ROI review paper?
Locked
 
|
Dear List,
does anyone know of a review paper treating the various ROI algorithms and their various strengths and weaknesses? JPK -- ******************************************* Jacob Pearson Keller Northwestern University Medical Scientist Training Program email: [hidden email] ******************************************* -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
|
|
Jacob,
I think your question may be too broad to answer. Can you be more specific? By "ROI algorithms" are you talking about various algorithms to identify some particular type of feature, or texture, or color? Are you asking specifically about medical applications? (versus satellite photo analysis?) About microscopy images? Red blood cells in a suspension? Stained nuclei in a tissue sample? Dual-stained nuclei with H&E? Histopathology images for cancer staging? 3-D feature detection in a stack of 2-D slices of CT-Scans or MRI-images? Counting? Area? Density? Shape detection? etc. Wade On 8/8/12 9:19 AM, Jacob Keller wrote: > Dear List, > > does anyone know of a review paper treating the various ROI algorithms and > their various strengths and weaknesses? > > JPK > -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
|
|
>
> Jacob, > > I think your question may be too broad to answer. > Can you be more specific? > Very good point--thanks for making it, and sorry for being unclear. What I have in mind is a general account of the various methods for the selection of ROIs, and not so much the processing thereafter. Examples are manual outlining, thresholding, using templates/masks from other images, etc. In other words, what methods can be used to define ROIs, either manually or computationally? Does that make more specific sense? Maybe there are so many ways that there is no way to review them all? Jacob -- ******************************************* Jacob Pearson Keller Northwestern University Medical Scientist Training Program email: [hidden email] ******************************************* -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
|
Hi Jacob,
In other words, what methods can be used to define ROIs, either manually > or computationally? This is still an extremely general problem, which is more technically referred to as segmentation. Searching Google Scholar for "segmentation methods review" yields over 637,000 results. I agree with Wade that it would be best to narrow your focus to a more specific case. What types of objects are you trying to segment? Are they generally shaped like tubes, blobs or clouds? A popular tube to segment is the neurite (e.g., http://fiji.sc/Simple_Neurite_Tracer). Popular blobs include many cells, nuclei, etc. (e.g., http://cellprofiler.org/). If you can be more specific, perhaps someone on this list is segmenting something similar to you, and can comment further on methods they have found effective. Regards, Curtis On Thu, Aug 9, 2012 at 9:01 AM, Jacob Keller <[hidden email] > wrote: > > > > Jacob, > > > > I think your question may be too broad to answer. > > Can you be more specific? > > > > Very good point--thanks for making it, and sorry for being unclear. What I > have in mind is a general account of the various methods for the selection > of ROIs, and not so much the processing thereafter. Examples are manual > outlining, thresholding, using templates/masks from other images, etc. In > other words, what methods can be used to define ROIs, either manually or > computationally? Does that make more specific sense? Maybe there are so > many ways that there is no way to review them all? > > Jacob > > > > -- > ******************************************* > Jacob Pearson Keller > Northwestern University > Medical Scientist Training Program > email: [hidden email] > ******************************************* > > -- > ImageJ mailing list: http://imagej.nih.gov/ij/list.html > -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
|
|
Jacob, since the code and logic behind a segmentation algorithm
can be arbitrarily complex, it's hard to know how to "count" or how to, well, segment or cluster the universe of segmentation techniques. It sounds more like you're looking for a taxonomy of techniques that are commonly used or already implemented in code and perhaps, for that matter, peer reviewed and therefore usable in research grants. (?) The ImageJ group is extremely helpful and full of good suggestions, but it would be easier if you could explain WHY you are interested in such a taxonomy. I have often found this group to respond to questions such as "How do I do X" with replies such as "OMG, don't do X, no one does X any more, do YZ it's much better ... and if it's a tissue sample, only do YZ if the pH is >=9... and here's a link to code that does that problem and works in Ames white mice." Or, even better, if you can attach an image, or better, a link to on-line image(s) of the type you are trying to segment, we can be much more specific about ways to tackle that and their pros and cons. Wade On 8/9/12 8:40 AM, Curtis Rueden wrote: > Hi Jacob, > > In other words, what methods can be used to define ROIs, either manually >> or computationally? > > This is still an extremely general problem, which is more technically > referred to as segmentation. Searching Google Scholar for "segmentation > methods review" yields over 637,000 results. > > I agree with Wade that it would be best to narrow your focus to a more > specific case. What types of objects are you trying to segment? Are they > generally shaped like tubes, blobs or clouds? A popular tube to segment is > the neurite (e.g., http://fiji.sc/Simple_Neurite_Tracer). Popular blobs > include many cells, nuclei, etc. (e.g., http://cellprofiler.org/). > > If you can be more specific, perhaps someone on this list is segmenting > something similar to you, and can comment further on methods they have > found effective. > > Regards, > Curtis > > > On Thu, Aug 9, 2012 at 9:01 AM, Jacob Keller <[hidden email] >> wrote: >>> Jacob, >>> >>> I think your question may be too broad to answer. >>> Can you be more specific? >>> >> Very good point--thanks for making it, and sorry for being unclear. What I >> have in mind is a general account of the various methods for the selection >> of ROIs, and not so much the processing thereafter. Examples are manual >> outlining, thresholding, using templates/masks from other images, etc. In >> other words, what methods can be used to define ROIs, either manually or >> computationally? Does that make more specific sense? Maybe there are so >> many ways that there is no way to review them all? >> >> Jacob >> >> >> >> -- >> ******************************************* >> Jacob Pearson Keller >> Northwestern University >> Medical Scientist Training Program >> email: [hidden email] >> ******************************************* >> >> -- >> ImageJ mailing list: http://imagej.nih.gov/ij/list.html >> > -- > ImageJ mailing list: http://imagej.nih.gov/ij/list.html -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
|
|
In reply to this post by Jacob Keller
Try Keith Price's bibliography.
http://iris.usc.edu/vision-notes/bibliography/twod277re1.html On Thu, Aug 9, 2012 at 7:01 AM, Jacob Keller <[hidden email] > wrote: > > > > Jacob, > > > > I think your question may be too broad to answer. > > Can you be more specific? > > > > Very good point--thanks for making it, and sorry for being unclear. What I > have in mind is a general account of the various methods for the selection > of ROIs, and not so much the processing thereafter. Examples are manual > outlining, thresholding, using templates/masks from other images, etc. In > other words, what methods can be used to define ROIs, either manually or > computationally? Does that make more specific sense? Maybe there are so > many ways that there is no way to review them all? > > Jacob > > > > -- > ******************************************* > Jacob Pearson Keller > Northwestern University > Medical Scientist Training Program > email: [hidden email] > ******************************************* > > -- > ImageJ mailing list: http://imagej.nih.gov/ij/list.html > -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
|
|
In reply to this post by Wade Schuette
Exactly right about the taxonomy concept. I am interested to know, let's
say for cells, how well the various segmentation techniques work, and what are their relative merits and demerits. The reason behind my question is that I am considering developing a new type of segmentation, and want to see how it would stack up against various others, make sure that there is a niche for it, make sure that no one else has already done it, and try to determine how generalizable it will be. JPK On Thu, Aug 9, 2012 at 11:32 AM, Wade Schuette <[hidden email]>wrote: > Jacob, since the code and logic behind a segmentation algorithm > can be arbitrarily complex, it's hard to know how to "count" > or how to, well, segment or cluster the universe of segmentation > techniques. > > It sounds more like you're looking for a taxonomy of techniques > that are commonly used or already implemented in code and > perhaps, for that matter, peer reviewed and therefore usable > in research grants. (?) > > The ImageJ group is extremely helpful and full of good suggestions, > but it would be easier if you could explain WHY you are interested > in such a taxonomy. > > I have often found this group to respond to questions such as > "How do I do X" with replies such as "OMG, don't do X, no > one does X any more, do YZ it's much better ... and if it's > a tissue sample, only do YZ if the pH is >=9... and here's > a link to code that does that problem and works in > Ames white mice." > > Or, even better, if you can attach an image, or better, > a link to on-line image(s) of the type you are trying to > segment, we can be much more specific about ways > to tackle that and their pros and cons. > > > Wade > > > > > > > > On 8/9/12 8:40 AM, Curtis Rueden wrote: > >> Hi Jacob, >> >> In other words, what methods can be used to define ROIs, either manually >> >>> or computationally? >>> >> >> This is still an extremely general problem, which is more technically >> referred to as segmentation. Searching Google Scholar for "segmentation >> methods review" yields over 637,000 results. >> >> I agree with Wade that it would be best to narrow your focus to a more >> specific case. What types of objects are you trying to segment? Are they >> generally shaped like tubes, blobs or clouds? A popular tube to segment is >> the neurite (e.g., http://fiji.sc/Simple_Neurite_**Tracer<http://fiji.sc/Simple_Neurite_Tracer>). >> Popular blobs >> include many cells, nuclei, etc. (e.g., http://cellprofiler.org/). >> >> If you can be more specific, perhaps someone on this list is segmenting >> something similar to you, and can comment further on methods they have >> found effective. >> >> Regards, >> Curtis >> >> >> On Thu, Aug 9, 2012 at 9:01 AM, Jacob Keller <[hidden email]. >> **edu <[hidden email]> >> >>> wrote: >>> >>>> Jacob, >>>> >>>> I think your question may be too broad to answer. >>>> Can you be more specific? >>>> >>>> Very good point--thanks for making it, and sorry for being unclear. >>> What I >>> have in mind is a general account of the various methods for the >>> selection >>> of ROIs, and not so much the processing thereafter. Examples are manual >>> outlining, thresholding, using templates/masks from other images, etc. In >>> other words, what methods can be used to define ROIs, either manually or >>> computationally? Does that make more specific sense? Maybe there are so >>> many ways that there is no way to review them all? >>> >>> Jacob >>> >>> >>> >>> -- >>> ********************************************* >>> Jacob Pearson Keller >>> Northwestern University >>> Medical Scientist Training Program >>> email: [hidden email] >>> ********************************************* >>> >>> -- >>> ImageJ mailing list: http://imagej.nih.gov/ij/list.**html<http://imagej.nih.gov/ij/list.html> >>> >>> -- >> ImageJ mailing list: http://imagej.nih.gov/ij/list.**html<http://imagej.nih.gov/ij/list.html> >> > > -- > ImageJ mailing list: http://imagej.nih.gov/ij/list.**html<http://imagej.nih.gov/ij/list.html> > -- ******************************************* Jacob Pearson Keller Northwestern University Medical Scientist Training Program email: [hidden email] ******************************************* -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
|
|
In reply to this post by David Webster
This is really helpful--thanks!
Jacob On Thu, Aug 9, 2012 at 11:46 AM, David Webster <[hidden email]>wrote: > Try Keith Price's bibliography. > > http://iris.usc.edu/vision-notes/bibliography/twod277re1.html > > On Thu, Aug 9, 2012 at 7:01 AM, Jacob Keller < > [hidden email] > > wrote: > > > > > > > Jacob, > > > > > > I think your question may be too broad to answer. > > > Can you be more specific? > > > > > > > Very good point--thanks for making it, and sorry for being unclear. What > I > > have in mind is a general account of the various methods for the > selection > > of ROIs, and not so much the processing thereafter. Examples are manual > > outlining, thresholding, using templates/masks from other images, etc. In > > other words, what methods can be used to define ROIs, either manually or > > computationally? Does that make more specific sense? Maybe there are so > > many ways that there is no way to review them all? > > > > Jacob > > > > > > > > -- > > ******************************************* > > Jacob Pearson Keller > > Northwestern University > > Medical Scientist Training Program > > email: [hidden email] > > ******************************************* > > > > -- > > ImageJ mailing list: http://imagej.nih.gov/ij/list.html > > > > -- > ImageJ mailing list: http://imagej.nih.gov/ij/list.html > -- ******************************************* Jacob Pearson Keller Northwestern University Medical Scientist Training Program email: [hidden email] ******************************************* -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
«
Return to ImageJ
|
1 view|%1 views
| Free forum by Nabble | Edit this page |