Tissue Image Analysis 2.0 Training Presentation AWS
Tissue Image Analysis 2.0 Training Presentation
Control #: 13MAN1231.A1 Effective Date: 12-Jul-12 ECO #: 3472
2012
Tissue IA 2.0 Training Presentation
1
Intended Use IMPORTANT: SlidePath applications are not cleared by the FDA, Health Canada, or in the EU for diagnostic or clinical use. All applications are intended solely for use in the research or educational setting, such as university or pharmaceutical development. These applications are described as Research Applications or Research Use Only.
2012
Tissue IA 2.0 Training Presentation
2
What is Tissue Image Analysis
2012
High-throughput web enabled image analysis solution for digital slides Control and review image analysis online anywhere using web browser Analyze whole slides, areas of interest or TMAs Queue multiple images for batch processing Automatically integrate data with existing TMA/Research data in Distiller and/or export to Excel spreadsheets
Tissue IA 2.0 Training Presentation
3
Introduction to Tissue Image Analysis
2012
Tissue IA 2.0 Training Presentation
4
OverView of Tissue IA
2012
Set preferences in Digital Image Hub Determine the color definition file What is positively stained Use pre defined deconvolution colour definition files Determine the algorithm preferences Select the preferences for the biomarker of interest Test the image analysis in a variety of fields and slides
Tissue IA 2.0 Training Presentation
5
OverView of TIA
2012
Run image analysis in Digital Image Hub Run on whole slides, TMAs (with appropriate licence) annotations Results are added as slide metadata Run high throughput image analysis in Distiller (with an applicable licence) Run on TMAs Results are automatically entered into records Each user will only have access to the preferences and color definition files they created
Tissue IA 2.0 Training Presentation
6
Optimizer Harness
2012
Tissue IA 2.0 Training Presentation
7
Optimizing Harness
Use the Harness to optimize the color definition files and algorithm setting before sending to high throughput analysis
Color Definition file
Algorithm settings
Preference File
High through put analysis 2012
Tissue IA 2.0 Training Presentation
8
Image Analysis Harness Optimizer
2012
In Digital Image Hub, select a slide In the slide viewer, zoom to 20x magnification and select 'Image Analysis Harness Optimizer'
Tissue IA 2.0 Training Presentation
9
How To Set Color Definition Files
2012
Tissue IA 2.0 Training Presentation
10
Color Definition File
2012
Determine what is positively stained Select 'Manage Color Definition files'
Tissue IA 2.0 Training Presentation
11
Color Definition Files
2012
A list of color definition files will appear Create, rename, copy, edit or delete files Preview results of color definition files Note: List of preferences is user specific, i.e. each user will only have access to their own preferences
Tissue IA 2.0 Training Presentation
12
PreLoaded Color Definition Files
2012
Default DAB Deconvolution processing allows for a More advanced method of color separation Preloaded Deconvolution Files: Deconvolution – Haematoxylin Deconvolution – AEC Deconvolution – DAB Deconvolution – Eosin Deconvolution – Fast Blue Deconvolution – Fast Red Deconvolution – Methyl Green
Tissue IA 2.0 Training Presentation
13
Create a Color Definition File
2012
Select pixels to include in the color definition file Green = pixels already included Select other colors to indicate positive pixels Use up / down arrow to increase / decrease number of pixels included
Tissue IA 2.0 Training Presentation
14
Inverted Screen
Use the 'T' key to toggle to the inverted screen Select pixels to be subtracted from the color definition file Red = pixels not included
color definition mask
2012
Inverted Screen mask
Tissue IA 2.0 Training Presentation
15
Test Color Definition Files
2012
Objective: record the full representation of positive color, regardless of location Test the color definition file in a variety of areas across a number of slides Test on areas of high and low intensity staining
Tissue IA 2.0 Training Presentation
16
Demonstration: Color Definition Files
2012
Tissue IA 2.0 Training Presentation
17
How To Set Algorithm Preferences
2012
Tissue IA 2.0 Training Presentation
18
Select the Algorithms There are two standard immunohistochemical algorithms Measure Stained Area Measure Stained Cells
2012
Tissue IA 2.0 Training Presentation
19
Select Algorithm Preference
2012
A list of preferences will be displayed Cannot create new preferences from scratch, must copy an existing default preference Test, rename, edit and delete Note: List of preferences is user specific, i.e. each user will only have access to their own preferences
Tissue IA 2.0 Training Presentation
20
Running the Algorithm on a Field of View or Area
2012
Tissue IA 2.0 Training Presentation
21
Field of View or Area
2012
Run the algorithm on the field of view (area which is visible in the viewer at 20x) Run algorithm on a selected area by selecting the square
Tissue IA 2.0 Training Presentation
22
Creating Preference File for Measure Stained Area Algorithm
2012
Tissue IA 2.0 Training Presentation
23
Select the Color Definition File
2012
Up to three classes/ color definition files can be defined Analysis will be carried out on all three classes simultaneously Choose either a pre loaded color definition file or one that has been created/edited by the user Use lock to freeze the preferences warning this action cannot be undone!
Tissue IA 2.0 Training Presentation
24
Input Parameters Goal: Adjust the input parameters to optimize the Algorithm for the biomarker of interest Changing the input parameters will affect the over all results
2012
Tissue IA 2.0 Training Presentation
25
Input Parameters Measurement Units Define the measurement units 0 = µm, 1= mm, 2 = pixels Calibration setting Set default calibration setting Range between 0-10 Use if there is no magnification information attached to the image (i.e. Jpeg images ) Deconvolution setting Select color deconvolution Enable = 1, select if deconvoluted color have been selected
2012
Tissue IA 2.0 Training Presentation
26
Tissue Threshold Intensity
Determine what is background and what is tissue Background Intensity Setting Select a value between 0 (black) and 255 (white) Typically set between 210 – 240
Example of original tissue
2012
Threshold set to 50
Threshold set to 210
Tissue IA 2.0 Training Presentation
Threshold set to 250
27
Deconvolution Threshold Setting Determine the intensity threshold for each of the color Classes defined Maximum intensity of a pixel in the Class to be considered positive Select a value between 0 (black) and 255 (white)
Example of original tissue
2012
Example of threshold setting 50
Tissue IA 2.0 Training Presentation
Example of threshold setting 200
28
Measure Stained Area Results
2012
Run the positive pixel algorithm on a field of view or area Back to refine preferences Reanalyse Store results will create an annotation with the results of the image analysis
Tissue IA 2.0 Training Presentation
29
Measure Stained Area Results
Measurement units
As defined during input settings
Total tissue area
Total area of the tissue which has been identified
Positive Area of Class A/B/C
Total area which is positive for each of the classes defined
2012
Tissue IA 2.0 Training Presentation
30
Measure Stained Area Results
Average staining intensity class A/B/C The modal value of a greyscale intensity histogram of the positively stained
pixels Used to calculate Staining Concentration
2012
Staining concentration Class A/B/C Measure of the concentration of the stain within the tissue Co-localization of Classes The area of overlap between classes Pearsons correlation coefficient The correlation between two classes
Tissue IA 2.0 Training Presentation
31
Tissue masks
Tissue mask represents the deconvoluted image for each class defined, threshold cut off for each class defined and total mask
Example of masks images where two color classes were applied. Class A: DeconvolutionHaematoxylin, Class B: Deconvolution -DAB
Original tissue
Deconvoluted DAB mask
Tissue threshold mask
Deconvoluted Haematoxylin mask Haematoxylin intensity threshold
DAB intensity threshold
Class 2012
Tissue IA 2.0 Training Presentation
Total mask: Haematoxylin : Red, DAB: Green, AB: Yellow 32
Tissue Masks
2012
Tissue masks apply primary colours to each class Class A: Red Class B: Green Class C: Blue The co-localization of these channels is a mix of both colours Class AB: Yellow Class BC: Magenta Class AC: Cyan Class ABC: Grey
Tissue IA 2.0 Training Presentation
33
Demonstration: Algorithm Preferences
2012
Tissue IA 2.0 Training Presentation
34
Measure Stained Cells Algorithm
2012
Tissue IA 2.0 Training Presentation
35
Measure Stained Cells
2012
Measure stained cells replacing the Nuclear, membrane and intercellular algorithms With the use of new input parameters including Nuclear Heterogeneity detection, TIA 2.0 has more advanced nuclear detection
Tissue IA 2.0 Training Presentation
36
Measure Stained Cells
Measure stained cells replacing the Nuclear, membrane and intercellular algorithms Analyse nuclear, cytoplasm or membrane staining separately or a combination of any two Up to three of the following (as long as two of them use the same color definition file) nuclear, cytoplasm or membrane marker
Set input parameters for each of the biomarkers
Nuclear Counterstain
Results output: Separate results for each of the biomarkers selected. For two biomarkers are defined, combination and co-localization results will be output
2012
Tissue IA 2.0 Training Presentation
37
Color Definition File
Select a color definition file for nuclear counterstain
AND
2012
Of the following: nuclear marker, cytoplasmic marker, membrane marker If all cellular compartments are selected, ensure that two of the compartments have the same colour definition file assigned
Tissue IA 2.0 Training Presentation
38
Input Parameters Measurement units Define the measurement units
0 = µm, 1= mm, 2 = pixels Tissue Threshold Determine what is background and what
is tissue Default calibration Set default calibration setting Range between 0-10
(use if there is no magnification provided i.e. Jpeg images)
2012
Tissue IA 2.0 Training Presentation
39
Nuclei Heterogeneity
How similar the nuclei are in the tissue Range between 0 – 4, 0 = nuclei are similar, >1 = increasing diversity in the tissue from darkest to lightest
0
2012
Increased diversity from darkest to lightest
Tissue IA 2.0 Training Presentation
4
40
Nuclei Heterogeneity Original tissue
2012
Nuclei Heterogeneity = 1
Tissue IA 2.0 Training Presentation
Nuclei Heterogeneity = 4
41
Strength of the Nuclear Counter Stain
Define if the counter stain is strong or weak Range between 0-2, 0= strong, 2 = weak
High Contrast
2012
Low intensity
Tissue IA 2.0 Training Presentation
42
Strength of the Nuclear Counter Stain Use 0 for high Contrast:
Use 2 for low contrast:
2012
Tissue IA 2.0 Training Presentation
43
Nuclear Window
Define the size of the window around the nucleus by setting its radius Values in units defined in measurement input , i.e. if measurement units are set to pixels, then this parameter is in pixels Affects nuclei per window and nuclear area per window settings SlidePath recommends this setting be left at 37 µm to start
radius
2012
Tissue IA 2.0 Training Presentation
44
Nuclear Area
Exclusion parameter Range between 0 -10,000 units squared as defined in measurement units
50µm2
150µm2
50µm2
150µm2
Area:: 100µm2
Area: 300µm2
2012
Nuclei of interest included The slider is set to 50µm and 150µm, nuclei of interest included as its area is 100µm2
Nuclei of interest excluded The slider is set to 50µm and 150µm, nuclei of interest excluded as its area is 300µm2
Tissue IA 2.0 Training Presentation
45
Nuclear Area Nuclear area set between 0- 1000
2012
Nuclear area set between 0- 10
Tissue IA 2.0 Training Presentation
46
Nuclei per window
2012
Exclusion parameter A “window” = as defined in the input setting Range: 0 – 1000 cells per window Cell will be counted if its centre point lies within the window
Tissue IA 2.0 Training Presentation
47
Nuclei Per window 5
20
Nuclei of interest excluded The slider is set to 5 and there are only 3 surround nuclei in the window (green)
5
20
Nuclei of interest included The slider is set to 20 and there are 10 surround nuclei in the window (green)
2012
Tissue IA 2.0 Training Presentation
48
Nuclei Per window Nuclei per window set between 0-100
2012
Nuclei per window set between 60-100
Tissue IA 2.0 Training Presentation
49
49
Nuclear Area Per Window
2012
Exclusion parameter Based on percentage of area within the window that is taken up by nuclei Range: 0 – 100% Nucleus is counted if its centre point lies within the window
Tissue IA 2.0 Training Presentation
50
Nuclear Area Per Window Nuclei of interest excluded Nuclei of interested excluded, as area taken up by surrounding nuclei is 10%
Nuclei of interest excluded Nuclei of interested excluded, as area taken up by surrounding nuclei is 80%
Nuclei of interest included Nuclei of interested included, as area taken up by surrounding nuclei is 70%
Nuclei of interest included Nuclei of interested included, as area taken up by surrounding nuclei is 70% 2012
Tissue IA 2.0 Training Presentation
51
Nuclear Area per Window % Nuclear per window set between 0-100%
2012
% Nuclear per window set between 10-50%
Tissue IA 2.0 Training Presentation
52
Cell Area
Eliminate cells on the basis of size Range between 0 – 10,000 (measurements units as defined during input parameters)
50µm2 150µm2
Cell of interest included The slider is set to 50µm and 150µm, cell of interest included as its area is 100µm2
Area: 100µm2
50µm2
Area::
2012
300µm2
150µm2
Cell of interest excluded The slider is set to 50µm and 150µm, cell of interest excluded as its area is 300µm2
Tissue IA 2.0 Training Presentation
53
Cell Area
Example of original tissue
2012
Threshold set from 0 -500
Tissue IA 2.0 Training Presentation
Threshold set between 0-50
54
Maximum Cell Radius
The maximum cell radius which should be included for analysis Range between 0 – 1000 (measurement as defined in ‘Measurement units’ input
parameters)
2012
Cell radius mainly influences how the cells are modeled (i.e. how much the cell boundary expands during the cell prediction)
Tissue IA 2.0 Training Presentation
55
Nuclear Staining Intensity Cutoff
All pixels with intensity higher than defined in deconvoluted nuclear staining image are considered negative Range 0-255
0
2012
255
Tissue IA 2.0 Training Presentation
56
Nuclear Staining Intensity Cutoff Nuclear staining intensity cut off set at 210
2012
Nuclear staining intensity cut off set at 150
Tissue IA 2.0 Training Presentation
57
% of Stained Pixels in a Nucleus Cutoff setting
Categorization parameter Range: 0 – 100% Nuclei with a lower percentage of positive staining will be categorized as negative
0
2012
100%
Tissue IA 2.0 Training Presentation
58
58
% of Stained Pixels in a Nucleus Cutoff setting Nuclear stained area cut off set 80%
2012
Nuclear stained area cut off set 10%
Tissue IA 2.0 Training Presentation
59
Strong/Moderate/Weak Nuclear Staining Intensity Cutoff
2012
Categorization parameter Range: 0 – 255 0 represents minimum intensity (black), 255 represents maximum intensity (white), in a greyscale intensity range
Tissue IA 2.0 Training Presentation
60
Strong/Moderate/Weak Nuclear Staining Intensity Cutoff
2012
Strong/Moderate Staining Intensity Cutoff Setting Nuclei with a level below this input are categorized as strong Moderate/Weak Staining Intensity Cutoff Setting Nuclei with a value above this input are categorized as weak Nuclei with values between are categorized as moderate
Tissue IA 2.0 Training Presentation
61
Strong/Moderate/Weak Nuclear Staining Intensity Cutoff Staining intensity low threshold set to 96, high threshold set to 196
2012
Staining intensity low threshold set to 163, high threshold set to 196
Tissue IA 2.0 Training Presentation
62
Nuclear Staining Filters
2012
Define if all, only positive, or only negative, nuclear staining should be included in results and output image Parameter works on cells not pixels Range 0-2, 0 representing all nuclei, 1 representing only positive nuclei, 2 representing only negative nuclei
Tissue IA 2.0 Training Presentation
63
Nuclear Staining Filter Include all cells and nuclei
2012
Include only positive cells
Tissue IA 2.0 Training Presentation
Only include negative cells
64
Cytoplasmic Input Parameters
Cytoplasmic parameters will only have an effect on Cytoplasmic input parameters if cytoplasmic stain is selected
Cytoplasmic input parameters Cytoplasmic Staining Intensity Cut-off % of Cytoplasmic Staining Area in a Cell Cut-off Strong/Moderate/Weak Cytoplasmic Staining Intensity Cut-off Cytoplasmic staining filter
2012
Tissue IA 2.0 Training Presentation
65
Cytoplasmic Staining Intensity Cutoff
All pixels with intensity higher than defined in deconvoluted cytoplasmic staining image are considered negative Range 0-255
0
2012
255
Tissue IA 2.0 Training Presentation
66
Cytoplasmic Staining Intensity Cutoff Original image
2012
Cytoplasmic stained intensity cut off 30
Tissue IA 2.0 Training Presentation
Cytoplasmic stained intensity cut off set 220
67
% of Cytoplasmic Staining Area in a Cell Cut-off
Categorization parameter Range: 0 – 100% % of Stained Pixels in cytoplasm Cutoff setting Cytoplasm with a lower percentage of positive staining will be categorized as negative
0
2012
100%
Tissue IA 2.0 Training Presentation
68
% of Cytoplasmic Staining Area in a Cell Cut-off Original tissue
2012
Cytoplasmic stained area cut off set to 75%
Tissue IA 2.0 Training Presentation
Cytoplasmic stained area cut off set to 95%
69
Strong/Moderate/Weak Cytoplasm Staining Intensity Cutoff
2012
Strong/Moderate Staining Intensity Cutoff Setting Cytoplasm with a level below this input are categorized as strong Moderate/Weak Staining Intensity Cutoff Setting Cytoplasm with a value above this input are categorized as weak Cytoplasm with values between are categorized as moderate
Tissue IA 2.0 Training Presentation
70
Strong/Moderate/Weak Cytoplasm Staining Intensity Cutoff Original tissue
2012
Strong/Moderate/Weak cytoplasm staining intensity cutoff set between 100-150 (From 0-100: strong, 100-150: moderate, 150-255: weak)
Tissue IA 2.0 Training Presentation
Strong/Moderate/Weak cytoplasm staining intensity cutoff set between 100-150 (From 0-160: strong, 160-220: moderate, 220-255: weak)
71
Cytoplasm Staining Filters
2012
Define if all, only positive, or only negative, Cytoplasm staining should be included in results and output image Range 0-2, 0 representing all cells, 1 representing only positive cells, 2 representing only negative cells
Tissue IA 2.0 Training Presentation
72
Membrane Input Parameters
Membrane parameters will only have an effect on membrane input parameters if cytoplasmic stain is selected
Membrane input parameters Membrane Staining Intensity Cut-off Percentage of Membrane Staining Area in a Cell Cut-off Strong/Moderate/Weak Membrane Staining Intensity Cut-off Membrane staining filter
2012
Tissue IA 2.0 Training Presentation
73
Membrane Staining Intensity Cutoff
All pixels with intensity higher than defined in deconvoluted membrane staining image are considered positive Range 0-255
0
2012
255
Tissue IA 2.0 Training Presentation
74
Membrane Staining Intensity Cutoff Original Tissue
2012
Membrane staining intensity cutoff set to 75
Tissue IA 2.0 Training Presentation
Membrane staining intensity cutoff set to 220
75
% of Stained Pixels in a Membrane Cutoff setting
Categorization parameter Range: 0 – 100% % of Stained Pixels in a membrane Cutoff setting Membrane with a lower percentage of positive staining will be categorized as negative
0
2012
100%
Tissue IA 2.0 Training Presentation
76
% of Stained Pixels in a membrane Cutoff setting Original tissue
2012
Membrane staining intensity cuttoff set to 75%
Tissue IA 2.0 Training Presentation
Membrane staining intensity cuttoff set to 95%
77
Strong/Moderate/Weak Membrane Staining Intensity Cutoff
2012
Strong/Moderate Staining Intensity Cutoff Setting Membrane with a level below this input are categorized as strong Moderate/Weak Staining Intensity Cutoff Setting Membrane with a value above this input are categorized as weak Membrane with values between are categorized as moderate
Tissue IA 2.0 Training Presentation
78
Strong/Moderate/Weak Membrane Staining Intensity Cutoff Original tissue
Strong/Moderate/Weak membrane staining intensity cutoff set between 100-150 (From 0-100: strong, 100-150: moderate, 150-255: weak)
Strong/Moderate/Weak membrane staining intensity cutoff set cutoff set between 160-220 (From 0-160: strong, 160-220: moderate, 220-255: weak)
Strong Membrane: Violet, Moderate membrane: Pink, Weak Membrane: Rose 2012
Tissue IA 2.0 Training Presentation
79
Membrane Staining Filters
2012
Define if all, only positive, or only negative, Membrane staining should be included in results and output image Range 0-2, 0 representing all nuclei, 1 representing only positive cells, 2 representing only negative cells
Tissue IA 2.0 Training Presentation
80
Membrane Filters All membrane include (setting= 0)
2012
Only included positive membrane setting = 1
Tissue IA 2.0 Training Presentation
Only included negative membrane setting = 2
81
Results
Measurement units
As defined during input settings
Total tissue area
Total area of the tissue which has been identified
Total number of cells
The total number of cells that have been identified in the tissue
2012
Tissue IA 2.0 Training Presentation
82
Results
Histoscore
Histoscore of the accepted nuclei
Total Accepted Nuclei
All nuclei that were not eliminated by exclusion criteria
Number and percentage of negative, weak, moderate, strong intensity nuclei
The total number nuclei Categorized as having negative, weak, moderate, strong intensity by value and percentage
2012
Tissue IA 2.0 Training Presentation
83
Results
Average nuclear staining intensity The modal value of a greyscale intensity histogram of the positively stained
pixels Used to calculate Staining Concentration
2012
Average nuclear Staining absorbance Measure of the absorbance of the stain within the tissue Percentage nuclear area in Tissue The percentage of nuclear area
Tissue IA 2.0 Training Presentation
84
Cytoplasmic Marker Results
Cellular H-Score for cytoplasmic staining Number and % of cells with negative cytoplasmic staining Number and % of cells with positive cytoplasmic staining Number and % of weak, moderate and strong intensity stained cytoplasm cells Average cytoplasmic staining intensity – of positive area only Average cytoplasmic staining absorbance - of positive area only % positive cytoplasmic area in tissue
Note: Only visible if cytoplasmic marker is selected
2012
Tissue IA 2.0 Training Presentation
85
Membrane Marker Results
Cellular H-Score for membrane staining - Number and % of cells with negative membrane staining Number and % of cells with positive membrane staining Number and % of weak, moderate and strong intensity stained membrane cells Average membrane staining intensity – of positive area only Average membrane staining absorbance – of positive area only % positive membrane area in tissue
Note: Only visible if membrane marker is selected
2012
Tissue IA 2.0 Training Presentation
86
Measure Stained Cells Mask
Original tissue
2012
Deconvoluted nuclear counterstain
Tissue IA 2.0 Training Presentation
Deconvoluted nuclear marker
87
Deconvoluted membrane marker
Detected Membrane
2012
Nuclear marker below set cutoff
Accepted nuclei and cell borders Membrane staining
Tissue IA 2.0 Training Presentation
Membrane marker below set cutoff
Co-localized Nuclear and
88
Testing Preference
2012
Once the algorithm preferences have been set, test on a variety of areas across a number of slides Test on areas with high and low intensity staining Once satisfied, lock the color definition file and the settings in the algorithm preferences The algorithm can now be run in high throughput
Tissue IA 2.0 Training Presentation
89
Demonstration: Measure Stained Cells
2012
Tissue IA 2.0 Training Presentation
90
Running Preference Files in High Throughput Analysis
2012
Tissue IA 2.0 Training Presentation
91
Image Analysis in Digital Image Hub
2012
Use Digital Image Hub to run image analysis on whole slides, TMAs and annotations Create custom jobs with exclusion regions, and merge results From the browse tab, select the images to be analyzed Send images to Tissue IA workflow
Tissue IA 2.0 Training Presentation
92
Select Algorithm and Preferences
2012
Select the algorithm to run, the desired preferences Select the areas to be analyzed Whole slide Annotations OpTMA cores (on provision of appropriate licence) Custom job
Tissue IA 2.0 Training Presentation
93
Select Algorithm and Preferences
2012
Select merged results Select analysis magnification Enter job name and description
Tissue IA 2.0 Training Presentation
94
Analysis Magnification Warning! Tissue IA has been designed to run at 20x magnification Positive pixel algorithm can be run at lower magnifications If the slide has been scanned at 40x magnification, please ensure that the slide is run at 20x
Note: very small annotations may not run
2012
Tissue IA 2.0 Training Presentation
95
Custom Job
2012
Select areas for analysis Exclude regions from analysis
Tissue IA 2.0 Training Presentation
96
Tissue IA Tab
2012
After submitting a job to IA, you are brought to the Tissue IA tab The Tissue IA tab consists of 5 subtabs Dashboard Completed jobs Queued jobs Failed jobs Custom jobs
Tissue IA 2.0 Training Presentation
97
Dashboard
2012
Overview of IA and job status
Tissue IA 2.0 Training Presentation
98
Completed Jobs Tab
2012
View list of completed jobs View results, download data, re-run, view folder, delete job Note: Job cannot be re-run if an image from the job has been deleted View the preferences and color definition file used
Tissue IA 2.0 Training Presentation
99
Results
2012
Image Analysis Results are stored as metadata View the results of the image analysis by: Info icon under completed jobs Info icon under browse tab Mousing over the annotations (most recent run results displayed) Click on the Excel icon to export the Image Analysis results into Excel On completed jobs page Under slide information View IA job history under slide information
Tissue IA 2.0 Training Presentation
100
Queued jobs
2012
Jobs submitted to Image Analysis are queued View queued jobs
Tissue IA 2.0 Training Presentation
101
Failed Jobs Tab
2012
Any jobs that failed are listed Can re-queue or re-create the IA job
Tissue IA 2.0 Training Presentation
102
Custom Jobs Tab
2012
All custom jobs in progress are listed here Work on jobs and then submit
Tissue IA 2.0 Training Presentation
103
Sending Annotations for High Throughput Analysis
2012
Tissue Tissue IA IA 2.02.0 Training Training Presentation Presentation © SlidePath 2012
104
High Throughput Analysis on Annotations
2012
Following drawing an annotation, right click over the annotations and select ‘analyse annotation’
Tissue IA 2.0 Training Presentation
105
High Throughput Analysis on Annotations
2012
Select the algorithm, algorithm preference and magnification which the algorithm should be run at Job Id will be returned for easy tracking of the job under completed jobs tab
Tissue IA 2.0 Training Presentation
106
Demonstration: High Throughput Analysis
2012
Tissue IA 2.0 Training Presentation
107
Important Information
2012
Tissue IA 2.0 Training Presentation
108
Multifocal Images
2012
For any image scanned in multiple planesimage analysis will be run on the middle plane
Tissue IA 2.0 Training Presentation
109
Freehand Annotations
2012
When drawing freehand annotations for analysis, please ensure that the annotation is sealed using the “seal annotation” tool Note: The annotation will not be included in analysis if not sealed
Tissue IA 2.0 Training Presentation
110
Out of Focus Images
2012
Images that have been scanned and are out of focus will affect the quality of Image Analysis results
Tissue IA 2.0 Training Presentation
111
Magnification and Annotations
2012
High throughput Image Analysis in Digital Image Hub can be run at different magnifications Warning: If annotations are small and run at 4x or 10x, they may fail to process. As a general rule, analysis should be run at a magnification equal or greater than the magnification the annotation was drawn at, to ensure that the analysis will complete successfully
Tissue IA 2.0 Training Presentation
112
Optional Algorithms
2012
Tissue IA 2.0 Training Presentation
113
Creating Preference for Measure Stained Area Fluorescence Algorithm (Optional)
2012
Tissue IA 2.0 Training Presentation
114
Measure Stained Area Fluorescence
2012
Analyses the number of positive pixels in a fluorescence images This algorithm is only applicable to supported fluorescence image formats in .scn This algorithm can be run on any magnification.
Tissue IA 2.0 Training Presentation
115
Select color Channel
2012
Select up to three color channels
Tissue IA 2.0 Training Presentation
116
Input Parameters
Intensity Threshold Channel A/B/C Adjust depending on the number of channels selected Range between 0-255 (0=Black, 255= White) Set the minimum and maximum intensity
thresholds for each of the input channels
2012
Define the measurement units 0 = µm, 1= mm, 2 = pixels Set default calibration setting Range between 0-10 Use when images don't have calibration information (i.e Jpeg) Use lock to freeze the preferences warning this action cannot be Undone!
Tissue IA 2.0 Training Presentation
117
Results
Measurement units As defined during input settings Positive Area of Channel A/B/C Total area which is positive for each of
the channel defined
Average staining intensity Channel A/B/B The modal value of a greyscale intensity
histogram of the positively stained pixels Used to calculate Staining Concentration
2012
Co-localization of Channel The area of overlap between classes Pearsons correlation coefficient The correlation between two channels
Tissue IA 2.0 Training Presentation
118
Measure Stained Area Fluorescence Masks
Original Image Fluorescence images are changed to a greyscale image Each channel represented as greyscale image
Image Channel Cutoff Image threshold for each channel using the defined input parameters Note: Accepted pixels are denoted by a white mask; black pixels are concerned background and are not included in the analysis
Total Mask Color mask representation of each separated channel and the over lapping between each channel
2012
Tissue IA 2.0 Training Presentation
119
Measure Stained Area Fluorescence Masks
Example of Fluorescence analysis; Channel A = Green spectrum, Channel B = DAPI, Channel C = Red spectrum
Original Tissue
Channel B greyscale image 2012
Channel A greyscale image
Chanel B threshold image
Channel C greyscale image
Tissue IA 2.0 Training Presentation
Chanel A threshold
Chanel C threshold image 120
Tissue Masks
2012
Tissue masks apply primary colours to each class Channel A: Red Channel B: Green Channel C: Blue The co-localization of these channels is a mix of both colours Channel AB: Yellow Channel BC: Magenta Channel AC: Cyan Channel ABC: Grey
Tissue IA 2.0 Training Presentation
121
Demonstration: Measure Stained Area Fluorescence
2012
Tissue IA 2.0 Training Presentation
122
Her2 for research purpose only Algorithm (optional)
2012
Tissue IA 2.0 Training Presentation
123
HER2 Algorithm Annotations
2012
Algorithm does not have contextual understanding of tissue Invasive regions must be annotated for analysis
Tissue IA 2.0 Training Presentation
124
HER2 Algorithm Input
2012
Algorithm preferences and color definition file are not editable
Tissue IA 2.0 Training Presentation
125
Her2
2012
Predicted HER2 score % Confidence in 0/1, 2+ and 3+ score Membrane Staining Absorbance % Membrane Positive Pixels % Continuously Stained Membrane
Tissue IA 2.0 Training Presentation
126
Her2 Results Predicted HER2 score % Confidence in 0/1, 2+ and 3+ score Membrane Staining Absorbance % Membrane Positive Pixels % Continuously Stained Membrane
Unprocessed image of breast tissue that has been immunohistochemically stained with antibodies probing for HER-2 protein expression 2012
Areas detected as positive for continuous membrane staining by image analysis are highlighted in green
Tissue IA 2.0 Training Presentation
127
Demonstration Her2
2012
Tissue IA 2.0 Training Presentation
128
Creating Preference File for Microvessel Detection Algorithm (Optional)
2012
Tissue IA 2.0 Training Presentation
129
Color File
2012
Select the correct color definition file from the drop down menu
Tissue IA 2.0 Training Presentation
130
Tissue Threshold Determine what is background and what is tissue Background Intensity Setting Select a value between 0 (black) and 255 (white) Typically set between 210 – 240
2012
Tissue IA 2.0 Training Presentation
131
Max Vessel Intensity
Set maximum vessel intensity This input allows segmentation of vessels from surrounding tissue Select a value between 0 (black) and 255 (white)
Maximum vessel intensity = 190
Maximum vessel intensity = 150
Yellow (0-1.5); orange (1.5-2), Blue: >2
2012
Tissue IA 2.0 Training Presentation
132
Set Minimum Vessel Size
Eliminate small vessels on the basis of size Range from 0-50,000 pixels Vessels with an area less than the input value will not be included in analysis
Minimum vessel size = 20
Minimum vessel size = 300
Yellow (0-1.5); orange (1.5-2), Blue: >2
2012
Tissue IA 2.0 Training Presentation
133
Set Maximum Vessel Size Eliminate large vessels on the basis of size Settings range from 0 – 50,000 pixels Vessels with an area greater than the input value will not be included in analysis
Maximum vessel size = 2000
Maximum vessel size = 4000
Yellow (0-1.5); orange (1.5-2), Blue: >2
2012
Tissue IA 2.0 Training Presentation
134
Set Maximum Vessel Aspect Ratio
Eliminate vessels on the basis of aspect ratio Settings range from 0 – 3000 Input parameter represents max(length/width)*100
Maximum vessel aspect ratio = 400
Maximum vessel aspect ratio = 1000
Yellow (0-1.5); orange (1.5-2), Blue: >2
2012
Tissue IA 2.0 Training Presentation
135
Separate Merged Vessels
0 = Separate merged vessels 1 = Do not separate
Separate merged vessels = 0
Separate merged vessels = 1
Yellow (0-1.5); orange (1.5-2), Blue: >2
2012
Tissue IA 2.0 Training Presentation
136
Merge Close Vessels
Set strength of merging for close vessel segments Setting range 0-10 This parameter connects close components and smoothens vessel edges Higher value = stronger merging of vessels
Merge close vessels = 1
Merge close vessels = 10
Yellow (0-1.5); orange (1.5-2), Blue: >2
2012
Tissue IA 2.0 Training Presentation
137
Microvessel Detection – Results Total Number of Vessels
Total number of identified vessels, with or without lumen Total Tissue Area in Pixels
Total number of pixels in the tissue Total Vessel Area with Lumen In Pixels:
Total number of pixels identified as vessel in the image, including regions identified as lumen Average Vessel Area In Pixels:
Average number of pixels per vessel in the image, excluding regions identified as lumen
Average Vessel Area with Lumen In Pixels:
Average number of pixels per vessel in the image, including regions identified as lumen 2012
Tissue IA 2.0 Training Presentation
138
Microvessel Detection – Results
Average Vessel Perimeter In Pixels:
Average number of pixels forming perimeter per vessel
Note: This output is not affected by the presence of vessel lumen Microvessel Density – Number of Vessels per Tissue Pixel:
Number of vessels per tissue pixel, including regions identified as lumen
Microvessel Density – Number of Vessels Without Lumen per Tissue Pixel:
Number of vessels per tissue pixel, excluding regions identified as lumen
2012
Tissue IA 2.0 Training Presentation
139
Tissue Masking
Original Tissue
Stained tissue (All tissue that is
Tissue Threshold
stained according to the color definition file)
All vessels not eliminated by size or aspect ratio Ratio : Yellow (0-1.5); orange (1.5-2) Blue: >2 2012
All vessels. Lumen in red (all identified vessels not
eliminated on the basis of size or aspect ratio)
Tissue IA 2.0 Training Presentation
140
Demonstration: Microvessel Detection
2012
Tissue IA 2.0 Training Presentation
141
Optional Modules
2012
Tissue IA 2.0 Training Presentation
142
Integration Toolkit
2012
The Integration Toolkit allows the user to upload customized algorithm, which have been developed for other image process programs
Tissue IA 2.0 Training Presentation
143
www.leica-microsystems.com/products/digital-pathology [email protected] +353 (0)1 8667830
2012
Tissue IA 2.0 Training Presentation
144
Control #: 13MAN1231.A1 Effective Date: 12-Jul-12 ECO #: 3472
2012
Tissue IA 2.0 Training Presentation
1
Intended Use IMPORTANT: SlidePath applications are not cleared by the FDA, Health Canada, or in the EU for diagnostic or clinical use. All applications are intended solely for use in the research or educational setting, such as university or pharmaceutical development. These applications are described as Research Applications or Research Use Only.
2012
Tissue IA 2.0 Training Presentation
2
What is Tissue Image Analysis
2012
High-throughput web enabled image analysis solution for digital slides Control and review image analysis online anywhere using web browser Analyze whole slides, areas of interest or TMAs Queue multiple images for batch processing Automatically integrate data with existing TMA/Research data in Distiller and/or export to Excel spreadsheets
Tissue IA 2.0 Training Presentation
3
Introduction to Tissue Image Analysis
2012
Tissue IA 2.0 Training Presentation
4
OverView of Tissue IA
2012
Set preferences in Digital Image Hub Determine the color definition file What is positively stained Use pre defined deconvolution colour definition files Determine the algorithm preferences Select the preferences for the biomarker of interest Test the image analysis in a variety of fields and slides
Tissue IA 2.0 Training Presentation
5
OverView of TIA
2012
Run image analysis in Digital Image Hub Run on whole slides, TMAs (with appropriate licence) annotations Results are added as slide metadata Run high throughput image analysis in Distiller (with an applicable licence) Run on TMAs Results are automatically entered into records Each user will only have access to the preferences and color definition files they created
Tissue IA 2.0 Training Presentation
6
Optimizer Harness
2012
Tissue IA 2.0 Training Presentation
7
Optimizing Harness
Use the Harness to optimize the color definition files and algorithm setting before sending to high throughput analysis
Color Definition file
Algorithm settings
Preference File
High through put analysis 2012
Tissue IA 2.0 Training Presentation
8
Image Analysis Harness Optimizer
2012
In Digital Image Hub, select a slide In the slide viewer, zoom to 20x magnification and select 'Image Analysis Harness Optimizer'
Tissue IA 2.0 Training Presentation
9
How To Set Color Definition Files
2012
Tissue IA 2.0 Training Presentation
10
Color Definition File
2012
Determine what is positively stained Select 'Manage Color Definition files'
Tissue IA 2.0 Training Presentation
11
Color Definition Files
2012
A list of color definition files will appear Create, rename, copy, edit or delete files Preview results of color definition files Note: List of preferences is user specific, i.e. each user will only have access to their own preferences
Tissue IA 2.0 Training Presentation
12
PreLoaded Color Definition Files
2012
Default DAB Deconvolution processing allows for a More advanced method of color separation Preloaded Deconvolution Files: Deconvolution – Haematoxylin Deconvolution – AEC Deconvolution – DAB Deconvolution – Eosin Deconvolution – Fast Blue Deconvolution – Fast Red Deconvolution – Methyl Green
Tissue IA 2.0 Training Presentation
13
Create a Color Definition File
2012
Select pixels to include in the color definition file Green = pixels already included Select other colors to indicate positive pixels Use up / down arrow to increase / decrease number of pixels included
Tissue IA 2.0 Training Presentation
14
Inverted Screen
Use the 'T' key to toggle to the inverted screen Select pixels to be subtracted from the color definition file Red = pixels not included
color definition mask
2012
Inverted Screen mask
Tissue IA 2.0 Training Presentation
15
Test Color Definition Files
2012
Objective: record the full representation of positive color, regardless of location Test the color definition file in a variety of areas across a number of slides Test on areas of high and low intensity staining
Tissue IA 2.0 Training Presentation
16
Demonstration: Color Definition Files
2012
Tissue IA 2.0 Training Presentation
17
How To Set Algorithm Preferences
2012
Tissue IA 2.0 Training Presentation
18
Select the Algorithms There are two standard immunohistochemical algorithms Measure Stained Area Measure Stained Cells
2012
Tissue IA 2.0 Training Presentation
19
Select Algorithm Preference
2012
A list of preferences will be displayed Cannot create new preferences from scratch, must copy an existing default preference Test, rename, edit and delete Note: List of preferences is user specific, i.e. each user will only have access to their own preferences
Tissue IA 2.0 Training Presentation
20
Running the Algorithm on a Field of View or Area
2012
Tissue IA 2.0 Training Presentation
21
Field of View or Area
2012
Run the algorithm on the field of view (area which is visible in the viewer at 20x) Run algorithm on a selected area by selecting the square
Tissue IA 2.0 Training Presentation
22
Creating Preference File for Measure Stained Area Algorithm
2012
Tissue IA 2.0 Training Presentation
23
Select the Color Definition File
2012
Up to three classes/ color definition files can be defined Analysis will be carried out on all three classes simultaneously Choose either a pre loaded color definition file or one that has been created/edited by the user Use lock to freeze the preferences warning this action cannot be undone!
Tissue IA 2.0 Training Presentation
24
Input Parameters Goal: Adjust the input parameters to optimize the Algorithm for the biomarker of interest Changing the input parameters will affect the over all results
2012
Tissue IA 2.0 Training Presentation
25
Input Parameters Measurement Units Define the measurement units 0 = µm, 1= mm, 2 = pixels Calibration setting Set default calibration setting Range between 0-10 Use if there is no magnification information attached to the image (i.e. Jpeg images ) Deconvolution setting Select color deconvolution Enable = 1, select if deconvoluted color have been selected
2012
Tissue IA 2.0 Training Presentation
26
Tissue Threshold Intensity
Determine what is background and what is tissue Background Intensity Setting Select a value between 0 (black) and 255 (white) Typically set between 210 – 240
Example of original tissue
2012
Threshold set to 50
Threshold set to 210
Tissue IA 2.0 Training Presentation
Threshold set to 250
27
Deconvolution Threshold Setting Determine the intensity threshold for each of the color Classes defined Maximum intensity of a pixel in the Class to be considered positive Select a value between 0 (black) and 255 (white)
Example of original tissue
2012
Example of threshold setting 50
Tissue IA 2.0 Training Presentation
Example of threshold setting 200
28
Measure Stained Area Results
2012
Run the positive pixel algorithm on a field of view or area Back to refine preferences Reanalyse Store results will create an annotation with the results of the image analysis
Tissue IA 2.0 Training Presentation
29
Measure Stained Area Results
Measurement units
As defined during input settings
Total tissue area
Total area of the tissue which has been identified
Positive Area of Class A/B/C
Total area which is positive for each of the classes defined
2012
Tissue IA 2.0 Training Presentation
30
Measure Stained Area Results
Average staining intensity class A/B/C The modal value of a greyscale intensity histogram of the positively stained
pixels Used to calculate Staining Concentration
2012
Staining concentration Class A/B/C Measure of the concentration of the stain within the tissue Co-localization of Classes The area of overlap between classes Pearsons correlation coefficient The correlation between two classes
Tissue IA 2.0 Training Presentation
31
Tissue masks
Tissue mask represents the deconvoluted image for each class defined, threshold cut off for each class defined and total mask
Example of masks images where two color classes were applied. Class A: DeconvolutionHaematoxylin, Class B: Deconvolution -DAB
Original tissue
Deconvoluted DAB mask
Tissue threshold mask
Deconvoluted Haematoxylin mask Haematoxylin intensity threshold
DAB intensity threshold
Class 2012
Tissue IA 2.0 Training Presentation
Total mask: Haematoxylin : Red, DAB: Green, AB: Yellow 32
Tissue Masks
2012
Tissue masks apply primary colours to each class Class A: Red Class B: Green Class C: Blue The co-localization of these channels is a mix of both colours Class AB: Yellow Class BC: Magenta Class AC: Cyan Class ABC: Grey
Tissue IA 2.0 Training Presentation
33
Demonstration: Algorithm Preferences
2012
Tissue IA 2.0 Training Presentation
34
Measure Stained Cells Algorithm
2012
Tissue IA 2.0 Training Presentation
35
Measure Stained Cells
2012
Measure stained cells replacing the Nuclear, membrane and intercellular algorithms With the use of new input parameters including Nuclear Heterogeneity detection, TIA 2.0 has more advanced nuclear detection
Tissue IA 2.0 Training Presentation
36
Measure Stained Cells
Measure stained cells replacing the Nuclear, membrane and intercellular algorithms Analyse nuclear, cytoplasm or membrane staining separately or a combination of any two Up to three of the following (as long as two of them use the same color definition file) nuclear, cytoplasm or membrane marker
Set input parameters for each of the biomarkers
Nuclear Counterstain
Results output: Separate results for each of the biomarkers selected. For two biomarkers are defined, combination and co-localization results will be output
2012
Tissue IA 2.0 Training Presentation
37
Color Definition File
Select a color definition file for nuclear counterstain
AND
2012
Of the following: nuclear marker, cytoplasmic marker, membrane marker If all cellular compartments are selected, ensure that two of the compartments have the same colour definition file assigned
Tissue IA 2.0 Training Presentation
38
Input Parameters Measurement units Define the measurement units
0 = µm, 1= mm, 2 = pixels Tissue Threshold Determine what is background and what
is tissue Default calibration Set default calibration setting Range between 0-10
(use if there is no magnification provided i.e. Jpeg images)
2012
Tissue IA 2.0 Training Presentation
39
Nuclei Heterogeneity
How similar the nuclei are in the tissue Range between 0 – 4, 0 = nuclei are similar, >1 = increasing diversity in the tissue from darkest to lightest
0
2012
Increased diversity from darkest to lightest
Tissue IA 2.0 Training Presentation
4
40
Nuclei Heterogeneity Original tissue
2012
Nuclei Heterogeneity = 1
Tissue IA 2.0 Training Presentation
Nuclei Heterogeneity = 4
41
Strength of the Nuclear Counter Stain
Define if the counter stain is strong or weak Range between 0-2, 0= strong, 2 = weak
High Contrast
2012
Low intensity
Tissue IA 2.0 Training Presentation
42
Strength of the Nuclear Counter Stain Use 0 for high Contrast:
Use 2 for low contrast:
2012
Tissue IA 2.0 Training Presentation
43
Nuclear Window
Define the size of the window around the nucleus by setting its radius Values in units defined in measurement input , i.e. if measurement units are set to pixels, then this parameter is in pixels Affects nuclei per window and nuclear area per window settings SlidePath recommends this setting be left at 37 µm to start
radius
2012
Tissue IA 2.0 Training Presentation
44
Nuclear Area
Exclusion parameter Range between 0 -10,000 units squared as defined in measurement units
50µm2
150µm2
50µm2
150µm2
Area:: 100µm2
Area: 300µm2
2012
Nuclei of interest included The slider is set to 50µm and 150µm, nuclei of interest included as its area is 100µm2
Nuclei of interest excluded The slider is set to 50µm and 150µm, nuclei of interest excluded as its area is 300µm2
Tissue IA 2.0 Training Presentation
45
Nuclear Area Nuclear area set between 0- 1000
2012
Nuclear area set between 0- 10
Tissue IA 2.0 Training Presentation
46
Nuclei per window
2012
Exclusion parameter A “window” = as defined in the input setting Range: 0 – 1000 cells per window Cell will be counted if its centre point lies within the window
Tissue IA 2.0 Training Presentation
47
Nuclei Per window 5
20
Nuclei of interest excluded The slider is set to 5 and there are only 3 surround nuclei in the window (green)
5
20
Nuclei of interest included The slider is set to 20 and there are 10 surround nuclei in the window (green)
2012
Tissue IA 2.0 Training Presentation
48
Nuclei Per window Nuclei per window set between 0-100
2012
Nuclei per window set between 60-100
Tissue IA 2.0 Training Presentation
49
49
Nuclear Area Per Window
2012
Exclusion parameter Based on percentage of area within the window that is taken up by nuclei Range: 0 – 100% Nucleus is counted if its centre point lies within the window
Tissue IA 2.0 Training Presentation
50
Nuclear Area Per Window Nuclei of interest excluded Nuclei of interested excluded, as area taken up by surrounding nuclei is 10%
Nuclei of interest excluded Nuclei of interested excluded, as area taken up by surrounding nuclei is 80%
Nuclei of interest included Nuclei of interested included, as area taken up by surrounding nuclei is 70%
Nuclei of interest included Nuclei of interested included, as area taken up by surrounding nuclei is 70% 2012
Tissue IA 2.0 Training Presentation
51
Nuclear Area per Window % Nuclear per window set between 0-100%
2012
% Nuclear per window set between 10-50%
Tissue IA 2.0 Training Presentation
52
Cell Area
Eliminate cells on the basis of size Range between 0 – 10,000 (measurements units as defined during input parameters)
50µm2 150µm2
Cell of interest included The slider is set to 50µm and 150µm, cell of interest included as its area is 100µm2
Area: 100µm2
50µm2
Area::
2012
300µm2
150µm2
Cell of interest excluded The slider is set to 50µm and 150µm, cell of interest excluded as its area is 300µm2
Tissue IA 2.0 Training Presentation
53
Cell Area
Example of original tissue
2012
Threshold set from 0 -500
Tissue IA 2.0 Training Presentation
Threshold set between 0-50
54
Maximum Cell Radius
The maximum cell radius which should be included for analysis Range between 0 – 1000 (measurement as defined in ‘Measurement units’ input
parameters)
2012
Cell radius mainly influences how the cells are modeled (i.e. how much the cell boundary expands during the cell prediction)
Tissue IA 2.0 Training Presentation
55
Nuclear Staining Intensity Cutoff
All pixels with intensity higher than defined in deconvoluted nuclear staining image are considered negative Range 0-255
0
2012
255
Tissue IA 2.0 Training Presentation
56
Nuclear Staining Intensity Cutoff Nuclear staining intensity cut off set at 210
2012
Nuclear staining intensity cut off set at 150
Tissue IA 2.0 Training Presentation
57
% of Stained Pixels in a Nucleus Cutoff setting
Categorization parameter Range: 0 – 100% Nuclei with a lower percentage of positive staining will be categorized as negative
0
2012
100%
Tissue IA 2.0 Training Presentation
58
58
% of Stained Pixels in a Nucleus Cutoff setting Nuclear stained area cut off set 80%
2012
Nuclear stained area cut off set 10%
Tissue IA 2.0 Training Presentation
59
Strong/Moderate/Weak Nuclear Staining Intensity Cutoff
2012
Categorization parameter Range: 0 – 255 0 represents minimum intensity (black), 255 represents maximum intensity (white), in a greyscale intensity range
Tissue IA 2.0 Training Presentation
60
Strong/Moderate/Weak Nuclear Staining Intensity Cutoff
2012
Strong/Moderate Staining Intensity Cutoff Setting Nuclei with a level below this input are categorized as strong Moderate/Weak Staining Intensity Cutoff Setting Nuclei with a value above this input are categorized as weak Nuclei with values between are categorized as moderate
Tissue IA 2.0 Training Presentation
61
Strong/Moderate/Weak Nuclear Staining Intensity Cutoff Staining intensity low threshold set to 96, high threshold set to 196
2012
Staining intensity low threshold set to 163, high threshold set to 196
Tissue IA 2.0 Training Presentation
62
Nuclear Staining Filters
2012
Define if all, only positive, or only negative, nuclear staining should be included in results and output image Parameter works on cells not pixels Range 0-2, 0 representing all nuclei, 1 representing only positive nuclei, 2 representing only negative nuclei
Tissue IA 2.0 Training Presentation
63
Nuclear Staining Filter Include all cells and nuclei
2012
Include only positive cells
Tissue IA 2.0 Training Presentation
Only include negative cells
64
Cytoplasmic Input Parameters
Cytoplasmic parameters will only have an effect on Cytoplasmic input parameters if cytoplasmic stain is selected
Cytoplasmic input parameters Cytoplasmic Staining Intensity Cut-off % of Cytoplasmic Staining Area in a Cell Cut-off Strong/Moderate/Weak Cytoplasmic Staining Intensity Cut-off Cytoplasmic staining filter
2012
Tissue IA 2.0 Training Presentation
65
Cytoplasmic Staining Intensity Cutoff
All pixels with intensity higher than defined in deconvoluted cytoplasmic staining image are considered negative Range 0-255
0
2012
255
Tissue IA 2.0 Training Presentation
66
Cytoplasmic Staining Intensity Cutoff Original image
2012
Cytoplasmic stained intensity cut off 30
Tissue IA 2.0 Training Presentation
Cytoplasmic stained intensity cut off set 220
67
% of Cytoplasmic Staining Area in a Cell Cut-off
Categorization parameter Range: 0 – 100% % of Stained Pixels in cytoplasm Cutoff setting Cytoplasm with a lower percentage of positive staining will be categorized as negative
0
2012
100%
Tissue IA 2.0 Training Presentation
68
% of Cytoplasmic Staining Area in a Cell Cut-off Original tissue
2012
Cytoplasmic stained area cut off set to 75%
Tissue IA 2.0 Training Presentation
Cytoplasmic stained area cut off set to 95%
69
Strong/Moderate/Weak Cytoplasm Staining Intensity Cutoff
2012
Strong/Moderate Staining Intensity Cutoff Setting Cytoplasm with a level below this input are categorized as strong Moderate/Weak Staining Intensity Cutoff Setting Cytoplasm with a value above this input are categorized as weak Cytoplasm with values between are categorized as moderate
Tissue IA 2.0 Training Presentation
70
Strong/Moderate/Weak Cytoplasm Staining Intensity Cutoff Original tissue
2012
Strong/Moderate/Weak cytoplasm staining intensity cutoff set between 100-150 (From 0-100: strong, 100-150: moderate, 150-255: weak)
Tissue IA 2.0 Training Presentation
Strong/Moderate/Weak cytoplasm staining intensity cutoff set between 100-150 (From 0-160: strong, 160-220: moderate, 220-255: weak)
71
Cytoplasm Staining Filters
2012
Define if all, only positive, or only negative, Cytoplasm staining should be included in results and output image Range 0-2, 0 representing all cells, 1 representing only positive cells, 2 representing only negative cells
Tissue IA 2.0 Training Presentation
72
Membrane Input Parameters
Membrane parameters will only have an effect on membrane input parameters if cytoplasmic stain is selected
Membrane input parameters Membrane Staining Intensity Cut-off Percentage of Membrane Staining Area in a Cell Cut-off Strong/Moderate/Weak Membrane Staining Intensity Cut-off Membrane staining filter
2012
Tissue IA 2.0 Training Presentation
73
Membrane Staining Intensity Cutoff
All pixels with intensity higher than defined in deconvoluted membrane staining image are considered positive Range 0-255
0
2012
255
Tissue IA 2.0 Training Presentation
74
Membrane Staining Intensity Cutoff Original Tissue
2012
Membrane staining intensity cutoff set to 75
Tissue IA 2.0 Training Presentation
Membrane staining intensity cutoff set to 220
75
% of Stained Pixels in a Membrane Cutoff setting
Categorization parameter Range: 0 – 100% % of Stained Pixels in a membrane Cutoff setting Membrane with a lower percentage of positive staining will be categorized as negative
0
2012
100%
Tissue IA 2.0 Training Presentation
76
% of Stained Pixels in a membrane Cutoff setting Original tissue
2012
Membrane staining intensity cuttoff set to 75%
Tissue IA 2.0 Training Presentation
Membrane staining intensity cuttoff set to 95%
77
Strong/Moderate/Weak Membrane Staining Intensity Cutoff
2012
Strong/Moderate Staining Intensity Cutoff Setting Membrane with a level below this input are categorized as strong Moderate/Weak Staining Intensity Cutoff Setting Membrane with a value above this input are categorized as weak Membrane with values between are categorized as moderate
Tissue IA 2.0 Training Presentation
78
Strong/Moderate/Weak Membrane Staining Intensity Cutoff Original tissue
Strong/Moderate/Weak membrane staining intensity cutoff set between 100-150 (From 0-100: strong, 100-150: moderate, 150-255: weak)
Strong/Moderate/Weak membrane staining intensity cutoff set cutoff set between 160-220 (From 0-160: strong, 160-220: moderate, 220-255: weak)
Strong Membrane: Violet, Moderate membrane: Pink, Weak Membrane: Rose 2012
Tissue IA 2.0 Training Presentation
79
Membrane Staining Filters
2012
Define if all, only positive, or only negative, Membrane staining should be included in results and output image Range 0-2, 0 representing all nuclei, 1 representing only positive cells, 2 representing only negative cells
Tissue IA 2.0 Training Presentation
80
Membrane Filters All membrane include (setting= 0)
2012
Only included positive membrane setting = 1
Tissue IA 2.0 Training Presentation
Only included negative membrane setting = 2
81
Results
Measurement units
As defined during input settings
Total tissue area
Total area of the tissue which has been identified
Total number of cells
The total number of cells that have been identified in the tissue
2012
Tissue IA 2.0 Training Presentation
82
Results
Histoscore
Histoscore of the accepted nuclei
Total Accepted Nuclei
All nuclei that were not eliminated by exclusion criteria
Number and percentage of negative, weak, moderate, strong intensity nuclei
The total number nuclei Categorized as having negative, weak, moderate, strong intensity by value and percentage
2012
Tissue IA 2.0 Training Presentation
83
Results
Average nuclear staining intensity The modal value of a greyscale intensity histogram of the positively stained
pixels Used to calculate Staining Concentration
2012
Average nuclear Staining absorbance Measure of the absorbance of the stain within the tissue Percentage nuclear area in Tissue The percentage of nuclear area
Tissue IA 2.0 Training Presentation
84
Cytoplasmic Marker Results
Cellular H-Score for cytoplasmic staining Number and % of cells with negative cytoplasmic staining Number and % of cells with positive cytoplasmic staining Number and % of weak, moderate and strong intensity stained cytoplasm cells Average cytoplasmic staining intensity – of positive area only Average cytoplasmic staining absorbance - of positive area only % positive cytoplasmic area in tissue
Note: Only visible if cytoplasmic marker is selected
2012
Tissue IA 2.0 Training Presentation
85
Membrane Marker Results
Cellular H-Score for membrane staining - Number and % of cells with negative membrane staining Number and % of cells with positive membrane staining Number and % of weak, moderate and strong intensity stained membrane cells Average membrane staining intensity – of positive area only Average membrane staining absorbance – of positive area only % positive membrane area in tissue
Note: Only visible if membrane marker is selected
2012
Tissue IA 2.0 Training Presentation
86
Measure Stained Cells Mask
Original tissue
2012
Deconvoluted nuclear counterstain
Tissue IA 2.0 Training Presentation
Deconvoluted nuclear marker
87
Deconvoluted membrane marker
Detected Membrane
2012
Nuclear marker below set cutoff
Accepted nuclei and cell borders Membrane staining
Tissue IA 2.0 Training Presentation
Membrane marker below set cutoff
Co-localized Nuclear and
88
Testing Preference
2012
Once the algorithm preferences have been set, test on a variety of areas across a number of slides Test on areas with high and low intensity staining Once satisfied, lock the color definition file and the settings in the algorithm preferences The algorithm can now be run in high throughput
Tissue IA 2.0 Training Presentation
89
Demonstration: Measure Stained Cells
2012
Tissue IA 2.0 Training Presentation
90
Running Preference Files in High Throughput Analysis
2012
Tissue IA 2.0 Training Presentation
91
Image Analysis in Digital Image Hub
2012
Use Digital Image Hub to run image analysis on whole slides, TMAs and annotations Create custom jobs with exclusion regions, and merge results From the browse tab, select the images to be analyzed Send images to Tissue IA workflow
Tissue IA 2.0 Training Presentation
92
Select Algorithm and Preferences
2012
Select the algorithm to run, the desired preferences Select the areas to be analyzed Whole slide Annotations OpTMA cores (on provision of appropriate licence) Custom job
Tissue IA 2.0 Training Presentation
93
Select Algorithm and Preferences
2012
Select merged results Select analysis magnification Enter job name and description
Tissue IA 2.0 Training Presentation
94
Analysis Magnification Warning! Tissue IA has been designed to run at 20x magnification Positive pixel algorithm can be run at lower magnifications If the slide has been scanned at 40x magnification, please ensure that the slide is run at 20x
Note: very small annotations may not run
2012
Tissue IA 2.0 Training Presentation
95
Custom Job
2012
Select areas for analysis Exclude regions from analysis
Tissue IA 2.0 Training Presentation
96
Tissue IA Tab
2012
After submitting a job to IA, you are brought to the Tissue IA tab The Tissue IA tab consists of 5 subtabs Dashboard Completed jobs Queued jobs Failed jobs Custom jobs
Tissue IA 2.0 Training Presentation
97
Dashboard
2012
Overview of IA and job status
Tissue IA 2.0 Training Presentation
98
Completed Jobs Tab
2012
View list of completed jobs View results, download data, re-run, view folder, delete job Note: Job cannot be re-run if an image from the job has been deleted View the preferences and color definition file used
Tissue IA 2.0 Training Presentation
99
Results
2012
Image Analysis Results are stored as metadata View the results of the image analysis by: Info icon under completed jobs Info icon under browse tab Mousing over the annotations (most recent run results displayed) Click on the Excel icon to export the Image Analysis results into Excel On completed jobs page Under slide information View IA job history under slide information
Tissue IA 2.0 Training Presentation
100
Queued jobs
2012
Jobs submitted to Image Analysis are queued View queued jobs
Tissue IA 2.0 Training Presentation
101
Failed Jobs Tab
2012
Any jobs that failed are listed Can re-queue or re-create the IA job
Tissue IA 2.0 Training Presentation
102
Custom Jobs Tab
2012
All custom jobs in progress are listed here Work on jobs and then submit
Tissue IA 2.0 Training Presentation
103
Sending Annotations for High Throughput Analysis
2012
Tissue Tissue IA IA 2.02.0 Training Training Presentation Presentation © SlidePath 2012
104
High Throughput Analysis on Annotations
2012
Following drawing an annotation, right click over the annotations and select ‘analyse annotation’
Tissue IA 2.0 Training Presentation
105
High Throughput Analysis on Annotations
2012
Select the algorithm, algorithm preference and magnification which the algorithm should be run at Job Id will be returned for easy tracking of the job under completed jobs tab
Tissue IA 2.0 Training Presentation
106
Demonstration: High Throughput Analysis
2012
Tissue IA 2.0 Training Presentation
107
Important Information
2012
Tissue IA 2.0 Training Presentation
108
Multifocal Images
2012
For any image scanned in multiple planesimage analysis will be run on the middle plane
Tissue IA 2.0 Training Presentation
109
Freehand Annotations
2012
When drawing freehand annotations for analysis, please ensure that the annotation is sealed using the “seal annotation” tool Note: The annotation will not be included in analysis if not sealed
Tissue IA 2.0 Training Presentation
110
Out of Focus Images
2012
Images that have been scanned and are out of focus will affect the quality of Image Analysis results
Tissue IA 2.0 Training Presentation
111
Magnification and Annotations
2012
High throughput Image Analysis in Digital Image Hub can be run at different magnifications Warning: If annotations are small and run at 4x or 10x, they may fail to process. As a general rule, analysis should be run at a magnification equal or greater than the magnification the annotation was drawn at, to ensure that the analysis will complete successfully
Tissue IA 2.0 Training Presentation
112
Optional Algorithms
2012
Tissue IA 2.0 Training Presentation
113
Creating Preference for Measure Stained Area Fluorescence Algorithm (Optional)
2012
Tissue IA 2.0 Training Presentation
114
Measure Stained Area Fluorescence
2012
Analyses the number of positive pixels in a fluorescence images This algorithm is only applicable to supported fluorescence image formats in .scn This algorithm can be run on any magnification.
Tissue IA 2.0 Training Presentation
115
Select color Channel
2012
Select up to three color channels
Tissue IA 2.0 Training Presentation
116
Input Parameters
Intensity Threshold Channel A/B/C Adjust depending on the number of channels selected Range between 0-255 (0=Black, 255= White) Set the minimum and maximum intensity
thresholds for each of the input channels
2012
Define the measurement units 0 = µm, 1= mm, 2 = pixels Set default calibration setting Range between 0-10 Use when images don't have calibration information (i.e Jpeg) Use lock to freeze the preferences warning this action cannot be Undone!
Tissue IA 2.0 Training Presentation
117
Results
Measurement units As defined during input settings Positive Area of Channel A/B/C Total area which is positive for each of
the channel defined
Average staining intensity Channel A/B/B The modal value of a greyscale intensity
histogram of the positively stained pixels Used to calculate Staining Concentration
2012
Co-localization of Channel The area of overlap between classes Pearsons correlation coefficient The correlation between two channels
Tissue IA 2.0 Training Presentation
118
Measure Stained Area Fluorescence Masks
Original Image Fluorescence images are changed to a greyscale image Each channel represented as greyscale image
Image Channel Cutoff Image threshold for each channel using the defined input parameters Note: Accepted pixels are denoted by a white mask; black pixels are concerned background and are not included in the analysis
Total Mask Color mask representation of each separated channel and the over lapping between each channel
2012
Tissue IA 2.0 Training Presentation
119
Measure Stained Area Fluorescence Masks
Example of Fluorescence analysis; Channel A = Green spectrum, Channel B = DAPI, Channel C = Red spectrum
Original Tissue
Channel B greyscale image 2012
Channel A greyscale image
Chanel B threshold image
Channel C greyscale image
Tissue IA 2.0 Training Presentation
Chanel A threshold
Chanel C threshold image 120
Tissue Masks
2012
Tissue masks apply primary colours to each class Channel A: Red Channel B: Green Channel C: Blue The co-localization of these channels is a mix of both colours Channel AB: Yellow Channel BC: Magenta Channel AC: Cyan Channel ABC: Grey
Tissue IA 2.0 Training Presentation
121
Demonstration: Measure Stained Area Fluorescence
2012
Tissue IA 2.0 Training Presentation
122
Her2 for research purpose only Algorithm (optional)
2012
Tissue IA 2.0 Training Presentation
123
HER2 Algorithm Annotations
2012
Algorithm does not have contextual understanding of tissue Invasive regions must be annotated for analysis
Tissue IA 2.0 Training Presentation
124
HER2 Algorithm Input
2012
Algorithm preferences and color definition file are not editable
Tissue IA 2.0 Training Presentation
125
Her2
2012
Predicted HER2 score % Confidence in 0/1, 2+ and 3+ score Membrane Staining Absorbance % Membrane Positive Pixels % Continuously Stained Membrane
Tissue IA 2.0 Training Presentation
126
Her2 Results Predicted HER2 score % Confidence in 0/1, 2+ and 3+ score Membrane Staining Absorbance % Membrane Positive Pixels % Continuously Stained Membrane
Unprocessed image of breast tissue that has been immunohistochemically stained with antibodies probing for HER-2 protein expression 2012
Areas detected as positive for continuous membrane staining by image analysis are highlighted in green
Tissue IA 2.0 Training Presentation
127
Demonstration Her2
2012
Tissue IA 2.0 Training Presentation
128
Creating Preference File for Microvessel Detection Algorithm (Optional)
2012
Tissue IA 2.0 Training Presentation
129
Color File
2012
Select the correct color definition file from the drop down menu
Tissue IA 2.0 Training Presentation
130
Tissue Threshold Determine what is background and what is tissue Background Intensity Setting Select a value between 0 (black) and 255 (white) Typically set between 210 – 240
2012
Tissue IA 2.0 Training Presentation
131
Max Vessel Intensity
Set maximum vessel intensity This input allows segmentation of vessels from surrounding tissue Select a value between 0 (black) and 255 (white)
Maximum vessel intensity = 190
Maximum vessel intensity = 150
Yellow (0-1.5); orange (1.5-2), Blue: >2
2012
Tissue IA 2.0 Training Presentation
132
Set Minimum Vessel Size
Eliminate small vessels on the basis of size Range from 0-50,000 pixels Vessels with an area less than the input value will not be included in analysis
Minimum vessel size = 20
Minimum vessel size = 300
Yellow (0-1.5); orange (1.5-2), Blue: >2
2012
Tissue IA 2.0 Training Presentation
133
Set Maximum Vessel Size Eliminate large vessels on the basis of size Settings range from 0 – 50,000 pixels Vessels with an area greater than the input value will not be included in analysis
Maximum vessel size = 2000
Maximum vessel size = 4000
Yellow (0-1.5); orange (1.5-2), Blue: >2
2012
Tissue IA 2.0 Training Presentation
134
Set Maximum Vessel Aspect Ratio
Eliminate vessels on the basis of aspect ratio Settings range from 0 – 3000 Input parameter represents max(length/width)*100
Maximum vessel aspect ratio = 400
Maximum vessel aspect ratio = 1000
Yellow (0-1.5); orange (1.5-2), Blue: >2
2012
Tissue IA 2.0 Training Presentation
135
Separate Merged Vessels
0 = Separate merged vessels 1 = Do not separate
Separate merged vessels = 0
Separate merged vessels = 1
Yellow (0-1.5); orange (1.5-2), Blue: >2
2012
Tissue IA 2.0 Training Presentation
136
Merge Close Vessels
Set strength of merging for close vessel segments Setting range 0-10 This parameter connects close components and smoothens vessel edges Higher value = stronger merging of vessels
Merge close vessels = 1
Merge close vessels = 10
Yellow (0-1.5); orange (1.5-2), Blue: >2
2012
Tissue IA 2.0 Training Presentation
137
Microvessel Detection – Results Total Number of Vessels
Total number of identified vessels, with or without lumen Total Tissue Area in Pixels
Total number of pixels in the tissue Total Vessel Area with Lumen In Pixels:
Total number of pixels identified as vessel in the image, including regions identified as lumen Average Vessel Area In Pixels:
Average number of pixels per vessel in the image, excluding regions identified as lumen
Average Vessel Area with Lumen In Pixels:
Average number of pixels per vessel in the image, including regions identified as lumen 2012
Tissue IA 2.0 Training Presentation
138
Microvessel Detection – Results
Average Vessel Perimeter In Pixels:
Average number of pixels forming perimeter per vessel
Note: This output is not affected by the presence of vessel lumen Microvessel Density – Number of Vessels per Tissue Pixel:
Number of vessels per tissue pixel, including regions identified as lumen
Microvessel Density – Number of Vessels Without Lumen per Tissue Pixel:
Number of vessels per tissue pixel, excluding regions identified as lumen
2012
Tissue IA 2.0 Training Presentation
139
Tissue Masking
Original Tissue
Stained tissue (All tissue that is
Tissue Threshold
stained according to the color definition file)
All vessels not eliminated by size or aspect ratio Ratio : Yellow (0-1.5); orange (1.5-2) Blue: >2 2012
All vessels. Lumen in red (all identified vessels not
eliminated on the basis of size or aspect ratio)
Tissue IA 2.0 Training Presentation
140
Demonstration: Microvessel Detection
2012
Tissue IA 2.0 Training Presentation
141
Optional Modules
2012
Tissue IA 2.0 Training Presentation
142
Integration Toolkit
2012
The Integration Toolkit allows the user to upload customized algorithm, which have been developed for other image process programs
Tissue IA 2.0 Training Presentation
143
www.leica-microsystems.com/products/digital-pathology [email protected] +353 (0)1 8667830
2012
Tissue IA 2.0 Training Presentation
144
