Confocal Microscopy and Striated Tool Marks

Report
Confocal Microscopy and Striated Tool Marks:
A Statistical Study and
Potential Software Tools For Practitioners
3
2
1
0
1
2
3
Carol Gambino, Patrick McLaughlin, Loretta Kuo, Peter Diaczuk, Gerard
Petillo, Frani Kammerman, Lauren Claytor, Peter Shenkin, Nicholas
Petraco, James Hamby and Nicholas D. K. Petraco
Outline
• Introduction
• Details of Our Approach
• Data acquisition
• Confocal Microscopy for Surfaces
• Surface/Profile Pre-processing
• Results of Statistical Discrimination/Error
Rate Estimates for Primer Shears
• Suggested Operating Guidelines
Introduction
• DNA profiling the most successful application of
statistics in forensic science.
• Responsible for current interest in “raising standards” of
other branches in forensics…??
• No protocols for the application of statistics to
comparison of tool marks.
• Our goal: application of objective, numerical pattern
comparison to tool marks
Caution: Statistics is not a panacea!!!!
Background Information
• All impressions made by tools and firearms can
be viewed as numerical patterns
– Machine learning trains a computer to recognize
patterns
• Can give “…the quantitative difference between an
identification and non-identification”Moran
• Can yield identification error rate estimates
Data Acquisition
• Obtain striation/impression
patterns from 3D confocal
microscopy
• Store files in ever expanding
database
• Data acquisition is labor
intensive and time consuming
• Data files will be made
available to practitioner
community through web
interface
Confocal Microscopy
Increasing surface height
All-in-Focus 2D Image
Overlay confocal “z-stack”
• 3D confocal image of portion of chisel striation pattern
Glock Fired Cartridges
Bottom of
Firing pin imp.
Glock 19 fired cartridge cases
Five Consecutively Manufactured Chisels
G. Petillo
Lead impression media
Striation patterns
generated at 32o
70 striation patterns total:
•20 for traditional comparison
•50 for confocal microscopy
G. Petillo
Experimental Research Design
Striation patterns
in lead
Striation patterns
in wax
Craftsman Screwdrivers
Extended ImageJ Functionality
2D profiles
3D surfaces
(interactive)
• Software can detect edges of significant “lines”:
1.0
• Or software can turn any profile into a “barcode”:
-0.5 0
0.0 1
0.5 2
barcode
-1.0-1
rep(0,profleingtlbarcode
[dproferih(deri
vi.lzeros]
v.zeros))
Experimental Research Design
0
1000
profile
2000
3000
1:length(barcode)
1:length(profil)
deriv.zeros
4000
5000
Website
Downloadable
• 3D surfaces
– ImageJ visualization
• 2D and 3D
– ImageJ measurements
• R scripts/programs for
statistical analysis
• Preprints of papers
• Striation pattern
processing:
Hamby 4
• Form removal
• Register and optionally
shift skewed profiles
• Use max CCF
• Optional filter surface
into waviness and
roughness components
• Cubic spline filter:
lc = 0.08 mm
Claytor 1
Primer Shear
Primer shear from the same Glock 19
Research
Results
Primer shear from two different
Glock 19s
Experimental Research Design
Mean total profile:
Mean “waviness”
profile:
Mean “roughness”
profile:
What Statistics Can Be Used?
• Statistical pattern comparison!
• Modern algorithms are called
machine learning
• Idea is to measure
features that
characterize
physical evidence
• Train algorithm to
recognize “major”
differences between
groups of features while taking into account
natural variation and measurement error.
Primer Shear
•
Primer shears (82-91 profiles)
– PCA-SVM, CPT at the 95% level of confidence
• Empirical error rate was 4.7%
• No “uninformative” intervals were returned
– PCA-SVM, HOO-CV
• Error rate estimate is 0.0%-4.4%, depending on
the number of replicates
– PLS-DA, Bootstrap (>10 replicates only)
• 95% confidence interval for error rate: [0%, 0%]
– PLS-DA, HOO-CV
• Error rate estimate is 0.0%-4.3%, depending on the number of replicates
•
•
Results so far are on par with expectations
More samples are being prepared for analysis
Preliminary suggested operating guidelines
• Visualization is MOST important
– Trained examiner assessment
– 3D microscopy and visualization
• For statistical analysis:
– # of replicates VERY important.
– Train “machine learning method” on suspect tool and
tools producing “similar marks” (close in data space)
• SVM, PLS-DA
– Get I.D. error rate estimates from various methods
• Large test sets, bootstrap, cross-validation
– Classify an unknown form a crime scene
• Use CPT for a level of confidence for the “association”
References
•
Biasotti AA. A statistical study of the individual characteristics of fired bullets. J Forensic Sci
1959;4(1):34-50.
•
Efron B, Tibshirani RJ. An introduction to the bootstrap. 1st ed. Boca Raton: Chapman &
Hall/CRC, 1993.
•
Geradts Z, Keijzer J, Keereweer I. A new approach to automatic comparison of striation
marks. J Forensic Sci 1994;39(4):974-80.
•
AFTE. Theory of Identification as it Relates to Toolmarks. AFTE J. 1998;30(1):89-8.
•
Moran B. A report on the AFTE theory of identification and range of conclusions for tool
mark identification and resulting approaches to casework. AFTE J 2002;34(2):227-35.
•
Vovk V, Gammerman A, Shafer G. Algorithmic learning in a random world. 1 ed. New York:
Springer, 2005.
•
Neel M, Wells M. A comprehensive statistical analysis of striated tool mark examinations.
Part 1: Comparing known matches to known non -matches. AFTE J 2007;39(3):176-98.
•
Gammerman A, Vovk V. Hedging predictions in machine learning. The Computer J
2007;50(7):151-77.
References
•
Schafer G, Vovk V. A tutorial on conformal prediction. J Machine Learning Research 2008;9:371421.
•
Howitt D, Tulleners F, Cebra K, Chen S. A calculation of the theoretical significance of matched
bullets. J Forensic Sci 2008;53(4):868-75.
•
Chumbley LS, Morris MD, Kreiser MJ, Fisher C, Craft J, Genalo LJ Davis S, Faden D, Kidd J.
2010. Validation of Tool Mark Comparisons Obtained Using a Quantitative, Comparative, Statistical
Algorithm. J Forensic Sci 2010;55(4):953-961.
•
Bachrach B, Jain A, Jung S, Koons RD. A statistical validation of the individuality and repeatability
of striated tool marks: Screwdrivers and tongue and groove pliers. J Forensic Sci 2010;55(1):348-57.
•
Chu W, Song J, Vorburger T, Yen J, Ballou S, Bachrach B. Pilot study of automated bullet signature
identification based on topography measurements and correlations. J Forensic Sci 2010;55(2):341-7.
•
Petraco N. Color Atlas of Forensic Toolmark Identification. 1st ed. Boca Raton: Chapman &
Hall/CRC, 2010.
•
Petraco NDK, Shenkin P, Speir J, Diaczuk P, Pizzola PA, Gambino C, Petraco N. Addressing the
National Academy of Sciences’ Challenge: A Method for Statistical Pattern Comparison of Striated
Tool Marks. J Forensic Sci 2011, (accepted).
•
Gambino C, McLaughlin P, Kuo L, Kammerman F, Shenkin P, Diaczuk P, Petraco N, Hamby J,
Petraco NDK. Forensic Surface Metrology: Tool Mark Evidence. Scanning, 2011 (accepted).
Acknowledgements
• National Institute of Justice
• New York City Police Department Crime Lab
• John Jay College of Criminal Justice
• Research Team:
• Ms. Alison Hartwell, Esq.
• Mr. Peter Diaczuk
• Ms. Lauren Claytor
• Ms. Carol Gambino
• Dr. James Hamby
•
• Helen Chan
• Dr. Thomas Kubic, Esq.
•
• Manny Chaparro
• Off. Patrick McLaughlin
•
• Aurora Ghita
• Mr. Jerry Petillo
•
• Eric Gosslin
• Mr. Nicholas Petraco
•
• Frani Kammerman
• Dr. Peter A. Pizzola
•
• Brooke Kammrath
• Dr. Jacqueline Speir
• Loretta Kuo
• Dr. Peter Shenkin
• Dale Purcel
• Mr. Peter Tytell
Stephanie Pollut
Rebecca Smith
Elizabeth Willie
Chris Singh
Melodie Yu
Greg Frasier
Website Information and Reprints/Preprints:
[email protected]
[email protected]

similar documents