Skip to first unread message
X. Qin
Dec 21, 2017, 10:41:37 PM12/21/17
to Icy imaging
Hi all,
I have some questions about the spot detection of Icy. I would like to know the information of the "small purple circles", the ROI of spot detection.
1. What is the size of ROI for each spot detection in pixel of Icy? Does the ROI size for each spot detection is fixed in spot detection (I will detect a lot of spots in one movie, will the ROI size be fixed?)?
2. How to Icy generate the intensity data for detections? How does it subtract the background? (after I do spot detection and analyze the trajectories, I export the intensity information for all the trajectories, I can see the intensity after background subtraction option. I would like to know how the software work to realize it. If anyone can provide some technical paper about it, it will be great!)
Best, and happy holidays!
X.Qin
X. Qin
Dec 21, 2017, 10:51:06 PM12/21/17
to Icy imaging
Hi.. By saying ROI, I was talking about the circles generated by the spots detection.
Stephane
Dec 27, 2017, 5:48:53 AM12/27/17
to Icy imaging
Hi X.Qin,
In fact the red circle marker are just barycenter representation of the detection and not the complete / real detection itself.
You can see real detection using the binary mask or ROI export (in the 'output' section). Intensity values are obtain from the detection mask, using real intensities value (no background subtraction). Background subtraction should be used with attention, i think it's better to compare results without background subtraction but it also depends from your images.
Hope this give you some hints !
I wish you nice Christmas holidays as well :)
Best,
- Stephane
In fact the red circle marker are just barycenter representation of the detection and not the complete / real detection itself.
You can see real detection using the binary mask or ROI export (in the 'output' section). Intensity values are obtain from the detection mask, using real intensities value (no background subtraction). Background subtraction should be used with attention, i think it's better to compare results without background subtraction but it also depends from your images.
Hope this give you some hints !
I wish you nice Christmas holidays as well :)
Best,
- Stephane
X. Qin
Dec 27, 2017, 6:10:08 AM12/27/17
to Icy imaging
Dear Stephane,
Thank you so much for your reply. I would like to make sure a few thingsIntensity values are obtain from the detection mask, using real intensities value (no background subtraction)
Does the "detection mask" mean the small irregular patterns in the following image?
If so, does it mean Icy counts the intensity just within these irregular masks?
Stephane
Dec 27, 2017, 6:41:45 AM12/27/17
to Icy imaging
Dear X.Qin,
I can't see your image but if this is the binary mask, then yeah, icy quantify intensities from the irregular detected masks.
About background subtraction, there is no any "background subtraction" directly inside Icy nor inside the Spot Detector, Spot Detector does background estimation to find the best spot detection but background is not removed...
There is several method to remove background, and there are not trivial generally speaking. In Icy we often recommend to not necessary try to remove it, but just be aware you have some background information and so take it in consideration.
ImageJ provide a default background removal tool : https://imagej.net/Rolling_Ball_Background_Subtraction
You can try to use it, but again we don't recommend to use it except if it works really well in your case and still allow to have comparable results.
Best,
- Stephane
I can't see your image but if this is the binary mask, then yeah, icy quantify intensities from the irregular detected masks.
About background subtraction, there is no any "background subtraction" directly inside Icy nor inside the Spot Detector, Spot Detector does background estimation to find the best spot detection but background is not removed...
There is several method to remove background, and there are not trivial generally speaking. In Icy we often recommend to not necessary try to remove it, but just be aware you have some background information and so take it in consideration.
ImageJ provide a default background removal tool : https://imagej.net/Rolling_Ball_Background_Subtraction
You can try to use it, but again we don't recommend to use it except if it works really well in your case and still allow to have comparable results.
Best,
- Stephane
X. Qin
Dec 27, 2017, 7:09:49 AM12/27/17
to Icy imaging
I am sorry, by saying "background subtraction" actually I meant the "DISK_BACKGROUND_CORRECTED" in the TrackManager:
I
If I export the data from this item, I think the background must have been "removed" or "corrected" in some sense. I would like to know how Icy does it. (whether the background used is a fixed value for the whole image, or it is different for all the detections. If the latter is the case, then how does Icy do it).
Thanks again for your nice reply!
Best,
Qin
X. Qin
Dec 27, 2017, 8:18:17 PM12/27/17
to Icy imaging
BTW, if here I choose the intensity type (in TrackManager, attached the screen capture above) to be "DISK" or "DISK_BACKGROUND_CORRECTED", does it mean the intensity is not counted from the "mask"? Because there is another choice "SPOT_MASK" there. .
Anyone knows what is the "DISK" in there?
Thanks!
Qin
On Friday, 22 December 2017 11:41:37 UTC+8, X. Qin wrote:
X. Qin
Dec 29, 2017, 6:50:57 AM12/29/17
to Icy imaging
Dear Stephane,
I am sending a cartoon explaining my understanding of "DISK" and "DISK_BACKGROUND_CORRECTED" for intensity export.
The blue irregular one is the mask that Icy generated. Icy of course can export the intensity just within this mask. In fact, there is indeed a item named "SPOT_MASK" in intensity profile export.
But I just wonder, how Icy works if we use "DISK" and "DISK_BACKGROUND_CORRECTED" to export intensity data.
My understanding is:
1. Icy can itself have a regular shaped "DISK" surrounding the mask. (as I draw in the cartoon, the red circle). And it can export the intensity data within this DISK if we choose it.
2. The background may be defined as those areas within the DISK but outside the mask, as shown in my drawing, the yellow area. I guess the "DISK_BACKGROUND_CORRECTED" may be the intensity within the DISK minus the intensity of the yellow area.
Let me know if you have any other idea about the definition of these two items. Looking forward to any reply on this issue.
My best regards,
Qin
On Wednesday, 27 December 2017 19:41:45 UTC+8, Stephane wrote:
Stephane
Jan 2, 2018, 7:46:51 AM1/2/18
to Icy imaging
Dear Xianan Xin,
Sorry for the very ate reply, in fact i'm currently in vacation (and a lot of people here during this period ^^).
I'm not the author of this plugin and to be honest i'm not sure about the "background corrected" intensity value... I hope the author of the plugin can bring some lights here.
If i'm correct, DISK is just referring the classic red spot disk (size of disk is fixed, not related to the mask i think) so it will compute the intensity from the spot mark only (not the full mask).
I will bring you more details when i will be back from vacation (in 2 days).
Best,
- Stephane
Sorry for the very ate reply, in fact i'm currently in vacation (and a lot of people here during this period ^^).
I'm not the author of this plugin and to be honest i'm not sure about the "background corrected" intensity value... I hope the author of the plugin can bring some lights here.
If i'm correct, DISK is just referring the classic red spot disk (size of disk is fixed, not related to the mask i think) so it will compute the intensity from the spot mark only (not the full mask).
I will bring you more details when i will be back from vacation (in 2 days).
Best,
- Stephane
nicolas
Jan 23, 2018, 5:04:13 PM1/23/18
to Icy imaging
Dear X. Qin,
I have just checked the source code of the plugin to be sure, but the options are the following:
* DISK
it computes pixel intensities over a disk area centered on the spot. The diameter used is specified in the field 'Diameter of the disk area in pixels'
* DISK_BACKGROUND_CORRECTED'
same thing as DISK, but this time a background value is subtracted to pixel intensities. The background value is computed over a doughnut shape area which inner and outer diameters are specified by the two fields 'Diameter of the disk area in pixels' and 'Diameter of the background disk area in pixels', respectively.
* 'GAUSSIAN_FIT2D'
A Gaussian (bell) shape is fitted to the spot and the amplitude of this bell shape is considered to be the spot intensity. You can specify the standard deviation of the Gaussian shape, but its center is fixed to the center of the original spot. The amplitude value is background corrected as there is a constant included in the equation of pixel intensity that is fitted to the data.
* 'SPOT_MASK'
in case you used the 'Spot detector' plugin to indentify your spots, a binary mask of the spot location is available (not just the locatio of its center of mass). Using this option you are computing pixel intensities for these identify pixels. I believe this would be pixels inside the blue shape in your example schematics.
Note that with the latter option there is not background correction applied. So, your example with a blue shape for foreground intensity, surrounded by a yellow circular area for background intensity is not currently implemented.
I hope this helps. Anyway, I am very happy you find this pluging useful for your work.
Best
Nicolas
I have just checked the source code of the plugin to be sure, but the options are the following:
* DISK
it computes pixel intensities over a disk area centered on the spot. The diameter used is specified in the field 'Diameter of the disk area in pixels'
* DISK_BACKGROUND_CORRECTED'
same thing as DISK, but this time a background value is subtracted to pixel intensities. The background value is computed over a doughnut shape area which inner and outer diameters are specified by the two fields 'Diameter of the disk area in pixels' and 'Diameter of the background disk area in pixels', respectively.
* 'GAUSSIAN_FIT2D'
A Gaussian (bell) shape is fitted to the spot and the amplitude of this bell shape is considered to be the spot intensity. You can specify the standard deviation of the Gaussian shape, but its center is fixed to the center of the original spot. The amplitude value is background corrected as there is a constant included in the equation of pixel intensity that is fitted to the data.
* 'SPOT_MASK'
in case you used the 'Spot detector' plugin to indentify your spots, a binary mask of the spot location is available (not just the locatio of its center of mass). Using this option you are computing pixel intensities for these identify pixels. I believe this would be pixels inside the blue shape in your example schematics.
Note that with the latter option there is not background correction applied. So, your example with a blue shape for foreground intensity, surrounded by a yellow circular area for background intensity is not currently implemented.
I hope this helps. Anyway, I am very happy you find this pluging useful for your work.
Best
Nicolas
X. Qin
Jan 23, 2018, 8:05:08 PM1/23/18
to Icy imaging
Hi Nicolas,
Thank you sooo much for your reply.
I have some further questions regarding your explanations:
1.
* DISK
it computes pixel intensities over a disk area centered on the spot. The diameter used is specified in the field 'Diameter of the disk area in pixels'
it computes pixel intensities over a disk area centered on the spot. The diameter used is specified in the field 'Diameter of the disk area in pixels'
What is the "center"(s) used here? (how does TrackManager define/find the "center"?)
2. What are default size of "DISK" and size of "doughnut" for background correction? (If I don't fill these two values, the TrackManager can still calculate these intensity data. The what is the "default" sizes?)
Best,
Qin
Reply all
Reply to author
Forward
0 new messages