1

Optical forward scattering for bacterial Optical forward scattering for bacterial colony differentiation and identification colony differentiation and identification ……..……..an interdisciplinary approachan interdisciplinary approach

Light scatteringProf. A. Bhunia (bhunia@purdue.edu)

l

Prof. D. HirlemanDr. Euiwon BaeNan BaiS. GuoDr. Nebeker, Ben BucknerSchool of Mechanical Engineering

11

Microbiology

Data analysisand processing

Dr. A. AroonnualAmanda BettassoDr. P. BanadaKarleigh HuffDr. Seung OhkAbrar Adil (UG)Amanda LathropDept of Food Science

Engineering

Prof. J. P. RobinsonDr. B. RajwaValery PatsekinBulent BayraktarDept of Basic MedicalSciences, Bindley Bioscience Center

USDA-ARSSupport/

collaboration

Collaborator: Gary Richards, USDA-ARS, DE

Oct 27-28, 2009

BARDOT BARDOT –– ReportingReporting BARDOT based analysis of serovars of E. coli,

Salmonella and Vibrio Detection of Salmonella from spiked peanut

butter Roadmap for multiRoadmap for multi--scale Forward Scattering scale Forward Scattering

Ph iPh i (FSP)(FSP)

22

PhenomicsPhenomics (FSP)(FSP) Generate scatter patterns for molds and yeast Determine scatter patterns for microcolonies for

early detection/identification System parameter optimization for best

performance (interfacing image analysis software)

BaBacterial cterial RRapid apid DDetection using etection using OOptical ptical Scattering Scattering TTechnology (BARDOT)echnology (BARDOT)

Laserλ: 635 nmBeam dia: 1 mm

Power: 0.1-0.2 mWCCD Pixel: 640x480

LaserPetri

Petri dish with bacterial colonies

CCD camera Cable to computer

CCD Pixel: 640x480

ComputerComputer

dish with

colonies

CCD chip

Scatter image

Banada et al. 2009. Biosens. Bioelectron. 24:1685-1692

Specifications

EnUrgaEnUrga

Second generation

pDimension: 24x20x17 inchWeight: 75 lbsOperationColony counts: line scan: 40 secScatter image: 5 sec/colony

3rd Generation BARDOT

4th generation!!!Advanced Bioimaging Systems(advancedbioimagingsystems.com)

•Analyzed scatter patterns of 10 different O-antigenic E coli strains

Analysis of scatter patterns of different O-antigenic E. coli strains

with selective and non-selective media(A. Aroonnual et .al.)

antigenic E. coli strains

•Media examined: Non-selective media (BHI) and selective media (SMAC and CT-SMAC)

•Growth conditions: 37oC for 12-13 h

2

Cryo-SEM photographs of coloniesL. monocytogenes L. innocua

L. mononocytogenes L. innocua

Banada et al. 2007. Biosens. Bioelectron. 22:1664

Banada et al. 2009. Biosens. Bioelectron. 24:1685

0

5

10

15

20

EPS Proteins

L. monocytogenes

L. innocua

0

100

200

300

400

500

600 L. mono L. innocua

Pentosesand uronic

acids

Hexoses

mg/m

l

µg/m

l

• Part of lipopolysaccharide (LPS) in gram-negative

bacteria (Lipid A, Core polysaccharide, and O-

polysaccharide)

• O-antigen specific chains determine specificity

O-antigen

http://bs.kaist.ac.kr/~mbtlab/Gram_Neg.jpg

http://www.mbl.edu/marine_org/images/animals/Limulus/blood/lpstyl01.gif

Gram-negative cell wall Lipopolysaccharide of Gram-negative bacteria

BARDOT images of E. coli colonies on BHI

O157:H7 DEL933

O157:H7 SEA13A53

12

12

E. coli strain Time (h) Images on BHI

O127:H6 EPEC

O142:H6 EPEC

O25:K98 NM ETEC

O78:H11 ETEC

12

12

12

12

O5 NM

O46:H38

12

12

E. coli strain Time (h) Images on BHI

BARDOT images of E. coli colonies on BHI

O111:H11

O91:H21

12

12

O157:H7 EDL933

O157:H7 SEA13A53

12

13

grey

grey

E. coli strain Time (h) Color Images on SMAC

BARDOT images of E. coli colonies on SMAC

O127:H6 EPEC

O142:H6 EPEC

O25:K98 NM ETEC

O78:H11 ETEC

12

13

12

13

grey

grey

pink

pink

O5 NM

O46:H38

13

13

grey

pink

E. coli strain Time (h) Color Images on SMAC

BARDOT images of E. coli colonies on SMAC

O46:H38

O111:H11

O91:H21

13

13

13

pink

grey

pink

3

BARDOT images of E. coli colonies on SMAC and CT-SMAC

O157:H7 EDL933

12

14

SMAC

CT SMAC

E. coli strain Time (h) Media BARDOT Images

O127:H6 EPEC

14

12

18

CT-SMAC

SMAC

CT-SMAC

BARDOT analysis of BARDOT analysis of SalmonellaSalmonella serovars serovars with selective and nonwith selective and non--selective mediaselective media

(A. Bettasso et .al.)(A. Bettasso et .al.)

Salmonella Salmonella serovarsserovars cultured overnight in cultured overnight in RappaportRappaport VassiliadisVassiliadis RR10 10 brothbroth

Spread plated on BHI (general) and XLD Spread plated on BHI (general) and XLD (selective) and Brilliant Green agars(selective) and Brilliant Green agars

Incubated at Incubated at 3737°°C for C for 1212hh Screened with BARDOTScreened with BARDOT

1414

Salmonella serovar scatter patterns Salmonella serovar scatter patterns on BHIon BHI

S. Typhimurium var copenhagen

S. Litchfield

BHI, 12 h, 37°C

S. Indiana

S. Typhi

S. Enteriditis PT14b

Salmonella Typhimurium on BHI

Salmonella serovar scatter patterns on BHISalmonella serovar scatter patterns on BHI

S. Agona

S. Cholerasuis

BHI/12 h/ 37°C

S. Heidelberg

S. Poona

S. Schottmuelleri

S. Seftenberg

S. Typhimurium var copenhagen

S. Litchfield

SalmonellaSalmonella on selective media (XLD)on selective media (XLD)

Salmonella

XLD/12h/37°C

S. Indiana

S. Typhi

S. Enteriditis PT14b

Salmonella Typhimurium on XLD

Salmonella on selective media (XLD)Salmonella on selective media (XLD)

S. Agona

S. Cholerasuis

S. Heidelberg

12h /XLD/ 37°C

S. Poona

S. Schottmuelleri

S. Seftenberg

4

Application of BARDOT for Application of BARDOT for SalmonellaSalmonelladetection/identification from peanut detection/identification from peanut

butter butter (A. Bettasso et al.) (A. Bettasso et al.) Peanut butter samples were inoculated with Peanut butter samples were inoculated with

100 100 CFU/CFU/2525g with g with SalmonellaSalmonella Typhimurium Typhimurium varvar CopenhagenCopenhagen

AnalysedAnalysed on Days on Days 00,,77,,1414,,2121, and , and 2828 Enriched for Enriched for 0 0 or or 1212h in RV brothh in RV broth Spread plated on XLD and BHI Spread plated on XLD and BHI Incubated for Incubated for 1212h at h at 3737°°CC Screened with BARDOTScreened with BARDOT

SalmonellaSalmonella detection from peanut butterdetection from peanut butterXLD BHI

0h enrich 12h enrich 0h enrich 12h enrich

Day 0

Day 7

Day 14

Day 21

Day 28

Vibrio cincinnatiensis

Vibrio

Vibrio alginolyticusVibrio campbelli

Vib i i i

V. cholerae O1 (6707)

V. cholerae non O1 (140-58)

Vibrio cincinnatiensis

Vibrio fluvialis

Vibrio mimicus

Vibrio parahaemolyticus

Vibrio anguillarumVibrio holisae

Vibrio orientalis

Vibrio metchinovic

V. cholerae O139 (CDC

V. cholerae O1E 16

V. cholerae non O1 140-58 18

V. cholerae O1 6707 24

Vibrio sp. BARDOT images on BHI1%NaClh

V. cholerae O139 CDC 14

V. parahemolyticus ATCC 17803 14

V. parahemolyticus

SPRC 10290 tdh+/trh+ 15

Vibrio sp. BARDOT images on BHI1%NaClh

Vp DIE 12052499 tdh+/trh+ O1:Kuk 16

Vp CPA 7081699 tdh+/trh-, O4:K8 13p ,

Vp BAC 983547 tdh-/trh-, O4:K55 11

Vp AQ4037 tdh-/trh+, O3:K6 12

Aroonnual et al.

Detection of Detection of Vibrio choleraeVibrio cholerae in presence in presence of other Vibrio of other Vibrio sppspp

2424

V. orientaris V. fluviaris V. cholerae nonO1

Mixture of Vibrio cultures (V. orientaris, V. fluviaris, and V. cholerae nonO1 on BHI+1%NaCl at 30oC for 18h.

Aroonnual et al.

5

The extendable version of the BARDOT The extendable version of the BARDOT systemsystem

Research laboratory systemResearch laboratory system Dynamic learningDynamic learning, n, nonon--exhaustive exhaustive

learning (could be also paired learning (could be also paired with mathematical modelling of with mathematical modelling of colony growth)colony growth)

Multiple spectral bandsMultiple spectral bands, m, multiple ultiple S l i t g ti

Tunable Light source

Backscatter detector

Fluorescence detector

Polarizers

Interferprofilosc

ence

l i

mag

er

polarizationspolarizations Integrated microincubation Integrated microincubation

systemsystem

Colony localization using confocal Colony localization using confocal profilometryprofilometry

Sample interogation area

Forward scatter detector

Polarizers

rometric

ometerFl

uore

sco

nfoc

al

PatternPattern--matching using orthogonal radial polynomials (Zernike, matching using orthogonal radial polynomials (Zernike, pseudopseudo--Zernike, Zernike, ChebyshevChebyshev…)…)

A B C D

Example: unsupervised hierarchical clustering Example: unsupervised hierarchical clustering ((SalmonellaSalmonella))

Kentucky

Agona

Thomasville

Typhimurium (Copenhagen)

40

6040 60

LD2

5

105 10

-5

0

-5 0

LD32

4

6 2 4 6

-4

-2

0

-4 -2 0

Linear Discriminant AnalysisMixture V. cholerae V. parahaemolyticus V. vulnificus

Aver

age

true

pos

itiv

e ra

te0

0.2

0.4

0.6

0.8

1.0

V. cholerae

V. parahaemolyticus

V vulnificus

Example: classification of Example: classification of VibrioVibriopathogens (single colonies)pathogens (single colonies)

SVM classifier

Scatter Plot Matrix

LD1

0

20

0 20

Average false positive rate0.0 0.2 0.4 0.6 0.8 1.0

0.0 V. vulnificus

Class Classifier Sensitivity Specificity Accuracy AUC

MixtureFLD 0.93 0.99 0.97 1.00

SVM-L 0.93 0.98 0.97 0.99

SVM-RBF 0.96 0.99 0.98 1.00

V. choleraeFLD 0.77 0.98 0.94 0.97

SVM-L 0.88 0.98 0.96 0.99

SVM-RBF 0.94 0.99 0.98 1.00

V. parahaemolyticusFLD 0.95 0.90 0.93 0.98

SVM-L 0.95 0.94 0.95 0.99

SVM-RBF 0.97 0.96 0.97 1.00

V. vulnificusFLD 1.00 1.00 1.00 1.00

SVM-L 1.00 1.00 1.00 1.00

SVM-RBF 0.99 1.00 1.00 1.00

Classification of plates containing Classification of plates containing VibrioVibrio pathogenspathogens (multiple colonies)(multiple colonies)

The plate is considered “negative” (or clean) if ALL the colonies are non-pathogenic (or they do not belong to the pathogenic classes of interest)

The plate is considered “positive” (or contaminated) if AT LEAST ONE of the colonies belongs to a pathogenic class.

This classifier can be tuned selecting proper cut-off levels to obtain desired sensitivity and specificity

Species Sensitivity Specificity

V. cholerae 1.00 1.00

V. parahaemolyticus 0.96 1.00

V. vulnificus 1.00 1.00

GrowthTime

2 hr 8 hr 24 hr

Scalar diffraction theoryDiscrete Dipole Approximation

Roadmap for multi-scale Forward Scattering Phenomics (FSP)

Tra

nsm

issi

on T

ype

Th

eory

Exp

erim

ent

&

Inst

rum

ent

yp pp

BARDOT w/ micro incubatorμ-BARDOT BARDOT

Reaction –diffusion model

6

Micro-scale : colony dipole model constructorModel Parameters: Bacteria No. = 200Bacteria Size = 0.855 µm x 0.315µm x 0.315µmColony Size = 13.5 µm x 1.35 µmDipole Spacing = 0.045 µmWavelength = 0.6328 µm (p-polarized plane wave with normal incidence)Bacteria Refractive Index = 1.38Extracellular Refractive Index = 1.40

Modeling Results from DDSCAT

FS is favored against BS with 1000 times stronger signal

Wavefront and Scattering Pattern Generated by Ray Optics Model (I)

Incoming wave(laser)

Bacterial colony as a amp/phase modulator

Scattered lightAt image plane

Wavefront and Scattering Pattern Generated by Ray Optics Model (II)

-2

Wavefront emerging from the colony imprinted with its 3D phase structure

information

Intensity field distribution of the propagating wave calculated by

Fresnel Diffraction

-2 -1.5 -1 -0.5 0 0.5 1 1.5 2

-1.5

-1

-0.5

0

0.5

1

1.5

2

ConclusionsConclusionsBARDOT show differential scatter signatures for BARDOT show differential scatter signatures for

different different E. coli E. coli serovars primarily on serovars primarily on nonselective medianonselective media

BARDOT show differential scatter signatures for BARDOT show differential scatter signatures for different different SalmonellaSalmonella serovars primarily on serovars primarily on

l i dil i diselective mediaselective mediaSalmonella was successfully detected from spiked Salmonella was successfully detected from spiked

peanut butter samples with or without peanut butter samples with or without enrichment from samples stored for enrichment from samples stored for 00, , 77, , 1414, , 21 21 and and 28 28 days at RT.days at RT.

Vibrio cholerae, V. parahemolyticus, V. vulnificus Vibrio cholerae, V. parahemolyticus, V. vulnificus can be detected from a mixture of other Vibriocan be detected from a mixture of other Vibrio

3535

The new version of the BARDOT systemThe new version of the BARDOT system

Sample interogation area

Red-laser light source

Forward scatter

Whi

te li

ght

imag

er

• Developed at Purdue University and funded by USDA

• Commercialized by Advanced Bioimaging Systems

• Fully customizable for any workflow/requirements

• Deployed at USDA July, 2009

scatter detector