Machine Learning for
Medical Image Analysis:
What, where and how?
Dr. Debdoot Sheet
Assistant Professor
Department of Electrical Engineering
Indian Institute of Technology Kharagpur
www.facweb.iitkgp.ernet.in/~debdoot/
A GREAT PIECE OF CAREER ADVICE
FOR EECS GRADUATES
INTERESTED IN MACHINE VISION
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 2
Market Scenario and Career
Media,
surveillance,
automotive,
graphics, etc.
($ 6 Billion)
63%
Medical Image
Analysis
($ 3.5 Billion)
37%
Market for Machine Vision
Systems in 2020
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 3
Modality
• X-ray
• Ultrasound
• Computed Tomography (CT)
• Magnetic Resonance (MRI)
• Nuclear Imaging (PET & SPECT)
Clinical Indications
• Radiology
• Cardiology
• Oncology
• Neurology
• Obstetrics & gynecology
• Breast mammography
End Users
• Hospitals
• Diagnostic centers
• Research centers
Report code: HIT 1309 and SE 2701 from www.marketsandmarkets.com
Medical Image Analysis (MedIA)?
• Is not Medical Imaging
• Is not Medicine
• Is not Image Processing
• Is not Computer Vision
• Is not Machine Learning
• Is not Software Engineering
• Is not Project Management
• Is not Creating Enterprises
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 4
Overview of MedIA
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 5
Last 35 years of MedIA
• Pre 1980 – 1984: Era
of Pattern Recognition
Analysis of 2D Images
• 1985 – 1991:
Knowledge based
Approaches
• 1992 – 1998: 3D
Images and Towards
Integrated Analysis
• 1999 – 2010: Machine
Learning with Shallow
Reasoning
• 2010 and Beyond:
Machine Learning
with Complex
Reasoning
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 6
Duncan, J.S.; Ayache, N., "Medical image
analysis: progress over two decades and the
challenges ahead," IEEE Trans. Pat. Anal., Mach.
Intell., vol.22, no.1, pp.85,106, Jan. 2000
Key Research (Business) Areas
• Macro - meso - micro imaging
• Medical image registration
• Organ localization
• Organ segmentation
• Medical visualization and
Augmented reality
• Virtual anatomy
• Digital angiography
• Optical and ultrasonic despeckling
• 3D optical microscopy
• Digital pathology
• Computational imaging
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 7
End use of MedIA
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 8
Tong, S.; Sheet, D.;
Bhuiyan, S.; Zequer
Diaz, M.; Taberne, A.,
"BME Trends Around
the World : From
Baby X to frugal
technologies, here's
what biomedical
engineers are
excited about in
2015. [From the
Editors]," IEEE
Pulse, vol.6, no.1,
pp.4-6, Jan.-Feb.
2015
MedIA Challenges in 2015
• Endoscopic vision for
Computer Assisted
Intervention (CAI)
• Statistical shape
modeling
• Liver ultrasound
tracking
• Gland segmentation in
histology
• Retinal cyst
segmentation
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 9
MedIA Challenges in 2015
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 10
CAREER ADVICES FOR MedIA
PHD ASPIRANTS
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 11
Find a (Research) Challenge
• Grand Challenges in Biomedical Image Analysis
– www.grand-challenges.org
• MICCAI
– www.miccai.org
– www.miccai2015.org
• ISBI
– www. biomedicalimaging.org
• SPIE Medical Imaging
– 2015 will host a grand-challenge
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 12
Tool(boxes) of the Trade
• BioDigital Human
– www.biodigital.com
• MeVisLab
– Medical image processing, analysis and visualization SDK
– http://www.mevislab.de/
• Python (x,y)
– Python 2.7 with scientific computing library for custom building tools
– https://code.google.com/p/pythonxy/
• Tortoise SVN
– Version control for managing large file software projects
– www.tortoisesvn.net/
• MicroDicomViewer
– Standalone Viewer for Dicom files
– www.microdicom.com/downloads.html
• CUDA
– Library for using NVIDIA GPUs
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 13
Where to Read for MedIA?
Specialist Literature
• Medical Image Analysis
• IEEE Trans. Medical Imaging
• IEEE Trans. Computational
Imaging
• IEEE J. Biomedical and
Health Informatics
• SPIE J. Medical Imaging
• Computerized Medical
Imaging and Graphics
General Reading
• IEEE Trans. Image
Processing
• IEEE Trans. Biomedical
Engineering
• Int. J. Computer Vision
• IET Image Processing
• IET Computer Vision
• ACM Trans. Graphics
• Proc. Int. Conf. Image. Proc.
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 14
Where to Read for ML?
Journal
• IEEE Trans. Pattern Analysis
and Machine Intelligence
• Machine Learning
• J. Machine Learning
Research
• IEEE Trans. Knowledge and
Data Engineering
• IEEE Trans. Neural Networks
• IEEE Trans. Sys. Man. Cyber.
Conferences
• Computer Vision and
Pattern Recognition (CVPR)
• Machine Learning confs.
– International (ICML)
– European (ECML)
– Asian (ACML)
• Neural Information
Processing System (NIPS)
• Computer Vision conf.
– International (ICCV)
– European (ECCV)
– Asian (ACCV)
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 15
Where to Listen and Socialize?
Workshops and Schools
• Medical Imaging Summer
School (MISS)
– MICCAI Society
• IEEE SPS Summer School on
Biomedical Image Analysis
– IEEE Signal Processing Soc.
• IEEE EMBS Int. Summer School
on Biomedical Imaging
– IEEE Engineering in Medicine
and Biology Society
• Biomedical Image Analysis
Summer School
Conferences
• Medical Image Computing and
Computer Assisted Intervention
(MICCAI)
• Int. Symp. Biomed. Imaging
(ISBI)
• SPIE Med. Imaging
• Information Processing in
Medical Imaging (IPMI)
• Information Processing in
Computer-Assisted Interventions
(IPCAI)
• Microscopy Conference (MC)
• Int. Conf. Sys., Med. Biol. (ICSMB)
– In IIT Kharagpur, Jan 2016
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 16
Some of the (BIG Group) Leaders
• Alejandro Frangi
(Sheffield)
• Alison Noble (Oxford)
• Badrinath Roysam
(Houston)
• Ben Glocker (Imperial)
• Bram van Ginneken
(Nijmegen)
• Gabor Fichtinger
(Queens)
• James Duncan (Yale)
• Michael Unser (EPFL)
• Michael Liebling (UCSB)
• Milan Sonka (Iowa)
• Nassir Navab (TUM+JHU)
• Nicholas Ayache (INRIA
Sophia Antipolis Cedex)
• Pierre Jannin (INRIA
Rennes)
• Polina Golland (MIT)
• Terry Peters (Robarts)
• Wiro Nissen (Erasmus
MC)
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 17
Biomedical Imaging
and Graphics
Where to Publish?
Conferences (Early Track)
• MICCAI
• ISBI
• SPIE Medical Imaging
• IPCAI
• IPMI
• CVPR
• ICCV / ECCV / ACCV
• ICIP
Journal (Mature)
• Medical Image Analysis
• IEEE Trans. Med. Imaging
• IEEE J. Biomed. Health Inf.
• IEEE Trans. Biomed. Engg.
• Comp. Med. Imag. Graph.
• SPIE Med. Imaging
• BMC Medical Imaging
• Int. J. Biomed. Imaging
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 18
Where to Look for (Creating) Jobs?
Incubation, Startup Grants
• Technology Business Incubation
(TBI) – DSIR
– Promoting Innovations in
Individuals, Start-ups and MSMEs
(PRISM)
• Cat I – Rs. 5 lcs (Idea pitchup)
• Cat II – Rs. 5-35 lcs (Tech.
Demonstrator)
• Biotechnology Industry Research
Assistance Council (BIRAC)
– Biotechnology Ignition Grant (BIG)
• Rs. 50 lcs (Tech. Demonstrator)
• Indo-US Science and Technology
Forum (IUSSTF)
– India+US Tech. Starter Endowment
• $ 400,000 (Rs. 2.5 Cr)
Investments
• Mix of Loan + Grant-in-Aid
– BIRAC + Department of
Electronics and Information
Technology (DEITY)
• Success of BIG
• 50% Grant to University + 50%
loan to Start-up
• Rs. 3 Cr. (Tech.
Commercialization)
• Angels and Investors
– India Angels Network
– Khosla Ventures
– Tata Ventures
– The Indus Entrepreneurs (TiE)
– Corporate Buy-outs
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 19
Take home message
Some Texts
• Toennies, Guide to medical
image analysis, 2012.
• Bankman, Handbook of Medical
Image Processing and Analysis,
2012.
• Dougherty, Medical Image
Processing techniques and
applications, 2013.
• M. N. Gurcan, "Histopathological
Image Analysis: A Review", IEEE
Rev. Biomed. Engg., vol. 2, pp.
147-171, 2009.
• J. A. Noble, “Ultrasound Image
Segmentation: A Survey”, IEEE
Trans. Med. Imaging, vol. 25, no.
8, pp. 987-1010, 2006.
MedIA in Classroom
• IIT Kharagpur:
http://www.bit.do/eemia
• IIT Bombay:
http://www.cse.iitb.ac.in/~su
yash/cs736/
• Carnegie Mellon:
http://www.cs.cmu.edu/~gal
eotti/methods_course/
• TU Munich:
http://campar.in.tum.de/Cha
ir/TeachingWiSe2014CAMP
ONE
24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 20

Machine Learning for Medical Image Analysis: What, where and how?

  • 1.
    Machine Learning for MedicalImage Analysis: What, where and how? Dr. Debdoot Sheet Assistant Professor Department of Electrical Engineering Indian Institute of Technology Kharagpur www.facweb.iitkgp.ernet.in/~debdoot/
  • 2.
    A GREAT PIECEOF CAREER ADVICE FOR EECS GRADUATES INTERESTED IN MACHINE VISION 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 2
  • 3.
    Market Scenario andCareer Media, surveillance, automotive, graphics, etc. ($ 6 Billion) 63% Medical Image Analysis ($ 3.5 Billion) 37% Market for Machine Vision Systems in 2020 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 3 Modality • X-ray • Ultrasound • Computed Tomography (CT) • Magnetic Resonance (MRI) • Nuclear Imaging (PET & SPECT) Clinical Indications • Radiology • Cardiology • Oncology • Neurology • Obstetrics & gynecology • Breast mammography End Users • Hospitals • Diagnostic centers • Research centers Report code: HIT 1309 and SE 2701 from www.marketsandmarkets.com
  • 4.
    Medical Image Analysis(MedIA)? • Is not Medical Imaging • Is not Medicine • Is not Image Processing • Is not Computer Vision • Is not Machine Learning • Is not Software Engineering • Is not Project Management • Is not Creating Enterprises 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 4
  • 5.
    Overview of MedIA 24June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 5
  • 6.
    Last 35 yearsof MedIA • Pre 1980 – 1984: Era of Pattern Recognition Analysis of 2D Images • 1985 – 1991: Knowledge based Approaches • 1992 – 1998: 3D Images and Towards Integrated Analysis • 1999 – 2010: Machine Learning with Shallow Reasoning • 2010 and Beyond: Machine Learning with Complex Reasoning 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 6 Duncan, J.S.; Ayache, N., "Medical image analysis: progress over two decades and the challenges ahead," IEEE Trans. Pat. Anal., Mach. Intell., vol.22, no.1, pp.85,106, Jan. 2000
  • 7.
    Key Research (Business)Areas • Macro - meso - micro imaging • Medical image registration • Organ localization • Organ segmentation • Medical visualization and Augmented reality • Virtual anatomy • Digital angiography • Optical and ultrasonic despeckling • 3D optical microscopy • Digital pathology • Computational imaging 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 7
  • 8.
    End use ofMedIA 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 8 Tong, S.; Sheet, D.; Bhuiyan, S.; Zequer Diaz, M.; Taberne, A., "BME Trends Around the World : From Baby X to frugal technologies, here's what biomedical engineers are excited about in 2015. [From the Editors]," IEEE Pulse, vol.6, no.1, pp.4-6, Jan.-Feb. 2015
  • 9.
    MedIA Challenges in2015 • Endoscopic vision for Computer Assisted Intervention (CAI) • Statistical shape modeling • Liver ultrasound tracking • Gland segmentation in histology • Retinal cyst segmentation 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 9
  • 10.
    MedIA Challenges in2015 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 10
  • 11.
    CAREER ADVICES FORMedIA PHD ASPIRANTS 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 11
  • 12.
    Find a (Research)Challenge • Grand Challenges in Biomedical Image Analysis – www.grand-challenges.org • MICCAI – www.miccai.org – www.miccai2015.org • ISBI – www. biomedicalimaging.org • SPIE Medical Imaging – 2015 will host a grand-challenge 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 12
  • 13.
    Tool(boxes) of theTrade • BioDigital Human – www.biodigital.com • MeVisLab – Medical image processing, analysis and visualization SDK – http://www.mevislab.de/ • Python (x,y) – Python 2.7 with scientific computing library for custom building tools – https://code.google.com/p/pythonxy/ • Tortoise SVN – Version control for managing large file software projects – www.tortoisesvn.net/ • MicroDicomViewer – Standalone Viewer for Dicom files – www.microdicom.com/downloads.html • CUDA – Library for using NVIDIA GPUs 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 13
  • 14.
    Where to Readfor MedIA? Specialist Literature • Medical Image Analysis • IEEE Trans. Medical Imaging • IEEE Trans. Computational Imaging • IEEE J. Biomedical and Health Informatics • SPIE J. Medical Imaging • Computerized Medical Imaging and Graphics General Reading • IEEE Trans. Image Processing • IEEE Trans. Biomedical Engineering • Int. J. Computer Vision • IET Image Processing • IET Computer Vision • ACM Trans. Graphics • Proc. Int. Conf. Image. Proc. 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 14
  • 15.
    Where to Readfor ML? Journal • IEEE Trans. Pattern Analysis and Machine Intelligence • Machine Learning • J. Machine Learning Research • IEEE Trans. Knowledge and Data Engineering • IEEE Trans. Neural Networks • IEEE Trans. Sys. Man. Cyber. Conferences • Computer Vision and Pattern Recognition (CVPR) • Machine Learning confs. – International (ICML) – European (ECML) – Asian (ACML) • Neural Information Processing System (NIPS) • Computer Vision conf. – International (ICCV) – European (ECCV) – Asian (ACCV) 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 15
  • 16.
    Where to Listenand Socialize? Workshops and Schools • Medical Imaging Summer School (MISS) – MICCAI Society • IEEE SPS Summer School on Biomedical Image Analysis – IEEE Signal Processing Soc. • IEEE EMBS Int. Summer School on Biomedical Imaging – IEEE Engineering in Medicine and Biology Society • Biomedical Image Analysis Summer School Conferences • Medical Image Computing and Computer Assisted Intervention (MICCAI) • Int. Symp. Biomed. Imaging (ISBI) • SPIE Med. Imaging • Information Processing in Medical Imaging (IPMI) • Information Processing in Computer-Assisted Interventions (IPCAI) • Microscopy Conference (MC) • Int. Conf. Sys., Med. Biol. (ICSMB) – In IIT Kharagpur, Jan 2016 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 16
  • 17.
    Some of the(BIG Group) Leaders • Alejandro Frangi (Sheffield) • Alison Noble (Oxford) • Badrinath Roysam (Houston) • Ben Glocker (Imperial) • Bram van Ginneken (Nijmegen) • Gabor Fichtinger (Queens) • James Duncan (Yale) • Michael Unser (EPFL) • Michael Liebling (UCSB) • Milan Sonka (Iowa) • Nassir Navab (TUM+JHU) • Nicholas Ayache (INRIA Sophia Antipolis Cedex) • Pierre Jannin (INRIA Rennes) • Polina Golland (MIT) • Terry Peters (Robarts) • Wiro Nissen (Erasmus MC) 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 17 Biomedical Imaging and Graphics
  • 18.
    Where to Publish? Conferences(Early Track) • MICCAI • ISBI • SPIE Medical Imaging • IPCAI • IPMI • CVPR • ICCV / ECCV / ACCV • ICIP Journal (Mature) • Medical Image Analysis • IEEE Trans. Med. Imaging • IEEE J. Biomed. Health Inf. • IEEE Trans. Biomed. Engg. • Comp. Med. Imag. Graph. • SPIE Med. Imaging • BMC Medical Imaging • Int. J. Biomed. Imaging 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 18
  • 19.
    Where to Lookfor (Creating) Jobs? Incubation, Startup Grants • Technology Business Incubation (TBI) – DSIR – Promoting Innovations in Individuals, Start-ups and MSMEs (PRISM) • Cat I – Rs. 5 lcs (Idea pitchup) • Cat II – Rs. 5-35 lcs (Tech. Demonstrator) • Biotechnology Industry Research Assistance Council (BIRAC) – Biotechnology Ignition Grant (BIG) • Rs. 50 lcs (Tech. Demonstrator) • Indo-US Science and Technology Forum (IUSSTF) – India+US Tech. Starter Endowment • $ 400,000 (Rs. 2.5 Cr) Investments • Mix of Loan + Grant-in-Aid – BIRAC + Department of Electronics and Information Technology (DEITY) • Success of BIG • 50% Grant to University + 50% loan to Start-up • Rs. 3 Cr. (Tech. Commercialization) • Angels and Investors – India Angels Network – Khosla Ventures – Tata Ventures – The Indus Entrepreneurs (TiE) – Corporate Buy-outs 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 19
  • 20.
    Take home message SomeTexts • Toennies, Guide to medical image analysis, 2012. • Bankman, Handbook of Medical Image Processing and Analysis, 2012. • Dougherty, Medical Image Processing techniques and applications, 2013. • M. N. Gurcan, "Histopathological Image Analysis: A Review", IEEE Rev. Biomed. Engg., vol. 2, pp. 147-171, 2009. • J. A. Noble, “Ultrasound Image Segmentation: A Survey”, IEEE Trans. Med. Imaging, vol. 25, no. 8, pp. 987-1010, 2006. MedIA in Classroom • IIT Kharagpur: http://www.bit.do/eemia • IIT Bombay: http://www.cse.iitb.ac.in/~su yash/cs736/ • Carnegie Mellon: http://www.cs.cmu.edu/~gal eotti/methods_course/ • TU Munich: http://campar.in.tum.de/Cha ir/TeachingWiSe2014CAMP ONE 24 June 2015 Intro to Machine Learning for Medical Image Analysis [Debdoot Sheet] - WMLMIA 20

Editor's Notes

  • #6 This slide had contents, the animations are hiding it. Play to see.