Dear Anastasia,
I’ve been looking at a lot of Tracula path.pd files and I’ve found that some probability distributions are only a single voxel wide, similar to the path.map file. The few none-zero voxels in these path.pd files have very high probability values. When an isosurface is generated for these tracts, it looks like a short thin blob somewhere in the usual tract distribution. I’ve seen descriptions in the archives of similar “short thin tracts,” but, from what I have seen, no one has offered a satisfying explanation for why these occur.
What I think is happening in these tracts is that a maximum-probability (or local maximum) path is found during a burn-in iteration and all following perturbations of that path are rejected. Since the probability value in the path.pd is equal to the number of sample paths intersecting that voxel, finding a local maximum early on results in a small number of very high-probability voxels. Consistent with this explanation, I’ve found that this issue occurs more frequently when nburnin is set to 1000 (default = 200). A similar issue can occur if a local maximum is found early during the sample iterations, and this results in a path.pd file containing a small number of voxels with very high values surrounded by a larger area of low-value voxels. When a 20% threshold is applied, the result is the same as when a local maximum occurs during a burn-in iteration.
Does my understanding of this issue seem correct?
None of this would be a problem if my only aim were to find the single path with the maximum a posteriori probability, but I’m concerned that the average and weighted_average sats for these tracts will be less accurate. Since these distributions include small fractions of the number of voxels included in most tract distributions, is it likely that the average and weighted_average stats from these narrow distributions are less representative of the whole tract and more subject to random noise?
Given these concerns, what type of overall path statistics do you think is most descriptive of a tract? Also, do you feel that higher nburnin and nsample values should lead to superior results? I would have thought this to be the case, but now it seems to me that setting either of these values too high will result in narrow probability distributions and bad statistics.
Thank you, Dillan
Hi Dillan - There's a work-around for this, see the reinit variable at the bottom of the sample config file: http://surfer.nmr.mgh.harvard.edu/fswiki/dmrirc
I'm hoping to make this happen automatically soon!
Best,
a.y
On Tue, 23 Aug 2016, Newbold, Dillan wrote:
Dear Anastasia,
I’ve been looking at a lot of Tracula path.pd files and I’ve found that some probability distributions are only a single voxel wide, similar to the path.map file. The few none-zero voxels in these path.pd files have very high probability values. When an isosurface is generated for these tracts, it looks like a short thin blob somewhere in the usual tract distribution. I’ve seen descriptions in the archives of similar “short thin tracts,” but, from what I have seen, no one has offered a satisfying explanation for why these occur.
What I think is happening in these tracts is that a maximum-probability (or local maximum) path is found during a burn-in iteration and all following perturbations of that path are rejected. Since the probability value in the path.pd is equal to the number of sample paths intersecting that voxel, finding a local maximum early on results in a small number of very high-probability voxels. Consistent with this explanation, I’ve found that this issue occurs more frequently when nburnin is set to 1000 (default = 200). A similar issue can occur if a local maximum is found early during the sample iterations, and this results in a path.pd file containing a small number of voxels with very high values surrounded by a larger area of low-value voxels. When a 20% threshold is applied, the result is the same as when a local maximum occurs during a burn-in iteration.
Does my understanding of this issue seem correct?
None of this would be a problem if my only aim were to find the single path with the maximum a posteriori probability, but I’m concerned that the average and weighted_average sats for these tracts will be less accurate. Since these distributions include small fractions of the number of voxels included in most tract distributions, is it likely that the average and weighted_average stats from these narrow distributions are less representative of the whole tract and more subject to random noise?
Given these concerns, what type of overall path statistics do you think is most descriptive of a tract? Also, do you feel that higher nburnin and nsample values should lead to superior results? I would have thought this to be the case, but now it seems to me that setting either of these values too high will result in narrow probability distributions and bad statistics.
Thank you, Dillan
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Hi Anastasia, My interpretation of the “reinit” parameter is that it is for situations where a narrow probability distribution is assumed to be incorrect. But how do you know whether it is indeed incorrect, or whether in fact the true equilibrium distribution is (correctly) very narrow?
In particular, my understanding of MCMC is that higher burn-in, and more sampling iterations are only a “good” thing. i.e., If we have the time and compute resources, we shouldn’t hesitate to increase them from their defaults, to help to make sure we are capturing the true equilibrium distribution. So, if increasing the nburnin and nsample values makes it more likely to find spatially narrow tract distributions, isn’t that a sign that the true distribution should indeed be narrow?
thanks, -MH
-- Michael Harms, Ph.D.
----------------------------------------------------------- Conte Center for the Neuroscience of Mental Disorders Washington University School of Medicine Department of Psychiatry, Box 8134 660 South Euclid Ave.Tel: 314-747-6173 St. Louis, MO 63110Email: mharms@wustl.edu
On 8/23/16, 4:44 PM, "freesurfer-bounces@nmr.mgh.harvard.edu on behalf of Anastasia Yendiki" <freesurfer-bounces@nmr.mgh.harvard.edu on behalf of ayendiki@nmr.mgh.harvard.edu> wrote:
Hi Dillan - There's a work-around for this, see the reinit variable at the bottom of the sample config file: http://surfer.nmr.mgh.harvard.edu/fswiki/dmrirc
I'm hoping to make this happen automatically soon!
Best,
a.y
On Tue, 23 Aug 2016, Newbold, Dillan wrote:
Dear Anastasia,
I’ve been looking at a lot of Tracula path.pd files and I’ve found that some probability distributions are only a single voxel wide, similar to the path.map file. The few none-zero voxels in these path.pd files have very high probability values. When an isosurface is generated for these tracts, it looks like a short thin blob somewhere in the usual tract distribution. I’ve seen descriptions in the archives of similar “short thin tracts,” but, from what I have seen, no one has offered a satisfying explanation for why these occur.
What I think is happening in these tracts is that a maximum-probability (or local maximum) path is found during a burn-in iteration and all following perturbations of that path are rejected. Since the probability value in the path.pd is equal to the number of sample paths intersecting that voxel, finding a local maximum early on results in a small number of very high-probability voxels. Consistent with this explanation, I’ve found that this issue occurs more frequently when nburnin is set to 1000 (default = 200). A similar issue can occur if a local maximum is found early during the sample iterations, and this results in a path.pd file containing a small number of voxels with very high values surrounded by a larger area of low-value voxels. When a 20% threshold is applied, the result is the same as when a local maximum occurs during a burn-in iteration.
Does my understanding of this issue seem correct?
None of this would be a problem if my only aim were to find the single path with the maximum a posteriori probability, but I’m concerned that the average and weighted_average sats for these tracts will be less accurate. Since these distributions include small fractions of the number of voxels included in most tract distributions, is it likely that the average and weighted_average stats from these narrow distributions are less representative of the whole tract and more subject to random noise?
Given these concerns, what type of overall path statistics do you think is most descriptive of a tract? Also, do you feel that higher nburnin and nsample values should lead to superior results? I would have thought this to be the case, but now it seems to me that setting either of these values too high will result in narrow probability distributions and bad statistics.
Thank you, Dillan
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
________________________________ The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.
Hi Michael - Even if it starts very close to the true max of the distribution, in practice it'll never stay put. MCMC will accept a new sample path with probability 1 if the new sample has greater probability than the previous sample, and it will also accept it with some very small probability if it doesn't. This means that it'll still explore the space around the max, even if it ends up returning to the max pretty quickly. So most likely there's something weird about the initial path if it doesn't move at all.
Best,
a.y
On Tue, 23 Aug 2016, Harms, Michael wrote:
Hi Anastasia, My interpretation of the “reinit” parameter is that it is for situations where a narrow probability distribution is assumed to be incorrect. But how do you know whether it is indeed incorrect, or whether in fact the true equilibrium distribution is (correctly) very narrow?
In particular, my understanding of MCMC is that higher burn-in, and more sampling iterations are only a “good” thing. i.e., If we have the time and compute resources, we shouldn’t hesitate to increase them from their defaults, to help to make sure we are capturing the true equilibrium distribution. So, if increasing the nburnin and nsample values makes it more likely to find spatially narrow tract distributions, isn’t that a sign that the true distribution should indeed be narrow?
thanks, -MH
-- Michael Harms, Ph.D.
Conte Center for the Neuroscience of Mental Disorders Washington University School of Medicine Department of Psychiatry, Box 8134 660 South Euclid Ave.Tel: 314-747-6173 St. Louis, MO 63110Email: mharms@wustl.edu
On 8/23/16, 4:44 PM, "freesurfer-bounces@nmr.mgh.harvard.edu on behalf of Anastasia Yendiki" <freesurfer-bounces@nmr.mgh.harvard.edu on behalf of ayendiki@nmr.mgh.harvard.edu> wrote:
Hi Dillan - There's a work-around for this, see the reinit variable at the bottom of the sample config file: http://surfer.nmr.mgh.harvard.edu/fswiki/dmrirc
I'm hoping to make this happen automatically soon!
Best,
a.y
On Tue, 23 Aug 2016, Newbold, Dillan wrote:
Dear Anastasia,
I’ve been looking at a lot of Tracula path.pd files and I’ve found that some probability distributions are only a single voxel wide, similar to the path.map file. The few none-zero voxels in these path.pd files have very high probability values. When an isosurface is generated for these tracts, it looks like a short thin blob somewhere in the usual tract distribution. I’ve seen descriptions in the archives of similar “short thin tracts,” but, from what I have seen, no one has offered a satisfying explanation for why these occur.
What I think is happening in these tracts is that a maximum-probability (or local maximum) path is found during a burn-in iteration and all following perturbations of that path are rejected. Since the probability value in the path.pd is equal to the number of sample paths intersecting that voxel, finding a local maximum early on results in a small number of very high-probability voxels. Consistent with this explanation, I’ve found that this issue occurs more frequently when nburnin is set to 1000 (default = 200). A similar issue can occur if a local maximum is found early during the sample iterations, and this results in a path.pd file containing a small number of voxels with very high values surrounded by a larger area of low-value voxels. When a 20% threshold is applied, the result is the same as when a local maximum occurs during a burn-in iteration.
Does my understanding of this issue seem correct?
None of this would be a problem if my only aim were to find the single path with the maximum a posteriori probability, but I’m concerned that the average and weighted_average sats for these tracts will be less accurate. Since these distributions include small fractions of the number of voxels included in most tract distributions, is it likely that the average and weighted_average stats from these narrow distributions are less representative of the whole tract and more subject to random noise?
Given these concerns, what type of overall path statistics do you think is most descriptive of a tract? Also, do you feel that higher nburnin and nsample values should lead to superior results? I would have thought this to be the case, but now it seems to me that setting either of these values too high will result in narrow probability distributions and bad statistics.
Thank you, Dillan
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Ok. We’ll give it a try. Just to confirm, you're agreeing that increasing nburnin and nsample is a “good” thing, right? (e.g., 1000, and 15000, respectively?) If anything, we should be less likely to get narrow tracts when using larger nburnin/nsample values, right?
thanks, -MH
-- Michael Harms, Ph.D.
----------------------------------------------------------- Conte Center for the Neuroscience of Mental Disorders Washington University School of Medicine Department of Psychiatry, Box 8134 660 South Euclid Ave.Tel: 314-747-6173 St. Louis, MO 63110Email: mharms@wustl.edu
On 8/23/16, 5:09 PM, "freesurfer-bounces@nmr.mgh.harvard.edu on behalf of Anastasia Yendiki" <freesurfer-bounces@nmr.mgh.harvard.edu on behalf of ayendiki@nmr.mgh.harvard.edu> wrote:
Hi Michael - Even if it starts very close to the true max of the distribution, in practice it'll never stay put. MCMC will accept a new sample path with probability 1 if the new sample has greater probability than the previous sample, and it will also accept it with some very small probability if it doesn't. This means that it'll still explore the space around the max, even if it ends up returning to the max pretty quickly. So most likely there's something weird about the initial path if it doesn't move at all.
Best,
a.y
On Tue, 23 Aug 2016, Harms, Michael wrote:
Hi Anastasia, My interpretation of the “reinit” parameter is that it is for situations where a narrow probability distribution is assumed to be incorrect. But how do you know whether it is indeed incorrect, or whether in fact the true equilibrium distribution is (correctly) very narrow?
In particular, my understanding of MCMC is that higher burn-in, and more sampling iterations are only a “good” thing. i.e., If we have the time and compute resources, we shouldn’t hesitate to increase them from their defaults, to help to make sure we are capturing the true equilibrium distribution. So, if increasing the nburnin and nsample values makes it more likely to find spatially narrow tract distributions, isn’t that a sign that the true distribution should indeed be narrow?
thanks, -MH
-- Michael Harms, Ph.D.
Conte Center for the Neuroscience of Mental Disorders Washington University School of Medicine Department of Psychiatry, Box 8134 660 South Euclid Ave.Tel: 314-747-6173 St. Louis, MO 63110Email: mharms@wustl.edu
On 8/23/16, 4:44 PM, "freesurfer-bounces@nmr.mgh.harvard.edu on behalf of Anastasia Yendiki" <freesurfer-bounces@nmr.mgh.harvard.edu on behalf of ayendiki@nmr.mgh.harvard.edu> wrote:
Hi Dillan - There's a work-around for this, see the reinit variable at the bottom of the sample config file: http://surfer.nmr.mgh.harvard.edu/fswiki/dmrirc
I'm hoping to make this happen automatically soon!
Best,
a.y
On Tue, 23 Aug 2016, Newbold, Dillan wrote:
Dear Anastasia,
I’ve been looking at a lot of Tracula path.pd files and I’ve found that some probability distributions are only a single voxel wide, similar to the path.map file. The few none-zero voxels in these path.pd files have very high probability values. When an isosurface is generated for these tracts, it looks like a short thin blob somewhere in the usual tract distribution. I’ve seen descriptions in the archives of similar “short thin tracts,” but, from what I have seen, no one has offered a satisfying explanation for why these occur.
What I think is happening in these tracts is that a maximum-probability (or local maximum) path is found during a burn-in iteration and all following perturbations of that path are rejected. Since the probability value in the path.pd is equal to the number of sample paths intersecting that voxel, finding a local maximum early on results in a small number of very high-probability voxels. Consistent with this explanation, I’ve found that this issue occurs more frequently when nburnin is set to 1000 (default = 200). A similar issue can occur if a local maximum is found early during the sample iterations, and this results in a path.pd file containing a small number of voxels with very high values surrounded by a larger area of low-value voxels. When a 20% threshold is applied, the result is the same as when a local maximum occurs during a burn-in iteration.
Does my understanding of this issue seem correct?
None of this would be a problem if my only aim were to find the single path with the maximum a posteriori probability, but I’m concerned that the average and weighted_average sats for these tracts will be less accurate. Since these distributions include small fractions of the number of voxels included in most tract distributions, is it likely that the average and weighted_average stats from these narrow distributions are less representative of the whole tract and more subject to random noise?
Given these concerns, what type of overall path statistics do you think is most descriptive of a tract? Also, do you feel that higher nburnin and nsample values should lead to superior results? I would have thought this to be the case, but now it seems to me that setting either of these values too high will result in narrow probability distributions and bad statistics.
Thank you, Dillan
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
________________________________ The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.
Yes, it shouldn't hurt.
On Tue, 23 Aug 2016, Harms, Michael wrote:
Ok. We’ll give it a try. Just to confirm, you're agreeing that increasing nburnin and nsample is a “good” thing, right? (e.g., 1000, and 15000, respectively?) If anything, we should be less likely to get narrow tracts when using larger nburnin/nsample values, right?
thanks, -MH
-- Michael Harms, Ph.D.
Conte Center for the Neuroscience of Mental Disorders Washington University School of Medicine Department of Psychiatry, Box 8134 660 South Euclid Ave.Tel: 314-747-6173 St. Louis, MO 63110Email: mharms@wustl.edu
On 8/23/16, 5:09 PM, "freesurfer-bounces@nmr.mgh.harvard.edu on behalf of Anastasia Yendiki" <freesurfer-bounces@nmr.mgh.harvard.edu on behalf of ayendiki@nmr.mgh.harvard.edu> wrote:
Hi Michael - Even if it starts very close to the true max of the distribution, in practice it'll never stay put. MCMC will accept a new sample path with probability 1 if the new sample has greater probability than the previous sample, and it will also accept it with some very small probability if it doesn't. This means that it'll still explore the space around the max, even if it ends up returning to the max pretty quickly. So most likely there's something weird about the initial path if it doesn't move at all.
Best,
a.y
On Tue, 23 Aug 2016, Harms, Michael wrote:
Hi Anastasia, My interpretation of the “reinit” parameter is that it is for situations where a narrow probability distribution is assumed to be incorrect. But how do you know whether it is indeed incorrect, or whether in fact the true equilibrium distribution is (correctly) very narrow?
In particular, my understanding of MCMC is that higher burn-in, and more sampling iterations are only a “good” thing. i.e., If we have the time and compute resources, we shouldn’t hesitate to increase them from their defaults, to help to make sure we are capturing the true equilibrium distribution. So, if increasing the nburnin and nsample values makes it more likely to find spatially narrow tract distributions, isn’t that a sign that the true distribution should indeed be narrow?
thanks, -MH
-- Michael Harms, Ph.D.
Conte Center for the Neuroscience of Mental Disorders Washington University School of Medicine Department of Psychiatry, Box 8134 660 South Euclid Ave.Tel: 314-747-6173 St. Louis, MO 63110Email: mharms@wustl.edu
On 8/23/16, 4:44 PM, "freesurfer-bounces@nmr.mgh.harvard.edu on behalf of Anastasia Yendiki" <freesurfer-bounces@nmr.mgh.harvard.edu on behalf of ayendiki@nmr.mgh.harvard.edu> wrote:
Hi Dillan - There's a work-around for this, see the reinit variable at the bottom of the sample config file: http://surfer.nmr.mgh.harvard.edu/fswiki/dmrirc
I'm hoping to make this happen automatically soon!
Best,
a.y
On Tue, 23 Aug 2016, Newbold, Dillan wrote:
Dear Anastasia,
I’ve been looking at a lot of Tracula path.pd files and I’ve found that some probability distributions are only a single voxel wide, similar to the path.map file. The few none-zero voxels in these path.pd files have very high probability values. When an isosurface is generated for these tracts, it looks like a short thin blob somewhere in the usual tract distribution. I’ve seen descriptions in the archives of similar “short thin tracts,” but, from what I have seen, no one has offered a satisfying explanation for why these occur.
What I think is happening in these tracts is that a maximum-probability (or local maximum) path is found during a burn-in iteration and all following perturbations of that path are rejected. Since the probability value in the path.pd is equal to the number of sample paths intersecting that voxel, finding a local maximum early on results in a small number of very high-probability voxels. Consistent with this explanation, I’ve found that this issue occurs more frequently when nburnin is set to 1000 (default = 200). A similar issue can occur if a local maximum is found early during the sample iterations, and this results in a path.pd file containing a small number of voxels with very high values surrounded by a larger area of low-value voxels. When a 20% threshold is applied, the result is the same as when a local maximum occurs during a burn-in iteration.
Does my understanding of this issue seem correct?
None of this would be a problem if my only aim were to find the single path with the maximum a posteriori probability, but I’m concerned that the average and weighted_average sats for these tracts will be less accurate. Since these distributions include small fractions of the number of voxels included in most tract distributions, is it likely that the average and weighted_average stats from these narrow distributions are less representative of the whole tract and more subject to random noise?
Given these concerns, what type of overall path statistics do you think is most descriptive of a tract? Also, do you feel that higher nburnin and nsample values should lead to superior results? I would have thought this to be the case, but now it seems to me that setting either of these values too high will result in narrow probability distributions and bad statistics.
Thank you, Dillan
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Anastasia Yendiki -
Harms, Michael -
Newbold, Dillan