5 - Spot Detection

5 - Spot Detection#

5.1 - Manual spot detection with the Multi-point Tool#

  1. read all the next steps before beginning

  2. open the beads_001.tif image in folder spot_detection (drag & drop, or File > Open)

  3. right-click (ctrl-click) on the Point Tool in the tool bar and select MultiTool (IJ:19.5)

  4. double-click on the Point Tool in the toolbar and customize to your liking (test on image, clear points by shift + a)

  5. count all beads by clicking on them one by one

  6. how long did it take you to click on all the beads (measure with stopwatch the time it took to complete the previous step)?

  7. add the points to the ROI Manager (e.g. press t on your keyboard)

    • NOTE: only one ROI will be added, it contains all the points

  8. save the ROI to file (in the ROI Manager window, More > Save)

  9. click measure in the ROI Manager to get number of spots

5.2 - Algorithmic spot detection with Find Maxima#

useful Find Maxima info on the image.sc forum.

  1. open the beads_001.tif image in folder spot_detection (drag & drop, or File > Open)

  2. Process > Find Maxima

  3. check Preview point selection

  4. try different values for Prominence and the three checkboxes

  5. try each of the possibilities in the pull-down (remember to click OK to apply your selection)

    • what is the difference and what could the different outputs be used for?

    • tip: shift + u allows you to pin sub-menus of the control panel for quick selection (e.g. the Process menu)

  6. add you detected maxima to the ROI Manager and save to file

    • using your preferred settings, set the Output type to Point Selection and press OK

    • add the points to the ROI Manager (e.g. press t on your keyboard)

      • NOTE: only one ROI will be added, it contains all the points.

    • save the ROI to file (in the ROI Manager window, More > Save).

  7. load saved ROIs from the manual detection exercise (5.1) (e.g. drag & drop on Fiji the ROIs .zip file)

  8. compare results: how many beads do you get and how does the number compare to your manual count?

5.3 - Automatic spot segmentation with thresholding#

  1. open the beads_001.tif image in folder spot_detection (drag & drop, or File > Open)

  2. Image > Adjust > Threshold (IJ:28.2.4)

    • select Otsu, then Apply

  3. Analyze > Set Measurements (IJ:30.7)

    • select Area, Mean gray value, Display label

    • click on OK

  4. Analyze > Analyze Particles (IJ:30.2)

    • select the Overlay Masks option in the Show dropdown menu

    • select Display results, Clear Results, Summarize, Add to Manager

  5. save the ROIs to file (in the ROI Manager window, More > Save)

  6. Compare results to the previous two approaches (5.1 and 5.2)

    • Do they differ significantly from each other?

    • Which performed better?

5.4 - Spot detection with noise#

  1. Repeat 5.2 with the with_noise_8000.tif image in folder spot_detection

    • consider smoothing the image first. You can try any of the filter in the Process > Filters menu:

      • e.g. Process > Filters > Gaussian Blur, Process > Filters > Mean, …

      • NOTE: the Smooth operation under the Process > Smooth is a mean filter with radius 1 (3x3 kernel)

  2. Repeat 5.3 with the with_noise_8000.tif image in folder spot_detection

    • consider smoothing the image first. You can try any of the filter in the Process > Filters menu:

      • e.g. Process > Filters > Gaussian Blur, Process > Filters > Mean, …

      • NOTE: the Smooth operation under the Process > Smooth is a mean filter with radius 1 (3x3 kernel)

  3. Compare and comment on the performance of the two methods

    • Which one is better?

    • Why/how is it better?

5.5 - Spot detection with variable background#

  1. Repeat 5.2 with the beads_001_ramp.tif image in folder spot_detection

  2. Repeat 5.3 with the beads_001_ramp.tif image in folder spot_detection

  3. Compare and comment on the performance of the two methods

    • Which one is better?

    • Why/how is it better?

  4. What could be done to the image to make thresholding work better?