What Is The Paraffin-Embedded Tissue Archive Project?

1
What Is The Paraffin-Embedded Tissue
ArchiveProject?

• Joint venture of the Biomedical Informatics
  Facility, the Department of Pathology and
  Laboratory Medicine and the Abramson Cancer
  Center
• Business Agreement with AIM Document Management
  to scan up to 600,000 Surgical Pathology Reports
  (SPRs) of paraffin-block archived tissues
• Plan to populate a relational database schema,
  caTIES, with information from the scanned SPRs
• Collaboration with the caTIES development team at
  the University of Pittsburgh to add TIFF images
  of pathology reports into the caTIES database

2
Paraffin-Embedded Tissue Archive Project Why Is
It Important To caBIG?

• There are vast quantities of paraffin embedded
  tissue in archives within Surgical Pathology
  Departments throughout the country
• The UPenn Department of Pathology and Laboratory
  Medicine has approximately 600,000
  paraffin-embedded tissue samples that date back
  to the 1940s
• The project will use microfiche that will be
  scanned to create TIFF images of all
  corresponding pathology reports, extract
  accession/subject IDs and full text using OCR
  (Optical Character Recognition)
• caTIES and De-ID will be used to chunk pathology
  reports and parse specific text strings
• The resulting dataset will be uploaded to the
  caTIES database, allowing use of the caTIES query
  interface and data model
• Conversion of the microfiche information for
  uploading into the caTIES relational database is
  required to expose these tissue samples to the
  caBIG community

3
Paraffin-Embedded Tissue Archive Project How Do
We Plan To Achieve The Project Goals?
Borrowed from john.curtin.edu.au/society/justice
4
Paraffin-Embedded Tissue Archive Project What
Has Been Accomplished Thus Far?

• All of Penns Paraffin related SPRs have been
  scanned by AIM Document Corp
• All accession numbers for these SPRs have been
  manually keyed to ensure accuracy
• Over 430,000 scanned SPRs have been received
• For each of these SPRs we have received 3 files
  a TIFF image file of the SPR, a Word document
  file of the OCRed SPR, and a text file of the
  manually entered accession number from the SPR
• A detailed QA analysis of over 1500 reports has
  been conducted

5
Paraffin-Embedded Tissue Archive Project Whats
Next?

• Test Run
• Run a roll rated good quality and a roll rated
  poor quality through caTIES to see how well
  they are coded
• Populate the caTIES schema with all of the
  Paraffin files
• Run all of the files through the caTIES pipeline
• Conduct quality assurance testing on the coded
  files
• Set up a link between the OCRed files and the
  TIFF image files

6
Paraffin-Embedded Tissue Archive ProjectGood
Quality OCRed File
7
Paraffin-Embedded Tissue Archive ProjectPoor
Quality OCR File