Hyperspectral microscopy imaging to analyze pathology samples with multicolors reduces time and cost

1
Hyperspectral microscopy imaging to analyze
pathology samples with multi-colors reduces time
and cost Michael L. Huebschmana,c, Kevin P.
Rosenblattd, and Harold R. Garnera,b,c
aMcDermott Center for Human Growth and
Development, bDepartments of Biochemisty and
Internal Medicine, cDivision for Transitional
Research UT Southwestern Medical Center, 5323
Harry Hines Blvd., Dallas, TX USA
75390-9185 dSealy Center for Molecular Medicine,
Institute of Translational Sciences, Department
of Biochemistry and Molecular Biology, Department
of Internal Medicine UT Medical Branch, 301
University Blvd., Galveston, TX USA 77555-1071
2
Introduction
Immunofluorescence Panel of Assays
Biomarkers Microtome tissue
sections Reduced Number of Panels Time
Cost Hyperspectral Microscopy Imaging
with 10 colors 2004 Expectations
Demonstration Experiments Analysis
Software for Validation
3
Our Hyperspectral Microscopy Imaging (HMI) system
is composed of a microscope, spectrograph,
motorized xy stage, CCD camera, light sources and
computer.
Excite 290nm 390nm Long Pass Emission gt420nm
4
Xanopath Analyses Programs De-convolves 1 to 10
standard spectra and a background spectrum using
Xanoscope Data Cubes.
Analysis software is written in (ITT Visual
Information Solutions) Interactive Data Language
(IDL) Step 1 - Load a Xanoscope Data Cube Step
2 - Select the number of standard spectra for
fitting. Step 3 Cell analysis
Automated - Blob algorithm
Manual - Mark cell ROI. Step 4 - Tissue
Mark Background ROI. Cell
Mark Background ROI Uses lowest
5 of pixels Step 6 - De-convolute with linear
curve fitting algorithm Step 7 - Save marker
fit coefficients as image layers in a Tiff
file. Step 8 - Calculate Morphology
Information Step 9 - Calculate Average and
Threshold data for each marker Step 10 - Save
Averages, Thresholds, and Morphology data in text
file. Step 11 - Print Q-reports (quantitative
reports) for Data Base Step 12 - Calculate and
save cell expression false color images of each
marker. Step 13 - Display and save image
comparison for Data Base. Step 14 Read all
data and print graphs for quantitative
comparison.
5
GUIs are used to pick data cubes and standard
spectra for fitting
6
14 colors and markers for colon cancer tissue
analysis
7
Normalized emission spectra used for 14
fluorochromes (adjusted for system efficiencies).
Pairs, AF488 and FITC Cy3 and AF555 and Cy5
and AF647 are almost the same.
8
Xanopath Analyses Programs De-convolves 1 to 10
standard spectra and a background spectrum using
Xanoscope Data Cubes.
Analysis software is written in (ITT Visual
Information Solutions) Interactive Data Language
(IDL) Step 1 - Load a Xanoscope Data
Cube Step 2 - Select the number of standard
spectra for fitting. Step 3 Cell analysis
Automated - Blob algorithm
Manual - Mark cell ROI. Step 4 -
Tissue Mark Background Cell
Mark Background ROI Uses
lowest 5 of pixels Step 6 - De-convolute with
linear curve fitting algorithm Step 7 - Save
marker fit coefficients as image layers in a Tiff
file. Step 8 - Cell Calculate Morphology
Information Step 9 - Calculate Average and
Threshold data for each marker Step 10 - Save
Averages, Thresholds, and Morphology data in text
file. Step 11 - Print Q-reports (quantitative
reports) for Data Base Step 12 - Calculate and
save cell expression false color images of each
marker. Step 13 - Display and save image
comparison for Data Base. Step 14 Read all
data and print graphs for quantitative
comparison.
9
Gray scale composite image of a region of a colon
cancer tissue slide stained with 4 color.
10
Xanopath Analyses Programs De-convolves 1 to 10
standard spectra and a background spectrum using
Xanoscope Data Cubes.
Analysis software is written in (ITT Visual
Information Solutions) Interactive Data Language
(IDL) Step 1 - Load a Xanoscope Data
Cube Step 2 - Select the number of standard
spectra for fitting. Step 3 Cell analysis
Automated - Blob algorithm
Manual - Mark cell ROI. Step
4 - Tissue Mark Background ROI
Cell Mark Background ROI
Uses lowest 5 of pixels Step 6 -
De-convolute with linear curve fitting
algorithm Step 7 - Save marker fit
coefficients as image layers in a Tiff file. Step
8 - Cell Calculate Morphology
Information Step 9 - Calculate Average and
Threshold data for each marker Step 10 - Save
Averages, Thresholds, and Morphology data in text
file. Step 11 - Print Q-reports (quantitative
reports) for Data Base Step 12 - Calculate and
save cell expression false color images of each
marker. Step 13 - Display and save image
comparison for Data Base. Step 14 Read all
data and print graphs for quantitative
comparison.
11
Typical de-convolution of the spectrum of one
pixel in a colon cancer tissue slide stained with
7 fluorochromes (AF350, DAPI, AF488, AF430,
AF546, AF594, AF647)
Data Spectrum
Fit Spectrum
Background Component
Marker Components
12
Sony CCD Camera picture mosaic at 40x compared
to the HMI 18 area mosaic scanned at 90x.
0.87mm2 of tissue
13
Xanopath Analyses Programs De-convolves 1 to 10
standard spectra and a background spectrum using
Xanoscope Data Cubes.
Analysis software is written in (ITT Visual
Information Solutions) Interactive Data Language
(IDL) Step 1 - Load a Xanoscope Data
Cube Step 2 - Select the number of standard
spectra for fitting. Step 3 Cell analysis
Automated - Blob algorithm
Manual - Mark cell ROI. Step 4
- Tissue Mark Background ROI
Cell Mark Background ROI
Uses lowest 5 of pixels Step 6 -
De-convolute with linear curve fitting
algorithm Step 7 - Save marker fit
coefficients as image layers in a Tiff file. Step
8 - Cell Calculate Morphology
Information Step 9 - Calculate Average and
Threshold data for each marker Step 10 - Save
Averages, Thresholds, and Morphology data in text
file. Step 11 - Print Q-reports (quantitative
reports) for Data Base Step 12 - Calculate and
save cell expression false color images of each
marker. Step 13 - Display and save image
comparison for Data Base. Step 14 Read all
data and print graphs for quantitative
comparison.
14
Mosaic images of the expression level of each
biomarker.
15
Heat map colored mosaics of the expression levels
of each biomarker normalized to the nuclear stain
mosaic.
16
Qualitative and quantitative analysis can be
performed the cell level
17
Conclusion
DAPI Staining Too Bright System Standard
Spectrum Measurements Cross talk of Adjacent
Standards for Controls Accumulations to Augment
Dynamic Range False Color Digital Stain Pathology
Validation Technique Potential to Save Time and
Cost