Autocrop
[[TOC]]
To extract large numbers of nuclei from a wide-field image, we have designed a process called autocrop here performed in 3D although this can also be achieved for 2D images. The basic principles of the autocrop process rely on a simple Otsu threshold method applied to the wide-field image. bounding box surrounding the 3D nuclei are defined to crop the nuclei from the wide-field image. To avoid to select to many objects which are not nuclei, a simple and scalable size filter was introduced. A second scalable filter allow to limit the number of multiple boxes for a given nucleus especially for large nuclei. Autocrop parameters are set by default but can can be modified prior to the analysis through a configuration file (config file).
First click on Autocrop NucleusJ 2 in NucleusJ2.0 plugin menu :
Then chose folder containing 1 or several images to crop as input and an output folder in the following window :
After cropping you will obtain in 3 folders and 1 results file :
- coordinates folder contain 1 text file per image analyzed. It contains 3D crop position relative to input image and a summarize of the analyze :
-
nuclei folder contain 1 file per crop obtain with the structure name InputImageName_CropNumber_ChannelNumber. Exemple : RawImage_0_C0.tif
-
zprojection folder contain 1 voxel max projection per input image with the structure name InputImageName_projection. We draw on this projection crop boxes and numbers corresponding, exemple :
- result_Autocrop_Analyse.csv contain a summarize of analyzed images (number of crop per image , OTSU threshold used ...) exemple :
By default there is a list of parameters defined (see next table). You can modify this parameter in graphical interface :
In the new window you can modify parameters :
Description of different parameters you can modify :
| Parameter name | Default value | Value type | Corresponding number in user interface | Description |
|---|---|---|---|---|
| xCropBoxSize | 40 | Int | 1 | Number of voxels added to xmin and xmax of the connected component defining the final box size in x |
| yCropBoxSize | 40 | Int | 2 | Number of voxels added to ymin and ymax of the connected component defining the final box size in y |
| zCropBoxSize | 20 | Int | 3 | Number of voxels added to zmin and zmax of the connected component defining the final box size in z |
| minVolumeNucleus | 1 | Int | 4 | Minimum volume of detected object |
| maxVolumeNucleus | 2147483647 | Int | 5 | Maximum volume of detected object |
| thresholdOSTUcomputing | 20 | Int | 6 | Minimal default OTSU threshold |
| channelToComputeThreshold | 0 | Int | 7 | Channel number used to compute OTSU threshold (Channel 1 is 0 etc) |
| slicesOTSUcomputing | 0 | Int | 8 | Slice start used to compute OTSU threshold |
| boxesPercentSurfaceToFilter | 50 | Int | 9 | Surface percent of boxes to groups them |
| boxesRegroupement | 10 | boolean | 10 | Activation of boxes regroupement |
| xcal | 1 | Double | 11 | X calibration value |
| ycal | 1 | Double | 12 | Y calibration value |
| zcal | 1 | Double | 13 | Z calibration value |
If you are using a special parameter configuration you can use a config file. You can select you config file before analyze :
Config file is text file containing 1 parameter per line composed of ParameterName:ParameterValue. In this example here we want to modify the x y boxes size of crop and the minimal OTSU threshold to detect object :