Data Capture in Cancer Registration Barry Plewa Analyst Programmer

1
Data Capture in Cancer RegistrationBarry
PlewaAnalyst Programmer
2
Traditional data capture process

• Data collection officers
• Paper forms (Cancer registration forms)
• Manual data entry
• 70,000 standard cancer registration forms scanned
  each year

3
One solution?
OpticalCharacter RecognitionTechnology
4
Scanning process

• Designed to enhance current process until fully
  automated registration in place
• Works alongside data entry process to facilitate
  information extraction

5
Scanning process

• Scan cancer registration form
• Interpret information on the form using a
  character recognition engine
• Manual verification
• Indexing and storage

6
Character recognition
7
Optimisation

• Clear areas for handwriting
• Each field either alpha or numeric to avoid
  difficult decisions. E.g., 5 and S
• Tick-boxes wherever possible
• Lookup fields

8
Character recognition

• Each field is given a formatEg. Date of birth
  expects NN/NN/NNNN
• Recognition engine attempts to read the
  information using the specified format.

9
Eyes and hands for forms
10
Eyes and hands for forms

• Scan documents
• Interpret characters

11
Eyes and hands for forms
12
Eyes and hands for forms
13
Eyes and hands for forms

• 90-95 character recognition rate
• Interprets at a rate of 1 form every 5-10 seconds
• Visual Basic used to store/index images
• Index data stored on Microsoft SQL Server

14
TCR Image Viewer
15
Problems/limitations?

• Handwriting
• Incorrect interpretation
• Manual verification required

16
Current uses?

• Cancer registration form image archiving
• DCO Tracing Project

17
Future Uses?

• Site/disease specific forms
• Extension of DCO tracing project
• Haematology registration
• Integration into cancer registration application