Open Source Radiology David Kelton, MD Feb 13, 2006
Outline My interest in OS Research Projects: 1) Survey OS in radiology 2) OS CAD at U of T
Background Interest “Thank goodness I’m a journalist and not a radiologist” Friedman, Thomas L. The World is Flat. New York: Farrar, Straus and Giroux 2005.
Overview of Imaging Infrastructure
Part 1: OS Opportunities in Rads Opportunities for OS in radiology: Ideal for standards DICOM (Digital Imaging and Communication in Medicine) Standard for handling, storing, and transmitting information File format definition and a network communications protocol IHE (Integrating the Healthcare Experience) Ideal for research tools Image analysis Teaching tools
Part1: Survey of OS Rads Community Dynamic sources Sourceforge.net Openrad.com DICOM SERVER: CTN DCMTK dcm4che Conquest MIND Dicomparser Dcmrouter Dicom3tools JDICOM Medwx MESA DICOM Viewer DicomWorks DICOMscope EViewBox AccuLite ezDICOM FP Image ImageJ Imread Irfanview32 Jivex MRIcro OSIRIS Sante Viewer Simple DICOM TomoVision XnView Amide NIH Image Scion Image OsiriX
Part 1: Survey of OS Rads Community Full PACS DIOWave CDIMEDIC Miniwebpacs Teaching Files myPACS.net MIRC SimpleMIRC Image Processing VTK ITK FSL AFNI SPM NeatMed ImageMagick XMedCon
Part 1: Barriers to OS in Radiology Barriers Identified to OS in Radiology: Erickson, Langer, Nagy. The Role of Open-Source Software in Innovation and Standardization in Radiology. JACR 2005;2: 927-931. 1) “White Hat” SCAR Conference (Project OS) 2) Active Community IT knowledge in physicians 3) “Red Hat” MedSphere Aycan
Part 1: Leading OS Rads Projects OsiriX viewer Increasing funding by Apple 8000+ users An emphasis on visual navigation (3D/4D/5D)
Part 1: Case Study OS in Radiology Beaumont Hospital in Ireland Est €13 million over 5 years Radiology specific: € 250,000 for OS RIS set-up Equivalent hospital in Ireland spent € 4.3 million commercial PACS Fitzgerald, B. Kenny, T. Lessons from a Large-Scale OSS Implementation. Twenty-fourth International Conference on Information Systems. Seattle,W ashington. 2003.
OS Radiology in Action
Part 2: OS CAD Computer-aided detection (CAD) is an emerging area of research and innovation in medical imaging to help cope with the huge increases in imaging data Number of CAD papers at RSNA (annual radiology meeting) 2000 – 55 2001 – 86 2002 – 134 2003 – 191 2004 – 244 2005 – 281 Basic concepts: Computer output as a second opinion Improve accuracy and consistency of radiological interpretation, especially in screening populations (breast, lung, colon cancers) Reduce reading time
Part 2: OS CAD Commercial History: FDA approval for mammography in 1998 In USA, >1500 CAD systems for breast cancer FDA approval for lung cancer in 2001 Under review: Colon cancer Renal cancer Liver cancer Cardiovascular disease
Part 2: OS CAD Is there a role for OS CAD? Basic technologies in CAD: Doi, K. Current status and future potential of computer-aided diagnosis in medical imaging. Br J Radiol. 2005; 78 Spec No 1: S3-S19. 1) image processing for detection and extraction of abnormalities 2) quantitation of image features for candidates of abnormalities 3) data processing for classification of image features between normals and abnormals 4) quantitative evaluation and retrieval of images similar to those of uknown lesions 5) observer performance using ROC analysis - Studies show a gain of 20% in early detection of breast cancers using CAD (controversial)
Part 2: OS CAD OS CAD tools: 1) image processing for detection and extraction of abnormalities 2) quantitation of image features for candidates of abnormalities ImageJ (Java) ITK/VTK (C++, funded $10 million by NLM) NeatVision (Java)
Part 2: OS CAD at U of T Where is the need? Reviewing Doi’s CAD technology components, there are two clear gaps: 1) Observer performance, or clinical validation 2) Normal/abnormal datasets to build/test these algorithms
Part 2: OS CAD at U of T Whitepaper Proposal: 1) Establish a CAD clinical testing network at U of T where OS algorithms can be clinically evaluated. Proposal is to refine the process utilizing our strengths (lung cancer CAD expertise) 2) Outline barriers (HIPAA, cost) to creating open database of image datasets with corresponding reports
Acknowledgements: Supervisor, Dr. Haider Project OS Committee