Upload
others
View
10
Download
0
Embed Size (px)
Citation preview
Medical Images Analysis Medical Images Analysis d P id P i 2564225642and Processing and Processing -- 2564225642
Emad Emad FatemizadehFatemizadeh
Sharif University of Technology Sharif University of Technology –– E. E. FatemizadehFatemizadeh
Course IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse Introduction
Course Information:Course Information:–– Type: GraduatedType: GraduatedType: Graduated Type: Graduated –– Credits: 3Credits: 3–– Prerequisites: Digital Image ProcessingPrerequisites: Digital Image ProcessingPrerequisites: Digital Image ProcessingPrerequisites: Digital Image Processing
Course IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse Introduction
Reference(s):Reference(s):–– Insight into Images: Principles andInsight into Images: Principles andInsight into Images: Principles and Insight into Images: Principles and
Practice for Segmentation, Practice for Segmentation, Registration, and Image AnalysisRegistration, and Image Analysis, , By: T. S. By: T. S. YooYoo, 2004 (Hardcopy), 2004 (Hardcopy)
–– Biomedical Images Analysis,Biomedical Images Analysis, by: by: R. R. M. M. RangayyaRangayya, 2004, eBook. , 2004, eBook.
–– Some papers!Some papers!–– DIP References!DIP References!
Course IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse Introduction
Evaluation:Evaluation:–– Final: 35%Final: 35%Final: 35% Final: 35% –– Homework: 20% (Mostly Simulation)Homework: 20% (Mostly Simulation)–– Research Project: 25%Research Project: 25%Research Project: 25% Research Project: 25%
Small Literature Survey (For example: Persian OCR)Small Literature Survey (For example: Persian OCR)In depth paper (one) study (Simulation and Judgment)In depth paper (one) study (Simulation and Judgment)Experiments on real dataExperiments on real data
–– TakeHomeTakeHome 20%20%
Course IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse Introduction
JournalsJournals::–– IEEE Transaction on Medical Imaging (TMI), IEEE PressIEEE Transaction on Medical Imaging (TMI), IEEE Press
Medical Image Analysis ElsevierMedical Image Analysis Elsevier–– Medical Image Analysis, Elsevier.Medical Image Analysis, Elsevier.–– Computerized Medical Imaging and Graphics (CMIG)Computerized Medical Imaging and Graphics (CMIG)–– IEEE Transaction on Biomedical Engineering. (TBE)IEEE Transaction on Biomedical Engineering. (TBE)–– IEEE Transaction on Image Processing (IP)IEEE Transaction on Image Processing (IP)–– IEEE Transaction on Pattern Analysis and Machine IEEE Transaction on Pattern Analysis and Machine
Intelligence (PAMI), IEEE Press.Intelligence (PAMI), IEEE Press.–– Pattern Recognition, (Pattern Recognition, (PergamonPergamon--Elsevier)Elsevier)–– Pattern Recognition Letters ( Elsevier)Pattern Recognition Letters ( Elsevier)
Course IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse Introduction
Course Contacts and Links:Course Contacts and Links:–– URL: URL: http:/ee.sharif.edu/~http:/ee.sharif.edu/~miapmiap
Course Lecture NotesCourse Lecture Notes–– Course Email: Course Email: [email protected]@gmail.com
l k b (l k b ( ))Electronic Homework submission (NOT .Electronic Homework submission (NOT .rarrar)!)!–– Submission rule: Submission rule:
Subject:Subject: MIAP0MIAP0nn::stdstdnumnumSubject: Subject: MIAP0MIAP0nn::stdstdnumnum–– My emails: My emails: fatemizadeh@[email protected]
fatemizadeh@[email protected]@[email protected]
Course IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse IntroductionCourse Introduction
SyllabusSyllabus::–– Introduction to Medical Images/Imaging Introduction to Medical Images/Imaging –– BrieflyBriefly
Introduction to Digital Image ProcessingIntroduction to Digital Image Processing–– Introduction to Digital Image ProcessingIntroduction to Digital Image Processing–– Segmentation (Intro to Classification ?)Segmentation (Intro to Classification ?)–– Enhancement/Enhancement/DenoisingDenoising–– RegistrationRegistration–– InterpolationInterpolation–– Medical Image Analysis Using PDEMedical Image Analysis Using PDEg y gg y g–– MultilodalMultilodal Image AnalysisImage Analysis–– Computer Aided Diagnosis (CAD) Computer Aided Diagnosis (CAD) –– Mostly MammographyMostly Mammography–– Medical Image Analysis Software (your task!)Medical Image Analysis Software (your task!)–– Medical Image Analysis Software (your task!)Medical Image Analysis Software (your task!)
Medical Images ModalitiesMedical Images ModalitiesMedical Images ModalitiesMedical Images Modalities
What is a medical image:What is a medical image:–– A geometric distribution of a certainA geometric distribution of a certainA geometric distribution of a certain A geometric distribution of a certain
physical/physiological property(physical/physiological property(iesies).).
ModalitiesModalitiesModalitiesModalities–– Several images from a certain region!Several images from a certain region!
Sharif University of Technology Sharif University of Technology –– E. E. FatemizadehFatemizadeh
Medical Images ModalitiesMedical Images ModalitiesMedical Images ModalitiesMedical Images Modalities
Concepts:Concepts:–– How to build images of internal organs ofHow to build images of internal organs ofHow to build images of internal organs of How to build images of internal organs of
body, nonbody, non--invasively.invasively.–– Image ModalitiesImage Modalitiesgg–– PrePre--processingprocessing–– PostPost--ProcessingProcessingPostPost ProcessingProcessing
Image ConstructionImage ConstructionImage ConstructionImage Construction
Goal:Goal:–– Draw images of a certain physicalDraw images of a certain physicalDraw images of a certain physical Draw images of a certain physical
property of subject anatomy.property of subject anatomy.–– Procedure in nonProcedure in non--invasive.invasive.–– Image Geometry:Image Geometry:
ProjectionProjectionjjTomographyTomography
Image ModalityImage ModalityImage ModalityImage Modality
Based on Interested Physical Property:Based on Interested Physical Property:–– XX--Ray (CT/Radiography)Ray (CT/Radiography)–– MRI (Magnetic Resonance Imaging)MRI (Magnetic Resonance Imaging)–– PET (Positron Emission Tomography)PET (Positron Emission Tomography)–– US (Ultra Sound)US (Ultra Sound)–– SPECT (Single Photon Emission CT)SPECT (Single Photon Emission CT)–– EIT (Electrical Impedance Tomography)EIT (Electrical Impedance Tomography)–– Video and etc.Video and etc.
PrePre ProcessingProcessingPrePre--ProcessingProcessing
Concepts:Concepts:–– Design optimum protocol for raw dataDesign optimum protocol for raw dataDesign optimum protocol for raw data Design optimum protocol for raw data
acquisition.acquisition.–– Image reconstruction from raw data.Image reconstruction from raw data.gg–– Noise and artifact reduction in raw data Noise and artifact reduction in raw data
space.space.pp
PostPost ProcessingProcessingPostPost--ProcessingProcessing
Concepts:Concepts:–– Noise and artifact reduction in imageNoise and artifact reduction in imageNoise and artifact reduction in image Noise and artifact reduction in image
space.space.–– Enhance images in Regions of Interest.Enhance images in Regions of Interest.g gg g–– Image partitioning to meaningful regions.Image partitioning to meaningful regions.–– Computer Aided Diagnosis (CAD)Computer Aided Diagnosis (CAD)Computer Aided Diagnosis (CAD)Computer Aided Diagnosis (CAD)–– Multimodality Image FusionMultimodality Image Fusion–– Virtual Reality (Virtual Surgery)Virtual Reality (Virtual Surgery)Virtual Reality (Virtual Surgery)Virtual Reality (Virtual Surgery)
Medical Images ModalitiesMedical Images ModalitiesMedical Images ModalitiesMedical Images Modalities
Medical Images Categories:Medical Images Categories:–– Number of channel:Number of channel:Number of channel:Number of channel:
Single channel (Only one property is Single channel (Only one property is acquired): CT , PET, USacquired): CT , PET, USMultichannel (More than one property are Multichannel (More than one property are acquired): MRIacquired): MRI
Sharif University of Technology Sharif University of Technology –– E. E. FatemizadehFatemizadeh
Medical Images ModalitiesMedical Images ModalitiesMedical Images ModalitiesMedical Images Modalities
Medical Images Categories:Medical Images Categories:–– Characteristic:Characteristic:Characteristic:Characteristic:
Anatomical: Static distribution of a certain Anatomical: Static distribution of a certain physical property, physical property, SkeletonSkeleton..Physiological/Functional: Functionality or Physiological/Functional: Functionality or Metabolism of organs, Metabolism of organs, Glucose consumption Glucose consumption in brainin brainin brainin brain..
Sharif University of Technology Sharif University of Technology –– E. E. FatemizadehFatemizadeh
Medical Images ModalitiesMedical Images ModalitiesMedical Images ModalitiesMedical Images Modalities
Medical Images Categories:Medical Images Categories:–– Geometry:Geometry:
Projective: A Straight line in the object will be Projective: A Straight line in the object will be mapped to a single point at images, mapped to a single point at images, Conventional RadiographyConventional Radiography..Conventional RadiographyConventional Radiography..Tomography: Cross section of object will be Tomography: Cross section of object will be imaged, imaged, Computerized TomographyComputerized Tomography..
Di i litDi i lit–– Dimensionality:Dimensionality:2D2D3D3D
Sharif University of Technology Sharif University of Technology –– E. E. FatemizadehFatemizadeh
3D3D
Medical Images ModalitiesMedical Images ModalitiesMedical Images ModalitiesMedical Images Modalities
Major Properties in Medical Images:Major Properties in Medical Images:–– XX--Ray TransmissionRay TransmissionXX Ray TransmissionRay Transmission–– Ultrasound Waves ReflectionUltrasound Waves Reflection–– Radioactive annihilationRadioactive annihilationRadioactive annihilationRadioactive annihilation–– Spin Density and Relaxation TimesSpin Density and Relaxation Times–– Optical (NonOptical (Non--Laser/Laser)Laser/Laser)–– Optical (NonOptical (Non--Laser/Laser)Laser/Laser)–– Electrical ConductanceElectrical Conductance
Sharif University of Technology Sharif University of Technology –– E. E. FatemizadehFatemizadeh
Medical Images ModalitiesMedical Images ModalitiesMedical Images ModalitiesMedical Images Modalities
XX--Ray Transmission:Ray Transmission:–– Simple Physics: Simple Physics: –– Absorption coefficient (Absorption coefficient (μμ) of X) of X--Ray photons (70Ray photons (70--120Kev)120Kev)
0L
TI I e μ−=Absorption coefficient (Absorption coefficient (μμ) of X) of X Ray photons (70Ray photons (70 120Kev) 120Kev) are displayed as image.are displayed as image.
–– Projection and Tomography are possible.Projection and Tomography are possible.–– Hazard: Yes!Hazard: Yes!Hazard: Yes! Hazard: Yes! –– Resolution: Very Good. Resolution: Very Good. –– SNR: Good.SNR: Good.–– Almost Static (except for Fluoroscopy and rarely used fCTAlmost Static (except for Fluoroscopy and rarely used fCTAlmost Static (except for Fluoroscopy and rarely used fCT Almost Static (except for Fluoroscopy and rarely used fCT
,Functional CT.,Functional CT.–– Good contrast for hard tissue (Bones)Good contrast for hard tissue (Bones)–– Low Contrast for soft tissue (Muscle, Tumors)Low Contrast for soft tissue (Muscle, Tumors)
Sharif University of Technology Sharif University of Technology –– E. E. FatemizadehFatemizadeh
( , )( , )
Medical Images ModalitiesMedical Images ModalitiesMedical Images ModalitiesMedical Images Modalities
XX--Ray Transmission:Ray Transmission:–– Examples:Examples:
C ti l R di hC ti l R di hConventional RadiographyConventional RadiographyComputerized Tomography (CT)Computerized Tomography (CT)Angiography: Some organs like as blood vessels Angiography: Some organs like as blood vessels enhanced through injection of contrast agentenhanced through injection of contrast agentenhanced through injection of contrast agentenhanced through injection of contrast agentDigital Subtraction Angiography (DSA): Difference of Digital Subtraction Angiography (DSA): Difference of two images of a single organs in the different two images of a single organs in the different conditions (Before and after contrast agent injection orconditions (Before and after contrast agent injection orconditions (Before and after contrast agent injection or conditions (Before and after contrast agent injection or two different Xtwo different X--Ray energy) are displayed.Ray energy) are displayed.Fluoroscopy: Watch oranges while the body is under XFluoroscopy: Watch oranges while the body is under X--Ray exposure.Ray exposure.
Sharif University of Technology Sharif University of Technology –– E. E. FatemizadehFatemizadeh
Medical Images ModalitiesMedical Images ModalitiesMedical Images ModalitiesMedical Images Modalities
Ultrasound Wave Reflection:Ultrasound Wave Reflection:–– Ultrasound Waves: Above 20KHz.Ultrasound Waves: Above 20KHz.–– Reflection times of incident ultrasound beam are related to Reflection times of incident ultrasound beam are related to
i i f h lli i f h llposition of the walls.position of the walls.–– Simple Physics: Simple Physics: –– Physical Characteristic in TomographyPhysical Characteristic in Tomography
Hazard: LowHazard: Low
x ct=
–– Hazard: LowHazard: Low–– Resolutions: Average (Different in two dimension)Resolutions: Average (Different in two dimension)–– SNR: BadSNR: Bad–– Anatomical and Dynamic (Movements of objects) but notAnatomical and Dynamic (Movements of objects) but notAnatomical and Dynamic (Movements of objects) but not Anatomical and Dynamic (Movements of objects) but not
metabolismmetabolism–– Problem with objects behind bone and air (lung)Problem with objects behind bone and air (lung)–– Need to access to the organs only from one side Need to access to the organs only from one side
(R fl ti )(R fl ti )Sharif University of Technology Sharif University of Technology –– E. E.
FatemizadehFatemizadeh
(Reflection)(Reflection)
Medical Images ModalitiesMedical Images ModalitiesMedical Images ModalitiesMedical Images Modalities
Ultrasound Wave Reflection:Ultrasound Wave Reflection:–– Examples:Examples:
AA mode: 1D imaging Eye’s Layersmode: 1D imaging Eye’s LayersAA--mode: 1D imaging, Eye’s Layers.mode: 1D imaging, Eye’s Layers.BB--mode (Sonography): 2D imaging, fetus, mode (Sonography): 2D imaging, fetus, Bladder, kidney, Prostate .Bladder, kidney, Prostate .CC d Ti Ch t i ti R hd Ti Ch t i ti R hCC--mode: Tissue Characterization, Research mode: Tissue Characterization, Research Application.Application.Doppler/Color Doppler: Blood Flow and Heart Doppler/Color Doppler: Blood Flow and Heart (V l d C it ) M it i(V l d C it ) M it i(Valve and Cavity) Monitoring.(Valve and Cavity) Monitoring.
Sharif University of Technology Sharif University of Technology –– E. E. FatemizadehFatemizadeh
Medical Images ModalitiesMedical Images ModalitiesMedical Images ModalitiesMedical Images Modalities
Radioactive annihilation:Radioactive annihilation:–– Source imaging: Source of radiation is located Source imaging: Source of radiation is located
inside body (Injection , inhalation and etc.)inside body (Injection , inhalation and etc.)inside body (Injection , inhalation and etc.)inside body (Injection , inhalation and etc.)–– Source radiation (consumption) distribution are Source radiation (consumption) distribution are
imaged.imaged.–– Special Drug for each organs (ISpecial Drug for each organs (I133133 for Thyroid)for Thyroid)Special Drug for each organs (ISpecial Drug for each organs (I for Thyroid)for Thyroid)–– Projection and Tomography are possibleProjection and Tomography are possible–– Hazard: Yes.Hazard: Yes.
R l ti LR l ti L–– Resolution: Low.Resolution: Low.–– SNR: Low.SNR: Low.–– Functional (Metabolism)Functional (Metabolism)
Sharif University of Technology Sharif University of Technology –– E. E. FatemizadehFatemizadeh
( )( )
Medical Images ModalitiesMedical Images ModalitiesMedical Images ModalitiesMedical Images Modalities
Radioactive annihilation:Radioactive annihilation:–– Example:Example:Example:Example:
Gamma Camera: Projection ImagingGamma Camera: Projection ImagingSPECT (Single Photon Emission Computerized SPECT (Single Photon Emission Computerized Tomography): TomographyTomography): TomographyPET (Positron Emission Tomography): Very PET (Positron Emission Tomography): Very interesting functional Imaginginteresting functional Imaginginteresting functional Imaging.interesting functional Imaging.
Sharif University of Technology Sharif University of Technology –– E. E. FatemizadehFatemizadeh
Medical Images ModalitiesMedical Images ModalitiesMedical Images ModalitiesMedical Images Modalities
Spin Density and Relaxation Times: Spin Density and Relaxation Times: –– Based on Magnetic Resonance Properties.Based on Magnetic Resonance Properties.–– Properties of Proton (HProperties of Proton (H++) spin are imaged.) spin are imaged.
Multichannel images:Multichannel images:–– Multichannel images:Multichannel images:PD (Proton Density)PD (Proton Density)T1: SpinT1: Spin--Lattice Relaxation Time.Lattice Relaxation Time.T2: SpinT2: Spin--Spin Relaxation Time.Spin Relaxation Time.
D t A i iti i t iD t A i iti i t i–– Data Acquisition is parametric:Data Acquisition is parametric:Several Protocols for imaging are possible.Several Protocols for imaging are possible.
–– Projection and Tomography are possible.Projection and Tomography are possible.–– Resolution: GoodResolution: Good–– SNR: GoodSNR: Good–– Hazard: Very Low (But banned for patients with Hazard: Very Low (But banned for patients with
ferromagnetic/Electrical/Magnetic Devices in their body)ferromagnetic/Electrical/Magnetic Devices in their body)–– High Contrast for soft tissue and Low for Hard tissue (bone)High Contrast for soft tissue and Low for Hard tissue (bone)
Sharif University of Technology Sharif University of Technology –– E. E. FatemizadehFatemizadeh
High Contrast for soft tissue and Low for Hard tissue (bone)High Contrast for soft tissue and Low for Hard tissue (bone)–– Static and Functional, both.Static and Functional, both.
Medical Images ModalitiesMedical Images ModalitiesMedical Images ModalitiesMedical Images Modalities
Spin Density and Relaxation Times:Spin Density and Relaxation Times:–– Examples:Examples:Examples:Examples:
MRI: Magnetic Resonance Imaging, Brain MRI: Magnetic Resonance Imaging, Brain Studies, Spin cord, Knee.Studies, Spin cord, Knee.fMRI: Functional MRI, Blood flow, brain.fMRI: Functional MRI, Blood flow, brain.MRA: Magnetic Resonance Angiography, MRA: Magnetic Resonance Angiography, Vessel StudiesVessel StudiesVessel Studies.Vessel Studies.
Sharif University of Technology Sharif University of Technology –– E. E. FatemizadehFatemizadeh
Medical Images ModalitiesMedical Images ModalitiesMedical Images ModalitiesMedical Images Modalities
Optical:Optical:–– Optical ReflectionOptical Reflection
H d N (P i U f i )H d N (P i U f i )–– Hazard: None (Patient Unconformity)Hazard: None (Patient Unconformity)–– Resolution: HighResolution: High–– SNR: HighSNR: HighSNR: HighSNR: High–– Examples:Examples:
EndoscopyEndoscopyLLLaryngoscopyLaryngoscopyColonoscopyColonoscopy
–– Optical Tomography found in research filesOptical Tomography found in research files
Sharif University of Technology Sharif University of Technology –– E. E. FatemizadehFatemizadeh
Medical Images ModalitiesMedical Images ModalitiesMedical Images ModalitiesMedical Images Modalities
Electrical Conductance:Electrical Conductance:–– Electrical Impedance Tomography (EIT)Electrical Impedance Tomography (EIT)–– Electrical Conductance (Resistance) is imaged.Electrical Conductance (Resistance) is imaged.–– Low ResolutionLow Resolution–– Low SNRLow SNR–– Hazard: Electrical Safety Problem.Hazard: Electrical Safety Problem.
Low PriceLow Price–– Low PriceLow Price–– TomographyTomography
Sharif University of Technology Sharif University of Technology –– E. E. FatemizadehFatemizadeh
Return to Medical Return to Medical ImagingImaging
Concepts:Concepts:–– How to build images of internal organs ofHow to build images of internal organs ofHow to build images of internal organs of How to build images of internal organs of
body, nonbody, non--invasively.invasively.–– Image ModalitiesImage Modalitiesgg–– PrePre--processingprocessing–– PostPost--ProcessingProcessingPostPost ProcessingProcessing
Image ConstructionImage ConstructionImage ConstructionImage Construction
Goal:Goal:–– Draw images of a certain physicalDraw images of a certain physicalDraw images of a certain physical Draw images of a certain physical
property of subject anatomy.property of subject anatomy.–– Procedure in nonProcedure in non--invasive.invasive.–– Image Geometry:Image Geometry:
ProjectionProjectionjjTomographyTomography
Image ModalityImage ModalityImage ModalityImage Modality
Based on Interested Physical Property:Based on Interested Physical Property:–– XX--Ray (CT/Radiography)Ray (CT/Radiography)–– MRI (Magnetic Resonance Imaging)MRI (Magnetic Resonance Imaging)–– PET (Positron Emission Tomography)PET (Positron Emission Tomography)–– US (Ultra Sound)US (Ultra Sound)–– SPECT (Single Photon Emission CT)SPECT (Single Photon Emission CT)–– EIT (Electrical Impedance Tomography)EIT (Electrical Impedance Tomography)–– Video and etc.Video and etc.
PrePre ProcessingProcessingPrePre--ProcessingProcessing
Concepts:Concepts:–– Design optimum protocol for raw dataDesign optimum protocol for raw dataDesign optimum protocol for raw data Design optimum protocol for raw data
acquisition.acquisition.–– Image reconstruction from raw data.Image reconstruction from raw data.gg–– Noise and artifact reduction in raw data Noise and artifact reduction in raw data
space.space.pp
PostPost ProcessingProcessingPostPost--ProcessingProcessing
Concepts:Concepts:–– Noise and artifact reduction in imageNoise and artifact reduction in imageNoise and artifact reduction in image Noise and artifact reduction in image
space.space.–– Enhance images in Regions of Interest.Enhance images in Regions of Interest.g gg g–– Image partitioning to meaningful regions.Image partitioning to meaningful regions.–– Computer Aided Diagnosis (CAD)Computer Aided Diagnosis (CAD)Computer Aided Diagnosis (CAD)Computer Aided Diagnosis (CAD)–– Multimodality Image FusionMultimodality Image Fusion–– Virtual Reality (Virtual Surgery)Virtual Reality (Virtual Surgery)Virtual Reality (Virtual Surgery)Virtual Reality (Virtual Surgery)