Validating A Currently Available Measuring Tool for Melanoma Gabor Hertz, M.D.
Jan 01, 2016
Ancillary uses could ease its adoption
• Skin cancers are common neoplasms in clinical practice (1/5)
• Melanoma • Treatment and
Prognosis depend on stage– <1 mm– 1-4 mm
Surgical Cases
Non-derm surgicals Skin
Design
• Retrospective validation study
• 18 cases 2006-2007
• Digitally imaged at varying optical mag.
• Measurements compared to clinically reported Breslow thickness
ResultsCase # Mean mBreslow Clinically Reported Breslow Diff Stage
07-1825 0.53 0.3 75% I
07-04175 0.41 0.5 18% I
07-05848 3.79 1.45 175% IIA
07-06267 0.5 0.56 10% I
07-06998 0.3 0.31 3% I
07-10490 0.91 1 9% IIA
07-11673 1.8 >1.5 20% IIB
07-13229 3.14 3.48 10% IIB
07-14139 0.86 0.88 2% I
07-17863 3.79 4 5% IIB
07-17899 1.33 1.5 11% IIA
07-17942 1.04 >1.0 4% IIA
07-17972 0.32 0.4 20% I
07-22303 4.58 4.5 2% IIC
08-00367 0.3 0.4 25% I
08-001640 0.26 0.38 32% I
08-03681 0.35 0.35 0% I
08-07804 0.35 .35-.4 13% I
Results
• Validating against stage micrometer– Avg 0.3 mm
discrapency (30%/1mm)
• Validating against 18 clinical case– Avg 0.12 mm
discrepency (24%)
Discrepency against stage micrometer
• Technical problems with initial installation as it focused only on telepath system
• Optical input was not initially configured/calibrated
Slightly tighter correlation against clinical measurements
• Likely due to additional confounding variables– Pathologic diagnosis/measurement is really
semiquantitative and known to be subject to large intra and interobserver variation.
Lessons
• Even off the shelf products require anticipation of ancillary uses.
• New technology has a learning curve
• Applying highly accurate measuring tools will not by itself improve an inherently poorly reproducible measurement (sampling and interpretation variables need to be addressed)
Next Steps
• Develop the required technical understanding of digital medical imaging in microscopy
• Determine method and timing of input image calibration.
• Identify and rely on intra-image controls• Develop logistical support to manage
confounding variables (sampling & Dx variation)
System Requirements
• Pentium processor
• At least Windows 98
• 256 MB RAM
• 200 MB Free space
• VGA monitor with 1024 x 768 resolution and 256 colors
• Microsoft Internet Explorer 3.02
Case #Mean mBreslow
Clinically Reported Breslow Diff Stage
07-1825 0.56 0.3 87% I
07-04175 0.5 0.5 0% I
07-05848 1.66 1.45 15% IIA
07-06267 0.16 0.56 72% I
07-06998 0.31 0.31 0% I
07-10490 0..97 1 3% IIA
07-11673 1.5 >1.5 0% IIB
07-14139 1 0.88 14% I
07-17899 1.26 1.5 16% IIA
07-17942 2.48 >1.0 148% IIA
07-17972 0.26 0.4 35% I
07-22303 4.22 4.5 6% IIC
08-00367 0.32 0.4 20% I
08-001640 0.38 0.38 0% I
08-03681 0.36 0.35 3% I
08-07804 0.41 .35-.4 3% I