Segmentation of objects when the signal intensity varies among them
Locked
 
|
<http://imagej.1557.x6.nabble.com/file/t382821/Example_bacteria_FM4-64.jpg>
<http://imagej.1557.x6.nabble.com/file/t382821/Example_bacteria_FM4-64_BINARY.jpg> <http://imagej.1557.x6.nabble.com/file/t382821/Example_bacteria_FM4-64_CLEAR.jpg> Hi, I am quite new to Fiji and have encountered a following challenge. I have a set of images of bacteria taken using 2 channels, DAPI (for DNA) and FM-64 (for cellular membrane). My goal is to do the segmentation so that I get a set of individual ROIs (one for each bacterium) based on DNA and another set of individual ROIs based on the membrane. And then I want to measure the features of the objects (bacteria) in the original image surrounded by ROIs. I constructed a macro that worked OKish for the DAPI channel but not for the FM4-64 channel. My problem is that I don’t manage to threshold all objects of interest (all bacteria) in one operation. When I try to apply a thresholding algorithm from the available list, some bacteria still remain below the threshold. I assume this is because the FM 4-64 signal is much stronger in some cells then in the others (see picture Example bacteria FM4-64). I tied to apply Filters (I liked Median the most) and Background Subtraction (radius 10, comparable to the size of my objects) before the thresholding and it helped a bit, but still the pale cells were left as a background. So what I did was that I ran a 2-step segmantation, the first step for the “bright” cells and the second for the “pale” cells. I am not sure if this an acceptable way to do it. Because I had to apply one thresholding algorithm at the first step, then after the measurements I just cleared the selected ROIs in the original image (which created some “black holes” in it, see picture Example bacteria FM4-64 CLEAR) . And ran another thresholding algorythm (a different one!) on the remaining objects… Another thing I tried was to use a Filter called Unsharp Mask or a CLAHE (Enhance Local Contrast) function, and most cells were identified as objects and thresholded. But since some of the cells form chains, those chains appeared as single particles, so I had to apply Watershed which created some artificial “septa” where they shouldn’t be (see picture Example bacteria FM4-64 BINARY). Maybe you could suggest any ideas for a better thresholding? Or pre-processing before the thresholding? Or other options then Watershed to “divide” the cells? -- Sent from: http://imagej.1557.x6.nabble.com/ -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
Re: Segmentation of objects when the signal intensity varies among them
|
|
Hi,
Given that you have identified individual bacteria from the dapi image and assuming that all bacteria are identified; take each individual dapi ROI, expand it a little (enlarge) and use the enlarged ROI on the FM-64 image. Use one of the automatic thresholding methods. You are no longer trying to segment the FM-64 image in one go, but piece by piece - knowing that each small area probably contains a single bacteria. When bacteria form chains - you know their size, so for each chain you can estimate how many bacteria are involved - starting at one end of the chain the look for division at steps matching the size of the bacteria - possibly a slight narrowing of the width. Jeremy Adler -----Original Message----- From: ImageJ Interest Group <[hidden email]> On Behalf Of ekamis Sent: Thursday, March 4, 2021 8:41 AM To: [hidden email] Subject: Segmentation of objects when the signal intensity varies among them <http://imagej.1557.x6.nabble.com/file/t382821/Example_bacteria_FM4-64.jpg> <http://imagej.1557.x6.nabble.com/file/t382821/Example_bacteria_FM4-64_BINARY.jpg> <http://imagej.1557.x6.nabble.com/file/t382821/Example_bacteria_FM4-64_CLEAR.jpg> Hi, I am quite new to Fiji and have encountered a following challenge. I have a set of images of bacteria taken using 2 channels, DAPI (for DNA) and FM-64 (for cellular membrane). My goal is to do the segmentation so that I get a set of individual ROIs (one for each bacterium) based on DNA and another set of individual ROIs based on the membrane. And then I want to measure the features of the objects (bacteria) in the original image surrounded by ROIs. I constructed a macro that worked OKish for the DAPI channel but not for the FM4-64 channel. My problem is that I don’t manage to threshold all objects of interest (all bacteria) in one operation. When I try to apply a thresholding algorithm from the available list, some bacteria still remain below the threshold. I assume this is because the FM 4-64 signal is much stronger in some cells then in the others (see picture Example bacteria FM4-64). I tied to apply Filters (I liked Median the most) and Background Subtraction (radius 10, comparable to the size of my objects) before the thresholding and it helped a bit, but still the pale cells were left as a background. So what I did was that I ran a 2-step segmantation, the first step for the “bright” cells and the second for the “pale” cells. I am not sure if this an acceptable way to do it. Because I had to apply one thresholding algorithm at the first step, then after the measurements I just cleared the selected ROIs in the original image (which created some “black holes” in it, see picture Example bacteria FM4-64 CLEAR) . And ran another thresholding algorythm (a different one!) on the remaining objects… Another thing I tried was to use a Filter called Unsharp Mask or a CLAHE (Enhance Local Contrast) function, and most cells were identified as objects and thresholded. But since some of the cells form chains, those chains appeared as single particles, so I had to apply Watershed which created some artificial “septa” where they shouldn’t be (see picture Example bacteria FM4-64 BINARY). Maybe you could suggest any ideas for a better thresholding? Or pre-processing before the thresholding? Or other options then Watershed to “divide” the cells? -- Sent from: http://imagej.1557.x6.nabble.com/ -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html När du har kontakt med oss på Uppsala universitet med e-post så innebär det att vi behandlar dina personuppgifter. För att läsa mer om hur vi gör det kan du läsa här: http://www.uu.se/om-uu/dataskydd-personuppgifter/ E-mailing Uppsala University means that we will process your personal data. For more information on how this is performed, please read here: http://www.uu.se/en/about-uu/data-protection-policy -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
|
|
Hi,
I don't think I see a problem with your 2-step segmentation, as long as it identifies your bacteria correctly and you document how you have done the segmentation. Your goal is to identify the bacteria and create ROIs for each and I understand that works. If you do some intensity measurements further down the line you would go back to the original image and measure all identified ROIs. It does not matter so much how you have identified these ROIs if they are correct. Best wishes Kees Dr Ir K.R. Straatman FRMS Advanced Imaging Facility University of Leicester www.le.ac.uk/advanced-imaging-facility<https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.le.ac.uk%2Fadvanced-imaging-facility&data=04%7C01%7Ckrs5%40leicester.ac.uk%7C485530d84c0e42d1de5608d88a42a407%7Caebecd6a31d44b0195ce8274afe853d9%7C0%7C0%7C637411366106508827%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=I6oSweJYMVqx%2B%2B3VtQ8cXgHDUy4%2F6zDmx%2FSd2lGC8bc%3D&reserved=0> ________________________________ From: Jeremy Adler Sent: 04 March 2021 08:17 Subject: Re: Segmentation of objects when the signal intensity varies among them Hi, Given that you have identified individual bacteria from the dapi image and assuming that all bacteria are identified; take each individual dapi ROI, expand it a little (enlarge) and use the enlarged ROI on the FM-64 image. Use one of the automatic thresholding methods. You are no longer trying to segment the FM-64 image in one go, but piece by piece - knowing that each small area probably contains a single bacteria. When bacteria form chains - you know their size, so for each chain you can estimate how many bacteria are involved - starting at one end of the chain the look for division at steps matching the size of the bacteria - possibly a slight narrowing of the width. Jeremy Adler -----Original Message----- From: ImageJ Interest Group On Behalf Of ekamis Sent: Thursday, March 4, 2021 8:41 AM Subject: Segmentation of objects when the signal intensity varies among them <https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2Ffile%2Ft382821%2FExample_bacteria_FM4-64.jpg&data=04%7C01%7Ckrs5%40leicester.ac.uk%7Cbf951a13591f4a7b42b908d8dee6eff0%7Caebecd6a31d44b0195ce8274afe853d9%7C0%7C0%7C637504430747511726%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=p9sSUfH4DliffDkb9ySpQJn5hq82Uf1mCLoRmgpRg7U%3D&reserved=0> <https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2Ffile%2Ft382821%2FExample_bacteria_FM4-64_BINARY.jpg&data=04%7C01%7Ckrs5%40leicester.ac.uk%7Cbf951a13591f4a7b42b908d8dee6eff0%7Caebecd6a31d44b0195ce8274afe853d9%7C0%7C0%7C637504430747521683%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=O53g8K17OVUmmMQx%2BIlW3BjOiX4OE%2FCLk1QxoXMsEBg%3D&reserved=0> <https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2Ffile%2Ft382821%2FExample_bacteria_FM4-64_CLEAR.jpg&data=04%7C01%7Ckrs5%40leicester.ac.uk%7Cbf951a13591f4a7b42b908d8dee6eff0%7Caebecd6a31d44b0195ce8274afe853d9%7C0%7C0%7C637504430747521683%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=VhD3Z%2B1tcRu7es9gMMAuSqiuALVbLw5E9isRMe3lKRE%3D&reserved=0> Hi, I am quite new to Fiji and have encountered a following challenge. I have a set of images of bacteria taken using 2 channels, DAPI (for DNA) and FM-64 (for cellular membrane). My goal is to do the segmentation so that I get a set of individual ROIs (one for each bacterium) based on DNA and another set of individual ROIs based on the membrane. And then I want to measure the features of the objects (bacteria) in the original image surrounded by ROIs. I constructed a macro that worked OKish for the DAPI channel but not for the FM4-64 channel. My problem is that I don’t manage to threshold all objects of interest (all bacteria) in one operation. When I try to apply a thresholding algorithm from the available list, some bacteria still remain below the threshold. I assume this is because the FM 4-64 signal is much stronger in some cells then in the others (see picture Example bacteria FM4-64). I tied to apply Filters (I liked Median the most) and Background Subtraction (radius 10, comparable to the size of my objects) before the thresholding and it helped a bit, but still the pale cells were left as a background. So what I did was that I ran a 2-step segmantation, the first step for the “bright” cells and the second for the “pale” cells. I am not sure if this an acceptable way to do it. Because I had to apply one thresholding algorithm at the first step, then after the measurements I just cleared the selected ROIs in the original image (which created some “black holes” in it, see picture Example bacteria FM4-64 CLEAR) . And ran another thresholding algorythm (a different one!) on the remaining objects… Another thing I tried was to use a Filter called Unsharp Mask or a CLAHE (Enhance Local Contrast) function, and most cells were identified as objects and thresholded. But since some of the cells form chains, those chains appeared as single particles, so I had to apply Watershed which created some artificial “septa” where they shouldn’t be (see picture Example bacteria FM4-64 BINARY). Maybe you could suggest any ideas for a better thresholding? Or pre-processing before the thresholding? Or other options then Watershed to “divide” the cells? -- Sent from: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2F&data=04%7C01%7Ckrs5%40leicester.ac.uk%7Cbf951a13591f4a7b42b908d8dee6eff0%7Caebecd6a31d44b0195ce8274afe853d9%7C0%7C0%7C637504430747521683%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=fFCXzbvSdA3CbgVwzLT9JQf2eZbYbLK6%2BJgCej4vghE%3D&reserved=0 -- ImageJ mailing list: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.nih.gov%2Fij%2Flist.html&data=04%7C01%7Ckrs5%40leicester.ac.uk%7Cbf951a13591f4a7b42b908d8dee6eff0%7Caebecd6a31d44b0195ce8274afe853d9%7C0%7C0%7C637504430747521683%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=c7FWOsMs3FOxv0mxu%2FGf5fWmb5CTHwW3byUwj5VTFSo%3D&reserved=0 När du har kontakt med oss på Uppsala universitet med e-post så innebär det att vi behandlar dina personuppgifter. För att läsa mer om hur vi gör det kan du läsa här: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.uu.se%2Fom-uu%2Fdataskydd-personuppgifter%2F&data=04%7C01%7Ckrs5%40leicester.ac.uk%7Cbf951a13591f4a7b42b908d8dee6eff0%7Caebecd6a31d44b0195ce8274afe853d9%7C0%7C0%7C637504430747521683%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=P8X5mVivq4nkIcy339qh4F2fkKUVlTYti9ymeRS02%2FI%3D&reserved=0 E-mailing Uppsala University means that we will process your personal data. For more information on how this is performed, please read here: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.uu.se%2Fen%2Fabout-uu%2Fdata-protection-policy&data=04%7C01%7Ckrs5%40leicester.ac.uk%7Cbf951a13591f4a7b42b908d8dee6eff0%7Caebecd6a31d44b0195ce8274afe853d9%7C0%7C0%7C637504430747521683%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=h4Bw64GRUgegGpF3X%2Bymx9HvkurchqrAwAPqZsps2c0%3D&reserved=0 -- ImageJ mailing list: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.nih.gov%2Fij%2Flist.html&data=04%7C01%7Ckrs5%40leicester.ac.uk%7Cbf951a13591f4a7b42b908d8dee6eff0%7Caebecd6a31d44b0195ce8274afe853d9%7C0%7C0%7C637504430747521683%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=c7FWOsMs3FOxv0mxu%2FGf5fWmb5CTHwW3byUwj5VTFSo%3D&reserved=0 -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
«
Return to ImageJ
|
1 view|%1 views
| Free forum by Nabble | Edit this page |