Input image Get Linked Channel : nuclei Set scaling/LUT Wiener Filter Smooth Plot Morphological Parameters Morphological operator Mask Image Copy Wiener Filter Smooth Set Segmentation Classifiers Simple Segmentation Set Segmentation Classifiers Rename Image Create ROI Threshold Subtract Background Threshold If Nothing EndIf 2D DFT Butterworth BP filter tiled Threshold Copy 2D Median filter Image Aritmetics In Place Close EndIf Wait for all inputs Select Inpaint Mask If Image Arithmetic In Place EndIf Crop in place Simple Segmentation Select Set scaling/LUT Branch Copy EndIf Set scaling/LUT 2D DFT Butterworth BP filter tiled Threshold Copy 2D Median filter Image Aritmetics In Place Close EndIf Wait for all inputs Select Inpaint Mask 2D DFT Butterworth BP filter tiled Threshold Copy 2D Median filter Image Aritmetics In Place Close EndIf Wait for all inputs Select Inpaint Mask Get Linked Channel : label1 Get Linked Channel : label2 Plot Segment Intensities Reevaluate Segments Copy Morphological operator Threshold Image Aritmetics In Place Close EndIf - seg Select Copy Rename Image Select by Number Select by Number Plot Segment Intensities Reevaluate Segments Copy Morphological operator Threshold Image Aritmetics In Place Close EndIf - seg Select Copy Rename Image Plot Segment Intensities Reevaluate Segments Copy Morphological operator Threshold Image Aritmetics In Place Close EndIf - seg Select Copy Rename Image 2D DFT Butterworth BP filter tiled Threshold Copy 2D Median filter Image Aritmetics In Place Close EndIf Wait for all inputs Select Inpaint Mask Get Linked Channel : label3 Copy Copy Copy Rename Image Plot Segment Intensities Reevaluate Segments Copy Morphological operator Threshold Image Aritmetics In Place Close EndIf - seg Select Copy Rename Image 2D DFT Butterworth BP filter tiled Threshold Copy 2D Median filter Image Aritmetics In Place Close EndIf Wait for all inputs Select Inpaint Mask Get Linked Channel : label4 Copy Subtract Background Subtract Background Subtract Background Subtract Background Subtract Background Select Fluor/OD Select Fluor/OD Select Fluor/OD Select Fluor/OD Optical density Optical density Optical density Optical density EndIf EndIf EndIf EndIf Get Linked Channel : nuclei Set Reference Image Set Reference Image Rename Image Copy Threshold Close Image Arithmetic In Place If Subtract Background EndIf If Subtract Background EndIf If Subtract Background EndIf If Subtract Background EndIf If Subtract Background EndIf Rename:nuclei Copy Optical density Threshold Threshold 2D DFT Threshold Optical density Threshold Threshold 2D DFT Threshold Optical density Threshold Threshold 2D DFT Threshold Optical density Threshold Threshold 2D DFT Threshold Add extra processing below Add extra processing above Add extra processing below Add extra processing above Add extra processing below Add extra processing above Add extra processing below Add extra processing above Run EXE Set Segmentation Classifiers Reevaluate Segments Set Segmentation Classifiers Copy ROIs Get Linked Channel Select by Number Rename Image Rename Image Rename Image Rename Image Link Windows Link Windows Link Windows Link Windows FAST Analysis Pipeline - Cellpose Segmentation 5 Fully Automated Senescence Test - AI This is a modified version of the original "AI Fluorescence and absorbance histometry using nuclear marker (1-4 labels - advanced background options)" V5 pipeline. Parameters have been changed to reflect the xGAL + EdU analysis. Assay protocol (see details on protcols.io https://dx.doi.org/10.17504/protocols.io.kxygx3ypwg8j/v1): 7.2 Determine dark current 9.1 Create fluorescence background reference image (optional) - pipeline: Create background reference image(s) for multiwell plate using median of wells 9.4 Create blank reference image for xGAL (required) - pipeline: Create BLANK reference image for multiwell plate using median 10.2 Set pipeline parameters: * Channel numbers for nuclei segmentation, labels * Subtract reference image background and shading correct first (all fluorescence channels): see above at #8.1 * Number of x and y tiles * Nucleus: Cellpose diameter (pixels) * Perinuclear ring width (pixels) label #3 * Optical density and shading: optional dark current intensity value: as determined at #7.2 (required, or use dark current reference image) 10.3 Verify that the pipeline parameters have been adjusted properly, it is recommended to run the pipeline on individual wells for each condition to be tested (e.g. a senescent and a control well) to ensure that: 1) The nuclear segmentation was performed correctly 2) The perinuclear rings to measure SA-B-Gal staining do not have excessive overlap with nearby perinuclear rings. Pipeline parameters can be iteratively modified and tested until the user is satisfied for both points above. 10.4 Run Pipeline (blue fast forward) drop-down menu, click "Run Pipeline on All Stage Positions" Generic description of the pipeline: Measurement of fluorescence intensities or optical densities over nuclei and perinuclear areas using a nuclear marker and 1-4 label channels. Nuclei are detected with neural network. The pipeline can run in single or tiled images and in whole wells with automatic suppression of outside of the well areas and debris (see “Suppress debris and outside of well areas”). Different measurements may be done in the same label channel, and each label can be defined as fluorescence or optical density (“Channel type”). The pipeline can also be executed on image series for faster analysis compared to frame-by-frame operation. Frames of an image series are scaled independently. For the SA-bGAL assay use "Densitometry with thin process suppression" as channel type and set the sensitivity of this at "Thin process suppression for optical density (maximal diameter in pixels)". This value is the maximal process diameter, thinner details will be suppressed. Thus a larger value provides greater suppression. This is performed by bandpass spatial filtering between the “Local background: Spatial filtering: Largest object size…” cut on and "Thin process suppression for optical density (maximal diameter in pixels)" cut off. Use additional rolling ball background subtraction. Using additional spatial filtering background subtraction will not have much further effect. Sample: fixed or live cells stained for nuclei e.g. with Hoechst 33342 or DAPI, and showing other fluorescence of interest. Input: 2-4 channels fluorescence image showing a nuclear marker and immuno- or other fluorescence visible over the nuclei or in perinuclear areas. Low or medium magnification 10-40x. May be tiled and whole well image. The area outside of the well must be bright in the fluorescence image for auto detection of the well. Whole slide scans are also workable using the Multi-Dimensional Open/Settings/Crop multi-resolution by ROIs, analyzing multiple regions/scenes loaded as an image series. Output: 1 morphological parameters of choice of nuclei and 4 intensity or optical density measurements in the same or different label channels. For intensity measurement the “Intensity measurement type of label #” statistics (Mean, Sum, Variance, Punctate over diffuse ratio) is calculated over nuclei or perinuclear area see (“Measure label # fluorescence over” and “Perinuclear ring width”). Output data are single cell values or histogram (see “Histogram or distribution calculation”). Note that the order of labels measured in the output worksheet may vary due to parallel processing. When “Position names and microplate worksheet output” is Yes, labels will show up in renumbered channels as: Ch0: Nuclei Morphology Ch1: Label #1 Ch2: Label #2 Ch3: Label #3 Ch4: Label #4 Reference image operation: by default, the calculation of nuclei is self-referenced. Thus, it is independently calculated for each image. This is set by the “Positive control reference image operation for nuclei detection” = ”Self-referenced” option. With the neural network version of this pipeline referece image may be used for rescaling intensities before nuclei detection. If cell density varies between samples, or if some samples have very few cells , use positive control reference. The positive control is one of the images from the data set with about maximum number of cells and positive cells. To make reference image set “Positive control reference image operation for nuclei detection” = ”Make reference image”. Process a positive control sample. The reference images will be minimized, and reused at the following operations. Then set “Positive control reference image operation for nuclei detection” = ”Use reference image” and process all data. To process a different data set, close first reference images by “File/Close All Reference Images” and make new reference images. Background subtraction: The background subtraction technique can be specified for each channel. Choose percentile-based global or local background subtraction methods. All channels use the same method, and the parameters are at the end of the pipeline parameter list. For median or special filtering, the object size must be bigger than the cell size. Tiled images: the background tiling pattern or vignetting is efficiently removed by spatial filtering if the recording was performed without overlap and image registration. Provide the number of tiles in x and y direction in the “Local background: Spatial filtering: Number of tiles in x” and “…y” parameters. Debris avoidance: to increase sensitivity, increase “Large debris removal sensitivity (%)” and decrease “Large debris minimum size (length in pixels)”. To decrease sensitivity, do the opposite. Debris avoidance may avoid cells or positive cells, if the cell count or positive cell count is too low. In this case set a small (non-zero) %, e.g. 1 as sensitivity. Use Cellpose GUI to fine tune thresholds. Start GUI in Pipelines/Using External Programs/Cellpose/Launch cellpose GUI. Save images to be loaded into the GUI use the Files/Export RGB Full Size Image, or by disabling temp folder deletion (see switch in the Run EXE function), and locating the temp folder. Cellpose will run faster if the same number of images are processed as an image series than one-by-one. Further speed increment can be achieved by increasing batch size and number of parallel processes, depending on the available system memory and the image size. Use batch size of 64 and 4 parallel processes with 24GB GPU. Adding more processing, image filtering to labels: Open pipeline window, unlock editing and insert image processing function in between the "Add extra processing below" and "Add extra processing above" placeholders. For this pick the function to be added by pressing the F icon in the toolbar of the Pipeline window. Place this function in an empty space, and then drag it and drop it on the "Add extra processing below" label. Add more processing, or reconnect the "Add extra processing above" label by dropping it on the last added function. This pipeline uses Cellpose 2.0. Pachitariu M, Stringer C. Cellpose 2.0: how to train your own model. Nat Methods. 2022;19(12):1634-41 https://github.com/MouseLand/cellpose Keywords: cytometry, histometry, nucleus, perinuclear, intensity, area, shape, histogram Version History V1: based on “Microplate whole well cell count (automatic well positioning, with positive control)” and “Seahorse well cell count with nuclear stain”. V3: Added control on how Excel Data is recorded Debris avoidance also removes zero edges resulting from stitching Position names are shown on labels Reference image background subtraction and shading correction Optional clipping of bright foci during segmentation Accidental loopback after background subtraction in label#3 removed Explicit gating for nuclei area New "Densitometry with thin process suppression" option for channel type and "Thin process suppression for optical density (maximal diameter in pixels)" for setting its sensitivity Added "Add extra processing below" and "Add extra processing above" placeholders Added Histogram Range parameters for each label V4: Cellpose flavor of the above pipeline. V5: ROIs are copied from original images, so ROIs may be drawn on the source images to gate the readout. Echa label can be named now.