New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Median otsu figures in example don't look good #321

Closed
Garyfallidis opened this Issue Jan 27, 2014 · 11 comments

Comments

Projects
None yet
4 participants
@Garyfallidis
Member

Garyfallidis commented Jan 27, 2014

Hi @matthieudumont and all,

It is difficult to see the background that was removed after brain extraction in this example
http://nipy.org/dipy/examples_built/brain_extraction_dwi.html#example-brain-extraction-dwi

also the figures are not shown in the standard way.

@matthieudumont

This comment has been minimized.

Contributor

matthieudumont commented Jan 29, 2014

Would that be a good way to view removed background?

median_otsu

Also, what would be the standard way to display the figures?

@matthieudumont

This comment has been minimized.

Contributor

matthieudumont commented Jan 29, 2014

Here's a test with rescaled values:
median_otsu_gray

Combined with jet color map:
median_otsu

@mdesco

This comment has been minimized.

Contributor

mdesco commented Jan 29, 2014

  1. Eyes should be pointing upwards
  2. Maybe try a histogram normalization before plotting. The idea is that we
    want to see the noise in the background.

Max

On Wed, Jan 29, 2014 at 9:18 AM, matthieudumont notifications@github.comwrote:

Would that be a good way to view removed background?

[image: median_otsu]https://f.cloud.github.com/assets/1280738/2030019/1f605874-88f0-11e3-85e1-c43e1c683c01.png

Also, what would be the standard way to display the figures?

Reply to this email directly or view it on GitHubhttps://github.com//issues/321#issuecomment-33587734
.

Maxime Descoteaux, PhD
Professor
Sherbrooke Connectivity Imaging Laboratory (SCIL)
Sherbrooke Molecular Imaging Center (CIMS)
Computer Science department
Sherbrooke University
2500 Boul. Université
Sherbrooke, Québec
J1K 2R1, Canada
phone: +1 819 821-8000 x66129
fax: +1 819 821-8200
http://scil.dinf.usherbrooke.ca

Scientific director
Visualization and Image Analysis Plateform (PAVI)
http://pavi.dinf.usherbrooke.ca/

@matthieudumont

This comment has been minimized.

Contributor

matthieudumont commented Jan 29, 2014

Histogram equalization:
median_otsu

@mdesco

This comment has been minimized.

Contributor

mdesco commented Jan 30, 2014

Ok. From what I see, the background around the brain is already set to
zero? This is not great for the purpose of the example.

Can you take the b0 image from the Sherbrooke data? The background would be
noise all over...

Max

On Wed, Jan 29, 2014 at 12:51 PM, matthieudumont
notifications@github.comwrote:

Histogram equalization:
[image: median_otsu_hist_eq]https://f.cloud.github.com/assets/1280738/2032356/fe2da6ca-890d-11e3-9ec6-c15dde4fb9ae.png

Reply to this email directly or view it on GitHubhttps://github.com//issues/321#issuecomment-33610376
.

Maxime Descoteaux, PhD
Professor
Sherbrooke Connectivity Imaging Laboratory (SCIL)
Sherbrooke Molecular Imaging Center (CIMS)
Computer Science department
Sherbrooke University
2500 Boul. Université
Sherbrooke, Québec
J1K 2R1, Canada
phone: +1 819 821-8000 x66129
fax: +1 819 821-8200
http://scil.dinf.usherbrooke.ca

Scientific director
Visualization and Image Analysis Plateform (PAVI)
http://pavi.dinf.usherbrooke.ca/

@matthieudumont

This comment has been minimized.

Contributor

matthieudumont commented Jan 30, 2014

Sherbrooke b0:
median_otsu_1 5t_histeq_2_1

Without histogram equalization:
median_otsu_1 5t_2_1

I had to reduce median_otsu's agressiveness to 2, 1 (instead of 4, 4) for this data.

@mdesco

This comment has been minimized.

Contributor

mdesco commented Jan 30, 2014

I let @Garyfallidis decide but I like the first set of images. We really
see the effect of the masking.

Max

On Thu, Jan 30, 2014 at 9:11 AM, matthieudumont notifications@github.comwrote:

Sherbrooke b0:
[image: median_otsu_1 5t_histeq_2_1]https://f.cloud.github.com/assets/1280738/2040543/4b0f2fea-89b8-11e3-99f0-1f7501feae1d.png

Without histogram equalization:
[image: median_otsu_1 5t_2_1]https://f.cloud.github.com/assets/1280738/2040550/558819b4-89b8-11e3-83e9-ce14d8dc53c8.png

I had to reduce median_otsu's agressiveness to 2, 1 (instead of 4, 4) for
this data.

Reply to this email directly or view it on GitHubhttps://github.com//issues/321#issuecomment-33690584
.

Maxime Descoteaux, PhD
Professor
Sherbrooke Connectivity Imaging Laboratory (SCIL)
Sherbrooke Molecular Imaging Center (CIMS)
Computer Science department
Sherbrooke University
2500 Boul. Université
Sherbrooke, Québec
J1K 2R1, Canada
phone: +1 819 821-8000 x66129
fax: +1 819 821-8200
http://scil.dinf.usherbrooke.ca

Scientific director
Visualization and Image Analysis Plateform (PAVI)
http://pavi.dinf.usherbrooke.ca/

@Garyfallidis

This comment has been minimized.

Member

Garyfallidis commented Jan 30, 2014

Yeah, go for the first one. Looks impressive! :)

@matthieudumont

This comment has been minimized.

Contributor

matthieudumont commented Jan 31, 2014

Allright, I will probably do my pull request on monday.

Thx for feedback!

@skoudoro

This comment has been minimized.

Member

skoudoro commented Jun 12, 2017

Should we close this one @Garyfallidis ?

@Garyfallidis

This comment has been minimized.

Member

Garyfallidis commented Jun 13, 2017

Sure.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment