Implementation of a Commercial Scalable, MultiUser Client Server Image Database for Pathology

1
Implementation of a Commercial Scalable,
Multi-User Client Server Image Database for
Pathology

• Michael Feldman, MD, PhD and John E. Tomaszewski,
  MDDepartment of Pathology Laboratory Medicine,
  University of Pennsylvania Health System,
  Philadelphia, PA 19104

2
Hardware

• Server HP Netserver, dual processing PIII, 1
  GB RAM, 100 GB HD array
• Client Windows NT or Windows 2000
• Image Capture
• Microscopic Polaroid Digital Microscope Camera
• Gross Nikon Coolpix 990
• Kodachrome Nikon LS2000
• Flatbed Epson

3
Software

• Database Server
• Single user MS Access
• Multi-user MS-SQL7, MS-SQL2000, Oracle
• Client
• ThumbsPlus (www.cerious.com)
• Developing ASP

4
Server Features

• Image database holds pointers to original image
  and can span network volumes
• Database creates and stores image Thumbnail as
  jpeg file
• Database stores image information including
• Keywords (Indexed and searchable)
• User defined fields (Indexed and searchable)
• EXIF data also searchable (also searchable)
• Annotations

5
Client Software

• Image capture (single or multiple) via TWAIN
  interface
• Automatic folder scanning
• Automatic keyword generation from words in
  filename
• Organize images within directory structure
  similar to Windows Explorer
• Basic Image enhancement integrated
• Image Gallery works like a 35 mm Light Box
• Built in Slide show feature for quick
  conferencing (tumor board) Show images directly
  from database without using PowerPoint

6
Thumbs Plus Client Software
Images displayed as thumbnail for viewing,
similar to giant Light Box

• Images organized by hierarchical subdirectory
• A. Organ System
• Service
• Autopsy
• Cytology
• Surgical Path

7
Client Query Tools
Search by Filename Keyword User defined
fields Direct SQL (assumes knowledge of tables
and SQL) Image Similarity (not great for
pathology)
8
Database Usage

• First 12 months of use
• 21,000 Images
• Reduced kodachrome costs by 15K/year
• Incorporate Images in Autopsy reports
• Incorporate images in private consults
• Image Database Functions
• Has allowed sharing of images easily and in real
  time for lecture, labs and general education
• Allows faculty and residents to produce talks in
  less time once images they need are in database
  (May need to scan in some part of there old
  kodachrome collection)
• Created a gross photography lecture series to
  teach gross diagnostic skills and basic digital
  photography skills to residents

9
Future Scope

• Develop ASP based query tool to allow non-client
  based systems to access database
• Develop transaction logging system to track which
  users look at which records for HIPPA compliance
• Using ASP front end to integrate images from
  database into web based educational tools for
  residents and medical students
• Integrate images with AP datawarehouse