signal and image processing for medical...

C O N F I D E N T I A L

©20

04 T

he M

athW

orks

, Inc

.

http://www.terasoft.com.tw/

Signal and Image Processing for Medical Applications

Nov 12th ,2004

Fusung Wang

©20

04 T

he M

athW

orks

, Inc

.


Today’s Agenda

• MATLAB in Biomedical Research

• The MATLAB Development and Deployment Platform

• Medical Image Processing

• Signal Processing for Medical Applications

• Wrap Up, Questions and Answers

©20

04 T

he M

athW

orks

, Inc

.


MathWorks Mission and Vision Accelerate innovation and discovery in engineering and science

MATLABA powerful, high-level language to develop algorithms, collect and analyze data, and visualize information

SimulinkA graphical system to model and simulate complex systems, and implement real-time and embedded systems

©20

04 T

he M

athW

orks

, Inc

.


MathWorks Products Drive Innovation in a Broad Range of Industries

Aerospace and DefenseAutomotiveBiotech, Pharmaceutical, and MedicalCommunications and SemiconductorEducationFinancial ServicesIndustrial Equipment and MachineryInstrumentationMedical Devices and Instrumentation

©20

04 T

he M

athW

orks

, Inc

.


3,500 universities teach students and conduct research using MathWorks products.More than 600 textbooks have been published for education and professional use, in 19 languages.

BiosciencesControlsSignal ProcessingImage Processing

Mechanical EngineeringMathematicsNatural SciencesEnvironmental Sciences

Wide Product Adoption Throughout Academia

©20

04 T

he M

athW

orks

, Inc

.


The MathWorks Product Family

Code Generation

Blocksets

Integrated for:Technical computing, data analysis, and visualizationSystem modeling and simulationImplementation of real-time embedded software

PC-Based Real-Time Systems

StateflowStateflowStateflowToolboxes

Data and Image Acquisition Cards

Instruments

Databases and Files

Desktop ApplicationsAutomated Reports

©20

04 T

he M

athW

orks

, Inc

.


Toolboxes

The MathWorks Product FamilyToolboxes for modeling, analysis, and computation

Specific functionality for data analysis, modeling, design, and other capabilities

• Bioinformatics• Statistics• Image Processing

• Neural Networks• Optimization• Compiler

• Symbolic Math• Curve Fitting • Filter Design

• Wavelet• Signal Processing• Fuzzy Logic

©20

04 T

he M

athW

orks

, Inc

.


Applications in Medical Imaging

FunctionalImaging

Microscopy

Visualization and 3D Reconstruction

Gel andMicroarray

Images

Cytology and

PathologyRadiological

Imaging

MultidimensionalImage Processing

gel image : Alan W. Partin, M.D., Ph.D., Johns Hopkins University School of Medicine

cancer cell image: Alan W. Partin, M.D., Ph.D., Johns Hopkins University School of Medicine

pathology image: Angelo M. DeMarzo, M.D. Ph.D., Johns Hopkins University School of Medicine

©20

04 T

he M

athW

orks

, Inc

.


Microscopy

Automatic segmentationAutomatic countingMorphometryMotilityDeblurring

Quantify number, area, color intensity, and shape of objects

©20

04 T

he M

athW

orks

, Inc

.


Radiology and Functional Imaging

SegmentationRegistrationImage enhancementSignal reconstructionDICOMStatistical analysisCo-registrationMotion correctionSpatial normalization

©20

04 T

he M

athW

orks

, Inc

.


Signal Processing in the Medical Industry

ECG analysis

Real-time biosignal amplifier

Analysis of PSG recordings

Real-time processing of EEG, EMG, ECG, and EOG

Hearing aid design

©20

04 T

he M

athW

orks

, Inc

.


ECG Analysis

Real-time data acquisitionEasy filter design and implementationDiverse automatic signal processing optionsSeveral signal visualization options

Acquire, filter, process, and visualize ECG signal

©20

04 T

he M

athW

orks

, Inc

.


Hearing Aid Design

Design components of system to be integratedTest and tune parameters on the flyCreate real-time test bench for rapid prototypingAutomatically generate code Test generated code on target

Design, tune, and implement model

©20

04 T

he M

athW

orks

, Inc

.


Customer Solutions: Institute for BiodiagnosticsNoninvasively Assessing Burn Injuries

The ChallengeDevelop technique to evaluate burns soon after injury to prevent irreversible tissue damage

The SolutionUsed NIR spectrometer and multiband CCD imager to measure hemoglobin levels and blood volume in shallow and deep epidermis tissueUsed MATLAB, Signal Processing Toolbox, Wavelet Toolbox, and Image Processing Toolbox to conduct burn simulations

The ResultsNon-invasive device to measure burn severity

Efficient test environment to support clinical trials and pursue FDA approval for new device

©20

04 T

he M

athW

orks

, Inc

.


What can you do with MATLAB?Explore your data.

Perform matrix computations (linear algebra).

Visualize complex data to gain insight.

Develop algorithms to solve problems.

Build custom data analysis tools with GUIs.

Understand how functions and algorithms work.

Integrate MATLAB with other software.

Develop models to simulate sensors, systems, and processes, especially with Simulink.

Share programs, data, and results with colleagues.

©20

04 T

he M

athW

orks

, Inc

.


MATLAB Java Interfacing

Access to the rich Java GUI building, database, and Web tools from the familiar MATLAB development platform.

Call Java directly from MATLAB (via interpreter).Embed Java routines and objects in MATLAB applications.Exchange data between MATLAB and Java.Access Java API class packages and third-party Java classes.

©20

04 T

he M

athW

orks

, Inc

.


Customer Solutions: Given Imaging

The ChallengeDevelop a minimally invasive diagnostic tool that substantially improves visual imaging of the small intestine

The SolutionUsed MathWorks tools for the feasibility study, development, and refinement phases of the image processing project

The ResultsFast FDA approval and time-to-market

©20

04 T

he M

athW

orks

, Inc

.


Customer Solutions: Beth Israel Deaconess

The ChallengeDevelop a safe and effective way to improve the resolution and speed of MR scans

The SolutionUse the Image Processing Toolbox and other MathWorks products to develop a technique for accessing multiple image components simultaneously

The ResultsFaster, more informative MR scansAbility to experiment with new approachesReduced programming time

©20

04 T

he M

athW

orks

, Inc

.


MATLAB supports Multi-modality Imaging

MRICTPETUltrasoundX-Ray…

MATLAB can incorporate images from several different data sources

©20

04 T

he M

athW

orks

, Inc

.


Image visualization, analysis, processing, and algorithm development

Image Processing Toolbox

Key Features:Interactive image viewerImage analysis functions, such as edge detection, feature measurement, and moreImage processing functions, such as enhancement, segmentation, morphology, and registrationSupport for numerous image and scientific file formats, including DICOM

©20

04 T

he M

athW

orks

, Inc

.


Pixel values and statistics ProfilesIntensity plots

Background removalSegmentationObject labelingFeature measurements

Image Processing Toolbox FunctionsEdge detectionMorphologySegmentation

Image deblurring

©20

04 T

he M

athW

orks

, Inc

.


Image Acquisition Toolbox

Key Features:Image and video streaming directly into MATLABHardware detection for supported video and image acquisition devices Device property configuration Background image acquisitionLive video previewingSupport for multiple hardware vendors

©20

04 T

he M

athW

orks

, Inc

.


Image Acquisition Toolbox

Supported Image Capture VendorsMatroxData Translation

Supported Windows Video DevicesUSB/FireWire cameras, DV camcorders, image capture boards, TV tuner cards, etc.Requires:

Windows Driver Model (WDM) or Video For Windows (VFW) driverMicrosoft DirectX 9.0 or later

Visit http://www.mathworks.com/products/imaq/ for more information

©20

04 T

he M

athW

orks

, Inc

.


Neuroimaging Case Study

Measure volume of white and gray matter in an MRI brain scan

©20

04 T

he M

athW

orks

, Inc

.


What this Case Study Shows

Data AccessMRI scan = set of DICOM files

Visualization2D slices3D volume

AnalysisSegmentation

Thresholding grayscale intensitiesCropping spatial dimensions3D morphological refinement

Volumetric measurements (density)

©20

04 T

he M

athW

orks

, Inc

.


Data Access

Patient examination conducted using MRI scanner3D volumetric region of head surrounding brainCoronal slices acquired as individual DICOM files

Parse DICOM filesIdentify image series associated with patient study

Read slice imagesPopulate XYZ matrix

Keep track of voxel sizeaspect ratio

©20

04 T

he M

athW

orks

, Inc

.


Demonstration

Enough with the slides, already … let’s see

some MATLAB!

©20

04 T

he M

athW

orks

, Inc

.


DICOM Support (acquired image files)

©20

04 T

he M

athW

orks

, Inc

.


Visualization

MATLAB Handle Graphics® engine Display complex plots, images and surfacesManipulate display properties (color, texture, lighting)

Interactive mouse-based controls (camera tools)MATLAB commands allow programmatic control

OpenGL supportHardware acceleration provides fast redrawReduces CPU burden

©20

04 T

he M

athW

orks

, Inc

.


2D Visualization

Explore image data [imview]

©20

04 T

he M

athW

orks

, Inc

.


2D VisualizationDisplay images [imshow] and intensity distributions [hist].

©20

04 T

he M

athW

orks

, Inc

.


3D Visualization

Display volume cross sections [slice,contourslice]Display surface of interior volume [isosurface,isocap]

©20

04 T

he M

athW

orks

, Inc

.


Custom Data AnalysisDownload sliceomatic from the MATLAB Central file exchange.

©20

04 T

he M

athW

orks

, Inc

.


Segmentation

Threshold by range of intensitiesBackground air, CSF & other (soft?) tissues too darkSkull & other (hard?) tissues too brightCustom GUI to aid parameter tuning (thresh_tool)

Crop by spatial dimensionsBrain mass not too low in head

Trim off small, residual fragmentsMorphological opening = erosion + dilation [imopen]Connected region size [bwlabel, regionprops, .Area]

Separate brain mass (white vs. gray matter) Use thresh_tool to help select proper threshold levelUse sliceomatic to display 3D results

©20

04 T

he M

athW

orks

, Inc

.


Outside of the Head

©20

04 T

he M

athW

orks

, Inc

.


Volumetric Measurements

scan_density = head_voxels / total_voxels

brain_density = brain_voxels / head_voxels

gray_fraction = gray_voxels / brain_voxelswhite_fraction = white_voxels / brain_voxels

©20

04 T

he M

athW

orks

, Inc

.


What this Case Study Showed

With MATLAB you can:Work with DICOM files to facilitate patient examinations.Display 2D slices as images – separately or in XYZ space.Display 3D surfaces containing interior volume with realistic rendering.Perform simple to complex image processing on 2D, 3D or N-D data.Segment features of interest from image background and other objects.Quantitatively measure blob sizes to volume densities.Build custom data analysis tools to solve scientific imaging problems.

©20

04 T

he M

athW

orks

, Inc

.


Why Use MATLAB for Medical Imaging?

Verified and trusted algorithms for a wide range of image processing applicationsIdeal for exploration and learningMore flexible and customizable than “point and click” softwareSaves time from developing and maintaining standard image processing algorithmsEnables you to present and share results graphically, in reportsand on the WebMulti-modality imaging capabilities

©20

04 T

he M

athW

orks

, Inc

.


Customer Solutions: Medrad

The ChallengeDesign an MRI vascular injection pump that dministers safe levels of pharmaceuticals to patients

The SolutionUse MathWorks tools to improve the pump’s pressure sensing technology

The ResultDesign time reduced by several monthsPrestigious industry award receivedFDA approval achieved

The Medrad Spectris Solaris contrast injection system

©20

04 T

he M

athW

orks

, Inc

.


Signal Processing Workflow

DataAnalysis &

Visualization

Data Acq& Import

AlgorithmDesign &Analysis

MathematicalModeling Embedded

SoftwareEmbeddedSoftware

Verification,HIL Test

Verification,HIL TestRapid

PrototypingRapid

Prototyping

CodeGeneration

CodeGeneration

EmbeddedHardware

EmbeddedHardware

SystemModeling,

Simulation andPartitioning

SystemModeling,


Systemcomponents

Environmenteffects

Embeddedalgorithms IM

PLEMEN

T

RESEARCH DESIGN TEST

©20

04 T

he M

athW

orks

, Inc

.


DSP, Comms,

Fixed Point Blocksets

Simulink

Simulink: Model-Based Design

MATLAB

Signal Processing

&Filter Design

Toolboxes

DataAnalysis &

Visualization

Data Acq& Import

AlgorithmDesign &Analysis

MathematicalModeling Embedded

SoftwareEmbeddedSoftware

Verification,HIL Test

Verification,HIL TestRapid

PrototypingRapid

Prototyping

CodeGeneration

CodeGeneration

EmbeddedHardware

EmbeddedHardware

SystemModeling,


SystemModeling,


Systemcomponents

Environmenteffects

Embeddedalgorithms

RESEARCH DESIGN

IMPLEM

ENT

TEST

©20

04 T

he M

athW

orks

, Inc

.


Over Half of All Embedded Software Projects are Completed Behind Schedule

On Schedule6.7%

Completed Behind Schedule

51.6%

Canceled19.0%

Ahead of Schedule

22.7%

Source: Software Development Times, December 2002Venture Development Corp. data

©20

04 T

he M

athW

orks

, Inc

.


MATLAB Connections

More than 300 add-on products and services from partners that complement and extend MathWorks products

Specialized third-party toolboxes for MATLABInterfaces to partners’ software and hardware productsSpecialized training courses and consulting servicesTurnkey systems providers that incorporate MathWorks products

©20

04 T

he M

athW

orks

, Inc

.


Your Obstacles and Our Solutions

Performing advanced and diverse analysis or processing requires several different toolsA lot of time is lost coding in a low-level language instead of focusing on solving the problem

Prototyping and testing takes too long

MATLAB provides one environment for all types of analysis (signal, image, statistics, wavelets, etc.)MATLAB’s high-level language allows you to focus on the problem solving with the advanced low-level implementation already done for youMATLAB’s capabilities allow you to create a test bench which allows you to test, tune, and implement your design quickly

©20

04 T

he M

athW

orks

, Inc

.


From a leading textbook author ...

"I have used a number of commercial image processing packages over the years, and prefer the (MathWorks Image Processing) toolbox for several reasons: the wide variety of functions it provides, the user’s ability to write additional functions with minimal effort, the quality of the software, and the high level of support."

- Rafael C. GonzalezUniversity of Tennessee

©20

04 T

he M

athW

orks

, Inc

.



Released in December 2003:

“Digital Image Processing Using

MATLAB”

by Gonzales, Woods, and Eddins

©20

04 T

he M

athW

orks

, Inc

.



Released in January 2004:

“Biosignal and Biomedical Image Processing

MATLAB-Based Applications”

by John L. Semmlow

©20

04 T

he M

athW

orks

, Inc

.


Further Information

Product Information and Demos Trials and technical literature are available at www.mathworks.com

MATLAB CentralFile exchange and newsgroup access for MATLAB and Simulink userswww.mathworks.com/matlabcentralAccess to comp.soft-sys.matlab

file exchange and newsgroup access forthe MATLAB & Simulink user community

signal and image processing for medical...

Documents