Preprocessing

Corsen like other image analysis tools need preprocessed images as input to get optimal quality results. Many preprocessing filters exists in image analysis:

  • Adjust brightness/contrast
  • Removing background/dust/saturated pixels
  • Resizing image
  • ...

Preprocessing use in Corsen article

To obtain data published in Corsen article (jourdren et al., 2010), we developed an image processing procedure for automatic extraction of both small particles and aggregates that could not be segmented together using a unique threshold.

The procedure is based on the generation of two distinct images, one containing the tiny objects, the other containing the brightest ones, that are then added up to retrieve all the significant signals.

After applying a threshold on the minimal grey value to remove cellular background, the tiny object image (I1) is obtained by performing a Laplacien filter (r=2), that enhances small particles signal, followed by a median filter (r=2 for mitochondria, r=1 for mRNA). The brightest particle image (I2) is obtained simply by a applying to the initial image a high threshold on the minimal grey value. Selection of the two thresholds needed for generation of I1 and I2 is automated using an analysis of the cell intensity histogram.

According to objects occupancy, the background thresholds values are chosen as follow: T=m+c.o with T, threshold gray value; m, cellular median intensity; o, cellular standard deviation of intensity; and c, a coefficient determined by object occupancy. Typically, for I1 generation, we set c=-2 for mitochondria objects and c E [-1,1] for RNA objects depending on their expression level. For I2 generation, the c value was raised of 2 units. Adding up of I1 and I2 results in an intensity enhanced image that can be submitted to isodata automatic threshold. After eroding (median filter with r=2 for mitochondria and r=1 for mRNA), this binary image was used as a mask on the initial image to set all background pixels to zero, whereas selected object pixels kept their initial intensity. This final image was analysed using Corsen 3D segmentation algorithm with the minimal threshold value set to 1.

Our image processing macro can be freely downloaded and use. As a sample we provide a demonstration cell image and its cell mask. The second script used to apply the particle mask is also available.

The following table summaries the steps of prepropressing we apply on our images before launching Corsen plug-in.

Cell Mask Cell image Processed image Particles mask Final Image
Image # 1 2 3 4 5
Image
Image sample demo_cellmask.tiff demo_cellimage.tiff processed_cellimage.tiff particles_mask.tiff processed_cellimage.tiff
Stack no yes yes yes yes
Image type 8 bits 16 bits 8 bits 8 bits 16 bits
Image values 2 values: 0, 255 0 to 65535 0 to 255 2 values: 0, 255 0 to 65535
Comment Mask of the original image. Creating e.g. by apllying a threshold on original image and a binarization. Original image. Output from microscope. Created from images 1 and 2 by using image processing macro with parameters (1,1,NO). Created from image 3 with ImageJ threshold (Image > Adjust > Threshold: Auto then Apply) then ImageJ filter median (Process > Filter > median: r=1). Created from images 4 and 2 by using apply mask macro. This image is ready to use with Corsen plug-in.