High Volume Slide Scanning Architecture and Applications

1
High Volume Slide ScanningArchitecture and
Applications

• Dr. André Huisman
• Department of pathology
• UMC Utrecht, The Netherlands
• a.huisman-4_at_umcutrecht.nl

2
Department of pathology UMCU

• UMC Utrecht
• gt1,000 beds
• gt10,000 employees
• Department of pathology
• 20.000 surgical pathology cases
• 156.000 glass slides (histology, cytology, IHC)
• 15 pathologists, 10 residents

3
Digital pathology - advantages

• Digital Archiving
• Instant access from multiple locations by
  multiple people
• No searching for slides
• Constant quality
• Telepathology
• Consultations, revisions and panels
• Education
• Research
• Automated image processing

4
Project background

• Clinico pathological conferences
• 900 meetings every year
• No (multi headed) microscope needed
• Quicker preparation of meetings
• No retrieval of glass slides from archive

5
Aim (2007)

• Digitize all diagnostic slides we have
  (prospectively)

6
Challenges

• Scanners
• Image size up to 1 GB x 500 slides per day
• No existing infrastructure present for storage of
  this size at our facility
• Image presentation and software integration
• Logistics

7
Scanners

• Different manufacturers
• Speed
• Focusing method
• Acquisition technique
• z-stack acquisition
• File format policy
• Application integration
• 2007 3D Histech (Zeiss), Aperio, Hamamatsu,
  Olympus (US Dmetrix, BioImagene)
• 2010 Leica, Menarini, Philips, Omnyx, BioImagene

8
Scanning logistics

• 3 Aperio XT scanners (120 slides per scanner)
• Morgue assistants
• Mark slides after scanning
• 6 Hours per run at 20x magnification (3
  minutes/scan)

9
Storage HSM

• HSM Hierarchical Storage Management
• Sun Microsystems (Oracle)
• 6 TB available on very fast fibre channel disks
• 120 TB available on tape (750 GB each)
• 2 Tape drives
• Completely transparent archiving and retrieval
  (robot)
• Access time from tape 1 - 3 minutes

10
Linking systems

• 1D Barcodes
• U-DPS reporting system
• LMS Laboratory Management System
• Spectrum Aperios image management solution
• Storage system
• Own development integration layer

11
Scanner workflow
Other images (e.g. macro)
Scanning
Order form scanner
Database
LIS
Image / Data server
Speech recognition
U-DPS
Storage
pathology users
12
Validation

• Aim validate diagnostic use of digital slides
• Method reevaluate diagnosis with same
  pathologist on scanned slide after washout period
  (1 year) for several organs
• Gold standard original diagnosis using
  traditional microscopy

13
Preliminary results validation

• GI tract discrepancy
• Different interpretation of abnormality
• Glass slide and WSI contained same information
• Skin discrepancy
• Clinical information not used for interpreting
  digital slide

14
Pitfalls of digital archive

• Costs
• Huge storage needs 40 TeraByte per year (over
  57,000 CDs)
• Largest storage in the UMCU
• Logistics of scanning up to 500 slides per day
• Currently scanning almost 24 / 7
• Place of scanning in process
• Speed of image retrieval
• Image compression (JPEG 2000?)
• Backup

15
Education

• All students view the same best slide
• Slide images can be integrated with
• Annotations
• Questions
• Macroscopic images
• Other multimedia
• Most UMCU microscopy practical sessions are
  digital
• Student satisfaction is high

16
Teleconsultation

• Place (small) slide scanners at different labs
• Upload digital slides to UMC Utrecht
• Aurora mScope Clinical
• www.pathoconsult.com

17
Digital slide panel discussions
18
Image processing research

• Image processing applications on virtual slides
• Detecting mitotic figures in breast cancer slides
• Use texture features
• Establishing histological grade in breast cancer
• Segmentation of individual nuclei (on HE stained
  slides)
• Detect points of interest
• Use marker-controlled watershed segmentation
• Post processing

19
Conclusions

• Routine scanning is possible and makes sense
• Future of pathology is digital
• Digital pathology is expensive
• Digital pathology is just starting..
• Together we are shaping Pathology 2.0

20
Discussion

• Limitations current system
• Cytology
• Speed
• Magnification (20x / 40x incidental)
• Backup
• Quality Control

21
Discussion

• Archive heavily used
• Educational use still growing
• Teleconsultation network growing
  (www.slideconsult.com)
• Need for standards
• DICOM / JPEG 2000
• Images, annotations and reports
• Mixing scanners and integration with other
  software platforms (middleware?)
• Image management central in workflow for
  pathologist?

22
Questions?!

• Huisman et al., Creation of a fully digital
  pathology slide archive by high-volume tissue
  slide scanning, Human Pathology, 2010 May 41(5)
  751-7
• a.huisman-4_at_umcutrecht.nl