Your image is now 8 bits, so the pixels can have now 2^8 values (range from 0-255) but….If over this image you do Threshold (Huang) and press apply two things happen: Since this image is uncalibrated Raw ID = ID (basically the sum of pixels contained in this selection) The values of this 4 pixels are the ones displayed. Lets consider the following image (10px x 10px) attached in tiff Your original image is 16bits, so each pixel can have up to 2^16 different values (from 0 to 65535) Please go to the ImageJ manual, read carefully and try to understand here the difference between measurements. These are really basic questions about what a pixel actually is, and the values it can take in grayscale images. Sorry for the length – I feel like many threads talk about thresholding and regular IntDen, but not my specific issue of whether an IntDen/RawIntDen from a binary image is worth using, or if RawIntDen values can be utilized and normalized to areas that I want to separately define. Would it be appropriate to take the RawIntDen and divide by the cell area, to get a normalized sum of pixels relative to that individual cell’s area? Issue 2: Is it ever appropriate to use/report RawIntDen for intensity? I want to compare the number of pixels (to reflect intensity) of the total amount of puncta among my different cells and treatments. (To clarify, I’m doing this because thresholding/Analyzing Particles alone can capture puncta I don’t want to measure) I want the IntDen of the particle ROI(s), but want to account for the total cell area. Thus, if I treat all of the particles as single ROI (seen in the righthand cell image in the screenshot, where puncta ROIs are drawn within each cell), then the measured IntDen value is reflective of the total area of those puncta outlines, as opposed to thresholding the puncta and analyzing particles within the ROI of the entire cell, making the IntDen relative to the cell’s area. My understanding is that the IntDen takes the area of the ROI into account (mean gray value * # of pixels). My workflow involves (using scaled images) thresholding for the larger puncta, setting minimum to exclude the smaller stuff I want to ignore, then creating particle ROIs that I then combine to reflect a single ROI that includes all of the particles for that cell – because I want to evaluate their measurement as a whole (total area, total intensity) rather than individual. I haven’t found any novice-friendly resources as to the underlying differences in intensity in regular images vs those made into binaries. Issue 1: When is it appropriate to Analyze Particles – Thresholding without clicking apply (with particles highlighted red)? Or Thresholding and clicking Apply to create a binary? Is any sort of intensity measurement from a binary image correct/accurate to use? The particle areas remain the same in either case, but not the IntDen (screenshot attached). Now, I’m interested in just measuring a subset of particles/puncta that appear to be inherently brighter than the rest. Thus, I turned to thresholding for all of the puncta (large and small, dim and bright) in the cell to ascertain these differences. When I measure the mean gray value or IntDen for the entire cell, I don’t feel like the differences are effectively captured because (I believe) the darkness of the cytoplasm skews the average gray value to the lower side, which should affect the mean gray value and the IntDen because both take ROI area into account. New ImagePlus("Overlay Mask",mask).I’m attempting to quantify the “brightness” of each of these larger and brighter puncta (spots/blobs), to see how different they are between drug vs no drug treatment, and how different the response to the drug is between different cell lines. Roi = new PolygonRoi(xpoints,ypoints,Roi.POLYGON) IJ.setAutoThreshold(img, "Huang dark no-reset") New ImagePlus("Threshold Mask_"+img.getTitle(),mask).show() Here is a JavaScript example that demonstrates the use of both the createThresholdMask() and createRoiMask() methods: // Threshold mask 1 You will need to be running ImageJ 1.52f19 or later for it to work with line selections and overlays. You can use the createRoiMask() method to create masks from ROIs or overlays. It would help if you could provide a MWE (Minimal Working Example), as a macro or script ( ). It is not clear what you are trying to do.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |