Artificial Intelligence in Label-free Microscopy: Biological Cell Classification by Time Stretch

151 Pages·2017·4.823 MB·English

by  Ata Mahjoubfar, Claire Lifan Chen, Bahram Jalali (auth.)

Ata Mahjoubfar · Claire Lifan Chen
Bahram Jalali
Artifi cial
Intelligence
in Label-free
Microscopy
Biological Cell Classifi cation by Time
Stretch
Artificial Intelligence in Label-free Microscopy
Ata Mahjoubfar • Claire Lifan Chen • Bahram Jalali
Artificial Intelligence
in Label-free Microscopy
Biological Cell Classification by Time Stretch
123
AtaMahjoubfar ClaireLifanChen
UniversityofCaliforniaLosAngeles UniversityofCaliforniaLosAngeles
LosAngeles,CA,USA LosAngeles,CA,USA
BahramJalali
UniversityofCaliforniaLosAngeles
LosAngeles,CA,USA
ISBN978-3-319-51447-5 ISBN978-3-319-51448-2 (eBook)
DOI10.1007/978-3-319-51448-2
LibraryofCongressControlNumber:2017937310
©SpringerInternationalPublishingAG2017
Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpartof
thematerialisconcerned,specificallytherightsoftranslation,reprinting,reuseofillustrations,recitation,
broadcasting,reproductiononmicrofilmsorinanyotherphysicalway,andtransmissionorinformation
storageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilarmethodology
nowknownorhereafterdeveloped.
Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication
doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant
protectivelawsandregulationsandthereforefreeforgeneraluse.
Thepublisher,theauthorsandtheeditorsaresafetoassumethattheadviceandinformationinthisbook
arebelievedtobetrueandaccurateatthedateofpublication.Neitherthepublishernortheauthorsor
theeditorsgiveawarranty,expressorimplied,withrespecttothematerialcontainedhereinorforany
errorsoromissionsthatmayhavebeenmade.Thepublisherremainsneutralwithregardtojurisdictional
claimsinpublishedmapsandinstitutionalaffiliations.
Printedonacid-freepaper
ThisSpringerimprintispublishedbySpringerNature
TheregisteredcompanyisSpringerInternationalPublishingAG
Theregisteredcompanyaddressis:Gewerbestrasse11,6330Cham,Switzerland
Preface
High-throughputreal-timeimagingandvisionsystemsforcaptureandidentification
of fast phenomena are among the most essential tools for scientific, industrial,
military,andmostimportantlybiomedicalapplications.Thekeychallengeinthese
instruments is the fundamental trade-off between speed and sensitivity of the
measurementsystemduetothelimitedsignalenergycollectedineachmeasurement
window. Based on two enabling technologies, namely, photonic time stretch and
opticalamplification,severalnovelhigh-throughputopticalmeasurementtoolsare
recently developed for applications such as volumetric scanning, vibrometry, and
flowcytometry.Here,weintroducetimestretchimaging,ahigh-contentcomputer
vision system developed for big data acquisition and analysis of images. Time
stretch imaging is able to capture quantitative optical phase and intensity images
simultaneously, enabling accurate surface inspection, volumetric scans, defect
detection,cellanalysis,andcancerdiagnostics.
We further describe a complete artificial intelligence pipeline for time stretch
microscopy that performs optical phase measurement, image processing, feature
extraction,and classification.Multiplebiophysicalfeaturessuchasmorphological
parameters,opticallosscharacteristics,andproteinconcentrationaremeasuredon
individual biological cells. These biophysical measurements form a hyperdimen-
sionalfeaturespaceinwhichsupervisedlearningisperformedforcellclassification.
The technology is in clinical testing for blood screening and circulating tumor
cell detection, as well as studying lipid-accumulating algal strains for biofuel
production. By integrating machine learning with high-throughput quantitative
imaging,thissystemachievesrecord-highaccuracyinlabel-freecellularphenotypic
screeningandopensupanewpathtodata-drivendiagnosis.
Furthermore,weexplainedareal-timeimagecompressiontechniqueperformed
in the optical domain to solve the big data challenge created by ultrafast mea-
surementsystems.Manyultrafastandhigh-throughputdataacquisitionequipment,
including time stretch imaging, produce a torrent of data in a short time, e.g.,
several gigabytes per second. Such a data volume and velocity place a burden on
data acquisition, storage, and processing and call for technologies that compress
images in opticaldomain and in real-time. As a solution, we have experimentally
v
vi Preface
demonstratedwarpedtimestretch,whichoffersvariablespectral-domainsampling
rate, as well as the ability to engineer the time-bandwidth product of the signal’s
envelope to match that of the data acquisition systems. We also show how to
design the kernel of the transform and, specifically, the nonlinear group delay
profile governed by the signal sparsity. Such a kernel leads to smart detection
with nonuniform spectral resolution, having direct utility in improvementof data
acquisition rate, real-time data compression, and enhancement of ultrafast data
captureaccuracy.
LosAngeles,CA,USA AtaMahjoubfar
ClaireLifanChen
BahramJalali
Acknowledgements
We would like to thank all of our publication coauthors, which our collaborative
works consist many chapters of this book. Chapter 2 is a version of IEEE 49th
AnnualConferenceonInformationSciencesandSystems(CISS)Proceedings,pp.
7086896,2015[1].Chapter3isaversionofAppliedPhysicsLetters,Vol.98,No.
10, 101107, 2011 [12]. Chapter 4 is a version of Proceedings of SPIE, Frontiers
in Ultrafast Optics: Biomedical, Scientific, and Industrial Applications XIII, San
Francisco, CA, 2013, 86110N [16]. Chapter 5 is a version of Biomedical Optics
Express, Vol. 4, No. 9, pp. 1618–1625,2013 [17]. Chapters 6 and 8 are based on
ScientificReports,Vol.6,pp.21471,2016[3].Chapter7isaversionofProceedings
of SPIE, High-Speed Biomedical Imaging and Spectroscopy: Toward Big Data
Instrumentation and Management II, San Francisco, CA, 2017, 100760J [2].
Chapter 9 is a version of PLoS ONE, Vol. 10, No. 4, pp. e0125106, 2015 [104].
Finally,Chapter10isaversionofScientificReports,Vol.5,pp.17148,2015[4].
vii
Contents
PartI TimeStretchImaging
1 Introduction................................................................. 3
2 TimeStretch................................................................. 7
2.1 TimeStretchImaging................................................ 7
2.2 CellClassificationUsingTimeStretchImaging.................... 9
2.3 Label-FreePhenotypicScreening ................................... 10
2.4 WarpedTimeStretchforDataCompression........................ 10
PartII InspectionandVision
3 Nanometer-ResolvedImagingVibrometer .............................. 15
3.1 Introduction........................................................... 15
3.2 ExperimentalDemonstration ........................................ 16
3.3 TheoreticalStudyoftheVibrometerPerformance ................. 18
3.4 ExperimentalResults................................................. 18
3.5 Conclusion............................................................ 19
4 Three-DimensionalUltrafastLaserScanner............................ 21
4.1 Introduction........................................................... 21
4.2 PrincipleofHybridDispersionLaserScanner...................... 22
4.3 ApplicationsofHybridDispersionLaserScanner.................. 23
PartIII BiomedicalApplications
5 Label-FreeHigh-ThroughputPhenotypicScreening................... 33
5.1 Introduction........................................................... 33
5.2 ExperimentalSetup .................................................. 35
5.3 ResultsandDiscussion............................................... 38
5.4 Conclusion............................................................ 41
ix
x Contents
6 TimeStretchQuantitativePhaseImaging............................... 43
6.1 Background........................................................... 43
6.2 TimeStretchQuantitativePhaseImaging........................... 45
6.2.1 Overview.................................................... 45
6.2.2 ImagingSystem............................................. 47
6.2.3 SystemPerformanceandResolvablePoints .............. 49
6.2.4 MicrofluidicChannelDesignandFabrication ............ 49
6.2.5 CoherentDetectionandPhaseExtraction................. 51
6.2.6 CellTransmittanceExtraction ............................. 56
6.2.7 ImageReconstruction ...................................... 57
6.3 ImageProcessingPipeline........................................... 59
6.3.1 FeatureExtraction.......................................... 59
6.3.2 MultivariateFeaturesEnabledbySensorFusion ......... 61
6.3.3 SystemCalibration ......................................... 63
6.4 Conclusion............................................................ 63
PartIV BigDataandArtificialIntelligence
7 BigDataAcquisitionandProcessinginReal-Time..................... 67
7.1 Introduction........................................................... 67
7.2 TechnicalDescriptionoftheAcquisitionSystem .................. 68
7.3 BigDataAcquisitionResults........................................ 70
7.4 Conclusion............................................................ 71
8 DeepLearningandClassification......................................... 73
8.1 Background........................................................... 73
8.2 MachineLearning.................................................... 74
8.3 Applications.......................................................... 76
8.3.1 BloodScreening:DemonstrationinClassification
ofOT-IIandSW-480Cells ................................. 76
8.3.2 Biofuel:DemonstrationinAlgaeLipidContent
Classification................................................ 77
8.4 FurtherDiscussionsinMachineLearning .......................... 80
8.4.1 LearningCurves............................................ 80
8.4.2 PrincipalComponentAnalysis(PCA)..................... 80
8.4.3 CrossValidation ............................................ 83
8.4.4 ComputationTime.......................................... 84
8.4.5 DataCleaning............................................... 84
8.5 Conclusion............................................................ 85
PartV DataCompression
9 OpticalDataCompressioninTimeStretchImaging................... 89
9.1 Background........................................................... 89
9.2 WarpedStretchImaging ............................................. 90
9.3 OpticalImageCompression ......................................... 91
Contents xi
9.4 ExperimentalDesignandResults ................................... 95
9.5 Conclusion............................................................ 99
10 DesignofWarpedStretchTransform.................................... 101
10.1 Overview.............................................................. 101
10.2 KernelDesign ........................................................ 104
10.2.1 SpectralResolution......................................... 106
10.2.2 GroupDelayProfileDesign................................ 110
10.2.3 SimulationModel........................................... 114
10.2.4 Spectrograms ............................................... 116
10.3 Discussion ............................................................ 116
10.4 Conclusion............................................................ 118
11 ConcludingRemarksandFutureWork ................................. 121
References......................................................................... 123
Index............................................................................... 131