Try   HackMD
tags: image inForm2.3 spotfire

Immunohistochemistry (IHC) image analysis


Tissue segmentation, cell segmentation, cell phenotyping using software InForm version 2.3

Batch Analysis
Click on Batch Analysis
Check all 'images to export' except for the two TIFF
Select JPEG as output image format
Check all boxes of 'Tables to export'
Click 'Browse' to select export directory
Click 'Add Images' to select testing images from a testing folder. In this run, five images from the RBH were used as training images, the other images were used as testing images.
Hit 'Run'. This applies the training algorithm to the images selected to segment tissue (into Tumor, Stroma, Glass, others), segment cells (into Nuclei and cytoplasm) and phenotype cells (into tumor cells, immune cells based on the staining of CD8, CD96. Immune cells can be CD8+, CD96+, or double positive).
'Batch progress' appears. This will take a long time.

Image Not Showing Possible Reasons
  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported
Learn More →


Visualise a marker intensity on IHC images

In this part, I applied different filters to the merged tissue segmentation file and visualised a marker using the R package imagick. One of the annotated images look like

Image Not Showing Possible Reasons
  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported
Learn More →

In this folder, entire CD8 mean >=7 is used as the filter, and entire CD8 mean is visualised
L:/Lab_MarkS/lunC/Immunohistochemistry_images/data_output/visual=entire-CD8-mean_filter=entire-CD8-mean/

In this folder, entire DANM1 mean >=7 is used as the filter, and entire DANM1 mean is visualised
L:/Lab_MarkS/lunC/Immunohistochemistry_images/data_output/visual=entire-DANM1-mean_filter=entire-DANM1-mean/

In this folder, entire CD8 mean >=7 and entire DANM1 mean >=7 are used as the filter, and entire CD8 mean is visualised
L:/Lab_MarkS/lunC/Immunohistochemistry_images/data_output/visual=entire-CD8-mean_filter=entire-CD8-mean_AND_entire-DANM1-mean/

In this folder, entire CD8 mean >=7 and entire DANM1 mean >=7 are used as the filter, and entire DANM1 mean is visualised
L:/Lab_MarkS/lunC/Immunohistochemistry_images/data_output/visual=entire-DANM1-mean_filter=entire-CD8-mean_AND_entire-DANM1-mean/

Mean cell counts and ratio results are exported in the file L:/Lab_MarkS/lunC/Immunohistochemistry_images/outcome/counts_cells_patientID.tsv

Column Definition
count_all_cells Number of cells per sample
count_cells_CD8_Mean_GTE_7 number of cells with Entire Cell CD8 Mean >=7 per sample
count_cells_DANM1_Mean_GTE_7 number of cells with Entire Cell DANM1 Mean >=7 per sample
count_cells_CD8_Mean_GTE_7_DANM1_Mean_GTE_7 number of cells with both Entire CD8 mean and Cell DANM1 Mean >=7 per sample

ID columns that are matched to patientID column in the cell count data are shown as bold; unmatched shown as italic

Image Not Showing Possible Reasons
  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported
Learn More →


Finding a cutpoint using survival analysis data

Determine optimal cutpoints for numerical variables in survival plots
Script location L:\Lab_MarkS\lunC\survival_analysis\scripts\Find_cutpoint_with_R.R