CELLPROFILER PIPELINE MANUAL
For pigment annotation, we used a manual thresholding method with a threshold of 0.8 and ‘typical diameter of objects” was set to 10 to 100 to reduce the number of objects identified. “Typical diameter of objects” was set to 10 to 40 pixels, as by default. To annotate nuclei, we decided to adopt the Otsu method with default parameters except for “threshold correction factor” for which we used value of 1.3 instead of the default 1.0 for more stringent annotation. IdentifyPrimaryObjects: Then, the Hematoxylin stain is used to annotate nuclei, and the pigment stain is used to annotate pigmented regions on the tile.
CELLPROFILER PIPELINE SOFTWARE
The deconvolution of Hematoxylin and Eosin is built-in the software and the pigment color was determined by choosing a custom color profile based on the pigmentation of our images (used image "pigment.png" in the same folder as the. UnmixColors: The pipeline starts by de-convolving the Hematoxylin, Eosin and Pigment signals and generating grayscale images indicating the location of each stain with white color.Our CellProfiler pipeline consists of the following steps: This file structure enables the CellProfiler pipeline to create an output folder for each slide. Files containing the paths to all tiles of the slides of interest are saved as well in a separate folder under the naming convention slide-name_path.txt.Each tile is named according to the formula: slide_name_x_y.jpeg, where x and y are the tile coordinates. 299x299 pixel tiles are saved in folders named after the corresponding slide.The input data need to be organized in the following manner: For our study, tiling is performed using the "0b_tileLoop_deepzoom4.py" script from the DeepPATH pipeline ( ). The pipeline is designed to run for every tile of an H&E slide. Genome Biology, 2006) to perform nuclear and cell annotation on melanoma histopathology slides.ĬellProfiler 3.1.8 version is used (Documentation: ).Īll scripts are developed for the "bigpurple" computing cluster of the NYU Grossman School of Medicine and appropriate SLURM headers are used in the. CellProfiler: image analysis software for identifying and This pipeline was developed using the publicly available CellProfiler software (Carpenter AE et al. This pipeline was developed for our study Kim & Nomikou et al, "A Deep Learning Approach for Rapid Mutational Screening in Melanoma" in biorxiv. A nuclear and cell annotation pipeline for histopathology H&E slides using CellProfiler.
0 kommentar(er)
