A Revolution in Cell Analysis

1
ImageStream Operator Training
A Revolution in Cell Analysis
2
The ImageStream System

• ImageStream Imaging Flow Cytometer
• Brightfield, darkfield, and 4 fluorescent images
  at gt15,000 cells/minute
• IDEAS Statistical Image Analysis Software
• Quantitative cellular image analysis and
  population statistics
• Novel Applications
• Translocation, co-localization, cell
  classification, cell cycle, apoptosis, etc.

3
Just in Case
AMNIS (Latin), Stream or Torrent
INSPIRE INstrument Software Processor for
Imaging Research Experiments
IDEAS Image Data Exploration and Analysis
Software
ASSIST Automated Suite of Systemwide
ImageStream Tests
4
ImageStream Workflow

• Experimental Design
• Instrument Calibration with SpeedBeads
• Data acquisition
• Data analysis

5
Experimental Design
A Revolution in Cell Analysis
6
ImageStream Experimental Design

• Considerations
• Selection of cell type
• Selection of probes
• Fluorescence control samples
• Sample prep requirements

7
Experimental Design Overview
44 mm
1) 44 mm channel width 2) Ideally 2x106 cells per
sample 3) Less for single color fluorescent
controls 4) Final Vol. 50 ml in 0.5 ml
microcentrifuge tube 5) Up to 4 fluorochromes,
all 488 excitable 6) 10-20 experimental samples
per hour
8
Fluorescence Controls

• Fluorescence crosstalk control samples
• Unlabeled and single color-labeled cells
• Cell type should be representative of
  experimental sample
• Single color labels should be identical to those
  used in the experimental file the fluorochrome
  MUST be identical
• DNA control separate and run last
• Collected with no brightfield
• Used to guide automated crosstalk correction of
  experimental files

9
Sample Preparation

• Sample Processing
• Follow standard flow cytometric methods for cell
  harvesting, incubation, washing and staining
  (including reagent titration)
• Final concentration of 4x107 cells per ml will
  run at approximately 75 cells per second
• Take care to balance fluorochrome staining
  intensities to avoid saturation of signal from
  bright stains at instrument setup conditions
  necessary for dim stains

10
Experimental Design Overview
44 mm
1) 44 mm channel width 2) Ideally 2x106 cells per
sample 3) Less for single color fluorescent
controls 4) Final Vol. 50 ml in 0.5 ml
microcentrifuge tube 5) Up to 4 fluorochromes,
all 488 excitable 6) 10-20 experimental samples
per hour
11
Calibration and SpeedBeads
A Revolution in Cell Analysis
12
SpeedBeads

• SpeedBeads - Instrument calibration and run-time
  system integrity
• Run-time system integrity
• Maintains continuous synchronization and
  autofocus independent of cell concentration or
  type
• Automatic Instrument Calibrations and Tests
• Optical, illumination, fluidic and camera systems
• Loaded at the beginning of each day and run
  continuously until shut down
• IR laser scattering characteristics monitored
• Automatically classified and not included in
  sample file

13
SpeedBeads and Image Quality

• The optical system focuses on the core stream
  that contains the sample
• Relative to the optics, the core can
• move back - front (z-axis focus)
• move left - right (x-axis core tracking)
• speed up or slow down (y-axis camera
  synchronization)
• SpeedBeads provide continuous feedback to update
  the objective stage position and camera line rate
  to maintain image quality

core with cells
sheath
Channel Image Deviation none
X Y
Z
cuvette
optics
14
Calibration using ASSIST
Automated Suite of Systemwide ImageStream Tests

• ASSIST is a fully automated suite of instrument
  calibrations tests that assures optimal
  performance.
• Tests all major subsystems using a uniform
  particle (SpeedBeads), and produces a report for
  daily quality assurance.
• Typically completes all calibration and tests in
  about 5 minutes.

15
Data Acquisition
A Revolution in Cell Analysis
16
INSPIRE Data Acquisition
INstrument Software Processor for Imaging
Research Experiments

• Brightfield, darkfield and fluorescent images
  collected in 6 channels with adjustable
  sensitivity.
• Brightfield in any channel to accommodate a
  variety of fluorochromes.
• Automated instrument calibration.
• On the fly images and scatter plots allow for
  quick sample assessment.

17
Instrument Run Sequence

1. Load sample
2. Run cells with SpeedBeads
3. Establish stable core fluidics
4. Establish appropriate instrument settings
5. Choose classifiers to distinguish cells, from
   debris
6. Collect data
7. Return sample (optional)
8. Flush sample syringe and lines
9. Load next sample

18
Instrument settings

• Choose channel for Brightfield (blocked for
  fluorescent control)
• Set laser power and/or camera stages to avoid
  camera pixel saturation by monitoring Peak
  Intensity plots
• Adjust laser height to maximize dynamic range of
  488 scatter intensity while still maintaining
  high fluorescence sensitivity
• For samples that contain abundant debris, set
  squelch value to reduce sensitivity of object
  detection so that debris is ignored.
• Note that the excitation detection conditions
  selected for the control (laser power, camera
  staging) MUST be used for the experimental
  samples

19
Choose Classifiers

• Detected objects can be classified in three ways
• Cells
• Beads
• Debris
• Only the Cells make it into your primary data
  file. You can save the bead and debris into
  separate files if you wish.
• Beads are automatically classified
• Cells can be classified based on object feature
  thresholds. Objects that fall outside the
  boundaries of any of the thresholds are
  classified as debris.

20
Instrument Shutdown

1. Optionally return experimental sample to tube
2. Change Sheath tank to Rinse
3. Run sterilize script
4. Powers off illumination (all)
5. Flushes lines
6. Cleans instrument automatically runs detergent,
   alcohol, bleach water

21
Data Analysis Opening Files
A Revolution in Cell Analysis
22
IDEAS Software
Image Data Exploration Analysis Software
Image Gallery
Tabular Data
200 params/cell population statistics object
values
Workspace
23
IDEAS Data Analysis Overview

• The ImageStream collects large numbers of digital
  images into a single file.
• Using image processing algorithms, specific
  features can be quantified from these images.
• IDEAS is the software tool used to analyze and
  report the image data acquired on the IS100
  instrument.

24
IDEAS Data Analysis Overview

• IDEAS Cell image-based informatics
• 6 images per cell, 30 standard features per
  image
• Customizable image display
• User definable features
• Features plotted on histograms or dot plots
• Images linked to plotted data points
• Populations can be created in many ways
• Standard region drawing tools
• Tagged populations
• Boolean combinations of multiple populations
• Multiparametric filter-based
• Full statistics repertoire
• Reporting via copy to clipboard, export to stats
  program
• Batch processing to apply analysis template to
  all files in an experiment

25
Files and their Structure

• Raw Image File (RIF)------------------------------
  --200 MB/10,000 events
• raw instrument data
• collection settings
• Compensated Image File (CIF)-------------------200
  MB/10,000 events
• Corrected for spectral crosstalk
• Corrected for offsets and gains from ASISST
• Determination of object boundaries (segmentation)
• Data Analysis File (DAF)--------------------------
  ---10 MB/10,000 events
• Image Gallery
• Work Area (graphs and specific images)
• Calculated features and statistics
• Saved state of analysis
• Uses the CIF as a database Keep track of where
  you store files! The DAF and related CIF must be
  in the same directory.

26
Opening Data Files
Raw image file (.rif)
Compensation matrix
Save corrected image file (.cif)

• Use file open in IDEAS to open a .rif
• Double click on a data file
• Cntl click on .daf R-click select open
• Open multiple instances of IDEAS to perform
  multi-sample analysis

Select a template
Data analysis file (.daf)
27
Spectral Compensation

• Navigate to an existing compensation matrix.
• Select the number of events to open. To get a
  quick look at the data. Opening 100 events will
  be faster then opening all the events.
• Advanced button reveals all the data file
  corrections that occur when creating the cif.

28
Data File Corrections

• Spectral compensation removes crosstalk.
• Spatial alignment assures direct overlap of each
  channel.
• Camera gain correction provides uniform light
  response.
• Dark current corrections set uniform background
  levels.
• Flow speed normalization corrects for small
  sensitivity changes as a result of flow speed
  changes.
• EDF (extended depth of field) is used only if
  data is collected with the EDF element.
• MTF (modulation transfer function) only used for
  sub-pixel alignments.

29
Corrected Image File (cif)

• Corrected image file has all file corrections
  applied and spectral crosstalk removed.
• Determination of object boundaries for each
  image have been made (segmentation).
• The .cif can be opened in any analysis template.
• Batching a .cif with an analysis template allows
  for quick re-analysis of archived data.

30
Spectral Compensation
Single color control samples used to calculate a
6x6 matrix.
Post-acquisition compensation is applied to
images on a pixel by pixel basis in IDEAS.
SSC Brightfield FITC
PE PE-Alexa610 Draq-5
31
IDEAS Templates

• Templates contain
• image display settings
• workspace histograms
• user defined and experiment specific feature
  calculations
• Gating and population logic
• Used in batch processing to provide uniform
  analysis for all experimental data files.

32
Data Analysis File (daf)

• The daf presents the saved state of the
  analysis.
• Uses the cif as its database for images and
  feature calculation.
• Includes all information from the template.

33
IDEAS Files Review

• Instrument creates Raw Image Files (rif)
• Corrections and segmentation are applied when a
  rif is opened
• Corrected images and segmentation masks are
  stored in a Corrected Image File (cif).
• A cif is loaded into IDEAS using a template file.
  Defined features are calculated for each object.
• Feature values and analysis results are saved in
  a Data Analysis File (daf).

34
Data Analysis IDEAS
A Revolution in Cell Analysis
35
IDEAS layout
The IDEAS window is divided into three major
areas the image gallery, the work area, the
statistics area
36
Image Display Properties

• Set display background and saturation levels
• Edit channel names
• Set color display properties
• Define masks to display and in which channels
• Edit gallery display views to look at
  combinations of channel and composite images
• Define composite images.

37
IDEAS Masks

• Masks
• Set of pixels that make up a region of interest
  in an image
• IDEAS automatically determines a sensitive mask
  for each channel of each object. This mask is
  best used for measuring the overall staining
  intensity detected in a given image. However,
  this mask may be less appropriate for other shape
  related features. IDEAS provides tools to create
  feature-appropriate custom masks
• The user can create custom masks in two ways
• Functionalize an existing mask (erode, dilate,
  fill, threshold, morphology)
• Make complex masks through boolean combinations
• These masks can then be used to build features

38
Masks Region of interest

• Masks define a region of interest for feature
  calculation.
• System masks are all inclusive.
• User defined masks target specific cell
  compartments.
• Complex masks use Boolean logic combine two or
  more masks.

39
IDEAS Masks

• Functionalize a mask
• Example threshold the system mask for the
  channel 6 image (nuclear stain) to constrain the
  region of interest to the region of dominant
  nuclear dye signal.

40
Feature Calculation

• Over 200 intensity and morphology based features.
• Feature calculator that allows for the
  development of novel cell classifiers.
• Each feature can be displayed as univariate or
  bivariate histograms.
• Population statistics are calculated for each
  population in the analysis.

41
IDEAS Plotting feature data

• Histogram image data linkage (click on a bin)

42
IDEAS Plotting feature data

• Scatter plots image data linkage (click on a
  dot)

43
IDEAS Populations

• Population creation (4 ways)
• Regions physically drawn on plots
• Tagging cells
• Boolean combinations
• Filtering (find-like-cells)
• Population display
• Virtual sort in the Image Gallery
• Show/Hide on existing scatter plots

44
IDEAS Creating Populations
Regions physically drawn on plots rectangle,
oval, polygon, line
45
IDEAS Creating Populations
Tagging cells population created by individually
selecting objects from the image gallery and/or
from scatter plots
46
IDEAS Creating Populations
Boolean combinations Complex populations can be
created by combining existing populations with
boolean operators (and, or, not)
47
IDEAS Creating Populations
Filtering (find-like-cells) Populations can be
created by instructing IDEAS to find all the
objects that have similar features to a given
cell or existing population.
48
IDEAS Displaying Populations
Virtual sort in the Image Gallery
49
IDEAS Displaying Populations

• Show/Hide a population on existing scatter plots

50
IDEAS Statistics

• Statistics automatically calculated for each
  feature
• Population stats Count, Total, Gated,
  Plotted
• Feature stats Mean, Median, Mode, Geometric
  Mean, Standard Deviation, CV, Variance
• Displayed under each plot and/or in the
  Statistics Area

51
IDEAS Data Reporting

• Reporting
• Copy image or image gallery to clipboard
• Copy graph and/or stats to clipboard
• Export stats to spreadsheet

52
IDEAS Data Reporting

• Copy image or image gallery to clipboard

Single image
Image Gallery (composite mode)
53
IDEAS Data Reporting

• Copy graph and/or stats to clipboard

Light mode
Dark mode
54
IDEAS Data Reporting

• Export stats or feature data to spreadsheet
• Export single image data to spreadsheet

55
IDEAS File tools

• File tools
• Batch processing
• Apply a compensation matrix and a saved analysis
  template to all files within a given experiment
• Merge files into one file
• Create smaller files from sub-populations
• Save data in .fcs format

56
IDEAS Data Analysis Review
IDEAS is the software tool used to analyze and
report the image data acquired on the
ImageStream. IDEAS allows the user to mine
existing features, create new features, plot
data, perform population statistical analysis and
customize image display.
57
ImageStream Operator Training
A Revolution in Cell Analysis
58
IDEAS Images

• Up to 6 images per object
• Displayed in the Image Gallery or Work Area
• Customizable Image display
• Linear and non-linear display transformation
• False color
• Composites
• Image Line and Region data

59
IDEAS Images
Linear and Non-linear image display transformation
Linear transform
Non-linear transform
60
IDEAS Images
False coloration of greyscale images and
composites
Channel Images of object 2
Bright Field and Composite
61
IDEAS Images
Image Line and Region data
62
IDEAS Features

• Features
• Are attributes related to each object image.
• Most of these features are derived from image
  processing algorithms, and most quantify
  morphologic aspects of the image.
• A large set of features is automatically
  calculated by IDEAS
• The user can create their own features through
  Boolean combinations and/or arithmetic operators
• Some feature rely on a mask or region of
  interest (discussed next). The user can
  calculate standard features based on customized
  masks

63
IDEAS Plotting feature data

• Plots images linked to plotted data
• Histograms and histogram overlays
• Scatter plots
• Linear, log, and Linear/Log transform plotting

64
Choice of Fluorochromes

• Example 488 excitation with BF in channel 2
• choose probes excitable with 488nm light
• 
  Molecular probes
• choose probes with max emission spectra in
  discrete channels
• balance brightness levels of individual probes
• e.g., titrate PI such that you dont saturate
  DNA detection with enough laser power
  to view weaker labels

65
Selection of cell type

• Cell image should fit within a camera channel
• width (88 pixels 44 mm diameter)
• bacteria 1 mm
• lymphocyte 10 mm
• cell lines 15-40 mm
• Cell must not clog instrument
• (250 mm will block tubing)
• Resolution 1.0 mm
• Can be adherent or suspension cells, but must
• be adaptable to flow

250 mm
tubing
channels
44 mm