself allows for specific tissue responses, including a “CSF-like” response function. So in a voxel contained oedematous WM, it would have a response function mixture of both CSF-like and WM-like response function. However, we should all be reminded there are multitudes of factors influencing tractography results (see my review article for detailed infor), so it is only one factor that determines your tracking results.

If you want, we are happy to provide further assistance/advice for your project. Perhaps, after you have tried Karawun conversion on Stealth, we can continue the discussion over email. You can email both myself and A/Prof Richard Beare, my senior colleague who is the co-founder and co-developer of Karawun
My email is: joseph.yang4@rch.org.au
richard.beare@mcri.edu.au

Many thanks,

Joseph

I know this is an old thread, but perhaps someone could answer my question: is it possible to perform tractography using the Brainlab internal tractography tool with the diffusion images imported from Karawun?

Hi,
Thanks for the interest in Karawun. Unfortunately the answer to that is no - see the “what Karawun is not section” in the publication (Karawun: a software package for assisting evaluation of advances in multimodal imaging for neurosurgical planning and intraoperative neuronavigation | SpringerLink).

In summary, Karawun can create 3 types of dicom - standard MR (for representing T1, FA, FLAIR etc) and two types of segmentation images (one for streamlines and one for label images or rois). We don’t attempt to recreate the dicom structures needed to represent a diffusion acquisition.

Unfortunately this does make it hard to compare results in a fine-grained way - e.g. brainlab tractography applied to state of the art distortion corrected data vs data without distortion correction.