data analysis department of laboratory medicine university of washington
TRANSCRIPT
Data Analysis
Department of Laboratory Medicine
University of Washington
Data Analysis
• Assess data quality– Remove artifacts
• Identify populations• Compare with normal
– Identify abnormal populations– Quantitate and evaluate immunophenotype
• Generate report
Assess Data Quality
Detector Optimization
Negative populations entirely on scale
Degeneration
Increase SSDecrease FS
08-03307
Degeneration
Decrease in intensity for many antigens
08-03307
Viability Gate
08-03307
Viability Gate
All cells Viable cells
08-03307
Sample Exhaustion
Air in system gives rise to many spurious signalsEvent gate to exclude non-real events
Laser Delay
Fluidic instability - Monitor events over time to detect
Laser Delay
Original
Gated
Doublet Discrimination
Doublet Discrimination• Doublets = > one cell in laser simultaneously
– High cell concentrations– Cell aggregates, sample preparation– High sample aspiration pressure
• Doublets have composite properties
• Can exclude using height, area, or width
Original07-04513
Example
Time07-04513
Example
Singlets07-04513
Example
Viable07-04513
Example
Determining Positivity
Determining Positivity
Incorrect Correct
07-08661
Population Identification
Cell Type Identification
Lymphocyte population identified by FS/SS gating
Cell Type Identification
Borowitz et al (1993) AJCP 100:534-40.Steltzer et al (1993) Ann NY Acad Sci 667:265-280
Lineage Identification
– CD19 for B cells and CD3 for T cells– Assumptions that may not always be correct– Always use at least two methods of identification
Compare with Normal
Normal B cell Maturation
Wood and Borowitz (2006) Henry’s Laboratory Medicine
Follicle Center B cells
08-01359
08-03324
Follicular Lymphoma
Follicular Hyperplasia
0.1% abnormal immature B cells
ALL MRD
06-01469
Data Analysis
• Data displayed as dot plots or histograms– Restrict to subset having high informational content
• Color discrete populations – Display information from other parameters– Allow rapid visual identification in multiple plots
• Display data in consistent manner– Pattern recognition