The Huygens software by default includes a tool to correct Hot Pixels and Cold Pixels. This tool uses a special algorithm to detect hot or cold pixels and replaces them with the median of their nearest neighbors.
Removal of hot and cold pixels is useful:
- For deconvolution; deconvolving images with hot or cold pixels influences the deconvolution, especially when these pixels are not due to strong noise, but due to deficiencies of the camera.
- To avoid false colocalization; when hot and cold pixels occur at the same position in each channel, this will affect the colocalization analysis.
- For visualization; hot and cold pixels disturb the contrast range, and could force the contrast of the image to be very low. This may cause low intensity objects of interest to become almost invisible.
We have come across artifacts due to the presence of hot pixels a few times. When the image seems black after deconvolution, try to remove hot pixels first and check, after repeating the deconvolution process, if the original contrast of the image is restored.
The hot & cold pixel remover interface consists of several parts. In the top left, the original image can be seen, whereas the top right shows the corrected image. By default a MIP projection will be shown in the XZ direction as this makes it easier to recognize hot pixels by eye. For cold pixel removal, a sum projection is used. Furthermore only a single channel is shown, as the correction must be checked per channel. In the bottom left a help box explains the general features of the hot & cold pixel remover. In the bottom center is the detection settings box, which shows the detection parameters and different correction and mask options. The bottom right shows a standard visualization settings box. At the very bottom a statusbar is shown with a progressbar.
|Tweaking the detection settings (Optional)
TThe most important setting is the top one: “Find & correct”. Here you can specify whether to look for and remove hot pixels, or cold pixels. Only one type of pixel aberration can be corrected at a time. Most of the time an image will only have hot pixels, so this is the standard mode. In case of an image with both hot and cold pixels (rare), it is recommended to remove the hot pixels first, and then remove the cold pixels in the result image. The locations that are flagged as hot / cold pixels will depend on the pixel detection settings. Whenever these settings are changed the shown correction changes accordingly. Also next to “Result” you will see the total number of unique hot / cold pixels that have been corrected. The settings are split up into two groups: The correlation settings and the sensitivity. The correlation settings allow you to specify whether there is some additional correlation between the hot / cold pixel locations, which the detection algorithm can use. If you are not sure if any correlations exist, you can simply disable all repeating settings. Next we will explain why and when these correlations can occur.
- Correlation along Z: When a camera defect is the cause of hot / cold pixels, this means that certain pixels in the 2D sensor grid will often produce a hot / cold pixel. When the image is recorded slice-by-slice (in the Z direction), this means that each Z slice has a high likelihood of having a hot / cold pixel at this exact location. This leads to “hot columns” and “cold columns” respectively, which show up as vertical lines in an XZ projection (like a barcode). Since this is almost always the case, the “along Z” flag is true by default.
- Correlation along T: For the same reason as above, when a time series is recorded with a single camera (as is usually the case in a time series), the hot / cold pixels will show up at the same location in each time frame. In this case the correlation along T option should be checked. Note that if the hot / cold pixels in a sensor change rapidly over time, this will no longer be true.
- Correlation along Channels: For the same reason as above, when the channels of a multi-channel image are recorded one by one, with the same camera, the hot / cold pixels will show up in the same locations in each channel. In this case the correlation along Channels option should be checked. Note that if the hot / cold pixels in a sensor change rapidly over time, this will no longer be true. When along channels is checked, the image windows will show all channels at once.
- Sensitivity: After properly specifying the repeating parameters the correction will generally already be good. However, the sensitivity of the tool can be fine-tuned for each channel using the sensitivity slider. When a channel is selected next to the sensitivity slider it is automatically shown in the image windows. When the sensitivity is lowered, fewer hot / cold pixels will be found, and vice versa. A higher sensitivity increases the odds of flagging regular pixels as hot / cold pixels, while a lower sensitivity increases the chance of flagging hot / cold pixels as regular pixels. Ideally you want to lower the sensitivity as much as possible before hot / cold pixels show up in the correction.
|Saving the result
When the corrections have been verified and (optional) the sensitivities have been fine-tuned for each channel, the correction can be saved by clicking the “Save correction” button. This will load the correction directly into the main Huygens window, from which it can be saved or edited further.
- Exporting a mask: When the corrections have been verified and (optional) the sensitivities have been fine-tuned for each channel, the locations of the hot / cold pixels can be saved by clicking the “Export mask” button. This will save the pixel mask as a .h5 file. This mask can then be imported back into the hot & cold pixel remover for correcting another image, or can be added to a deconvolution template to correct a whole series of images with identical hot or cold pixels.
- Importing a mask: A previously determined mask can be imported from file by clicking the “Import mask” button. This will override the currently used correction with this old mask. The effect of the correction with this mask can be checked, and if it is good the correction can be saved as normal. Note that when the detection settings are changed, these will take priority again and the old mask will no longer be used.
|Removing hot or cold pixels in the batch processor
A hot or cold pixel removal step can be added to a deconvolution template using either the “Edit deconvolution template” tool or the internal deconvolution template editor in the batch processor. In either case you must navigate to the pre-processing options (“Pre” tab). Adding a hot or cold pixel removal step simply means importing a previously created mask from file. When the specified mask is valid, it will light up green, and a pixel correction step is added to the deconvolution template. If the path is invalid, it will light up red and the pixel correction step is skipped. Whenever this occurs you will be notified of the reason for this invalidity.
Images were used with permission of Dr. Rebecca Lee and Genevieve Phillips, Fluorescence Microscopy Shared Resource, University of New Mexico School of Medicine, Albuquerque, USA