LV Mechanics Challenge - Discussion LV Mechanics Challenge - Discussion
Combination View Flat View Tree View
LV Mechanics Results
toggle
Dear Vicky, Alistair and Martyn,

Thank you for your last message! This is to confirm that we at McGill are ok with your plan to post current simulation results on the website with participants anonymized. We are working to update the workshop paper to include better simulation results. We understand that you want to save the comparison results that you compiled for the journal version of the article

We are aware that some of our errors (in position estimation) particularly for two of the canine data sets might have to do with the way we handled/extrapolated position constraints based on the 4 points at the base for which you provided position coordinates for all frames.

Given that you would like us to update our results (for the LNCS version) one thing I wanted to ask is should we produce "improved" results (for the LNCS version) how can we tell whether they are in fact improved without comparison against the ground truth?

Best,
Kaleem Siddiqi
(McGill)
Flag Flag
RE: LV Mechanics Results
10/27/14 12:17 PM as a reply to Anonymous.
Dear Kaleem

Thank you for posting questions on this forum.

Regarding the reassessment of your updated results, it was a very good suggestion and we are currently discussing this among ourselves as we have some internal deadlines to meet, so we need to figure out how to prioritise this in order for you to meet your deadline too.

Regarding the displacement boundary conditions, the displacement for each node at ED is calculated with respect to the diastasis frame (reference frame) and the displacement at ES is calculated with respect to the ED frame. Does this make sense? If not, please write back to me and I can further clarify.

Vicky

Anonymous:
Dear Vicky, Alistair and Martyn,

Thank you for your last message! This is to confirm that we at McGill are ok with your plan to post current simulation results on the website with participants anonymized. We are working to update the workshop paper to include better simulation results. We understand that you want to save the comparison results that you compiled for the journal version of the article

We are aware that some of our errors (in position estimation) particularly for two of the canine data sets might have to do with the way we handled/extrapolated position constraints based on the 4 points at the base for which you provided position coordinates for all frames.

Given that you would like us to update our results (for the LNCS version) one thing I wanted to ask is should we produce "improved" results (for the LNCS version) how can we tell whether they are in fact improved without comparison against the ground truth?

Best,
Kaleem Siddiqi
(McGill)
Flag Flag
RE: LV Mechanics Results
11/6/14 9:58 AM as a reply to Anonymous.
Dear Kaleem

Regarding your request for reassessing updated simulation results, we had an internal discussion and unfortunately, we won't able to do this before November 16th. In terms of your current models, I think the error is perhaps largely due to a global translation in the long-axis. If you can ensure that your present simulation closely adhered to the boundary conditions we provided, I am sure your simulated models will be improved.

After November 16th, we will carry out comparisons of the new results from all participants with the ground truth, and the outcome of that comparison will likely to be released early next year.

Hope you can understand.

Vicky

Anonymous:
Dear Kaleem

Thank you for posting questions on this forum.

Regarding the reassessment of your updated results, it was a very good suggestion and we are currently discussing this among ourselves as we have some internal deadlines to meet, so we need to figure out how to prioritise this in order for you to meet your deadline too.

Regarding the displacement boundary conditions, the displacement for each node at ED is calculated with respect to the diastasis frame (reference frame) and the displacement at ES is calculated with respect to the ED frame. Does this make sense? If not, please write back to me and I can further clarify.

Vicky

Anonymous:
Dear Vicky, Alistair and Martyn,

Thank you for your last message! This is to confirm that we at McGill are ok with your plan to post current simulation results on the website with participants anonymized. We are working to update the workshop paper to include better simulation results. We understand that you want to save the comparison results that you compiled for the journal version of the article

We are aware that some of our errors (in position estimation) particularly for two of the canine data sets might have to do with the way we handled/extrapolated position constraints based on the 4 points at the base for which you provided position coordinates for all frames.

Given that you would like us to update our results (for the LNCS version) one thing I wanted to ask is should we produce "improved" results (for the LNCS version) how can we tell whether they are in fact improved without comparison against the ground truth?

Best,
Kaleem Siddiqi
(McGill)
Flag Flag
RE: LV Mechanics Results
11/11/14 4:32 AM as a reply to Anonymous.
Dear Vicky and co-organisers of the mechanical challenge,

we are currently running the simulations to complete the three dogs we did not process in time for the workshop in Boston. We will try to send you the 16th of November the LNCS version including them.

As far as we've understood from your e-mail, you plan to gather results you will have the 16th of November as the final ones for the journal version. If this is right, we think it would be nice to have a bit more time, maybe until the end of the year, to play with the data and then start processing and writing beginning of next year. We do not see such a hurry for the journal version and it would be fairer for the participants. Besides, we need to think if just a collation of the results will be enough for a good journal. Which is going to be the message of the paper? Something that could help cardiac mechanics modellers would be desirable. For doing so, we should try to think if there is a better way to present the work of everyone rather than just serially.

Again, thanks a lot for your huge effort on leading this.

Take care,

Oscar, Jazmin, Mariano
Flag Flag
RE: LV Mechanics Results
11/13/14 10:37 PM as a reply to Anonymous.
Hi Oscar, Jazmin, Mariano,

I understand your situation, and I for one would like requests like this to be accommodated (as we did during the original submission process for this LV mechanics challenge). However, this raises several potential issues, and there seem to be differing views on such extensions amongst participants.

One issue with this approach, as we discovered during the STACOM workshop, is that the results submitted by some groups were generated using methods that were not fully documented in their STACOM workshop papers (which had been submitted several weeks earlier). I recall that in some cases different boundary constraints were used, and in other cases different fibre orientation fields were used. If the final submitted results end up being generated using undocumented methods, then this would most likely cause a problem for publication of a joint paper based on those results. The submission date for the final workshop papers for the LNCS volume is a hard deadline, so with the above in mind it was decided that the same hard deadline should apply for the final submission of results. Vicky has tried to give plenty of prior-warning of the deadline over email.

Following the workshop, a suggestion was made to submit a paper based on the original submissions of results, in order to fully document the scientific process and developments during this challenge. In my opinion (and I think I speak for Vicky/Alistair), there is certainly scientific merit to this approach from a learning perspective, however I do not believe that multiple sets of collated results would sit well in a good international journal. We agreed that we could present the final submissions in the joint journal paper, while publishing anonymised comparison results using the originally submitted results on the challenge website.

In addition to this, I think it would be great if there was a way that participants could add their learning perspectives associated with the challenge to the website (e.g. via this forum) for the wider benefit of the community, but we shall see what eventuates.

Best wishes,
Martyn.

PS. One thing I ask is for everyone to 'play nicely' on this rather open forum :-) Standard community forum rules apply please - keep it positive, etc. It'd be a shame to have to moderate the forum,but there is a risk that anonymous posts are allowed at present. I felt such a good, open-natured buzz at the STACOM workshop and would like for us to do whatever we can as a community to protect and foster this style of communication.
Flag Flag
Organisers Organisers
  1. Alistair Young
    (University of Auckland, NZ)
  2. Martyn Nash
    (University of Auckland, NZ)
  3. Vicky Wang
    (University of Auckland, NZ)

 

Silver Sponsors

Elite Sponsors