Using a virtual reality Powerwall to view virtual slides

The VR Powerwall at the University of Leeds School of Computing. A researcher is viewing a gigapixel virtual slide. The image shows a piece of skin tissue (the epidermis is the thin pink line on the surface running diagonally across the image). There is an intradermal naevus ("mole") centrally. The area of tissue in the image above is just 7.5mm x 4mm in real life.

Background

Histopathology is the medical specialty concerned with diagnosing disease by examining tissue. Histopathologists use microscopes to examine glass slides containing pieces of tissue - for example to diagnose breast or colon cancer.

Left: A conventional light microscope as used in hospitals all over the world. . Right: A glass slide containing a pathology specimen - in this case a section of the gut containing a tumour between 3 and 7 o'clock.

Now pathology is undergoing a digital revolution - so called "virtual slides" are enormous digital images produced by scanning glass pathology slides at very high resolution (200,000 dots per inch). The resulting images are gigapixels in size - far larger than images taken by even the best digital cameras.
However because of their size, viewing virtual slides on conventional monitors is inefficient - up to 60% less so than a microscope. One reason for this is the small display size of a computer monitor compared to a light microscope.
A Virtual Reality Powerwall is a novel user interface device which allows display of large high resolution images.
The Powerwall at the University of Leeds School of Computing (pictured at the top of this page) comprises an array of 28 LCD screens controlled by an 8 node computer cluster. The resulting image is over 3.5 by 1.5 metres in size, but has the same resolution as a conventional computer screen. This allows, for example, a whole gastrointestinal biopsy to be seen at once at a resolution which shows every individual cell in detail.
You can see 2 videos showing pathologists using the Powerwall below.

Videos of pathologists using the VR Powerwall to diagnose a skin biopsy (left) and to count mitoses in cancer cells (right). Note how quickly the image can be navigated despite the techinical challaneges involved in rendering 50 megapixels. On the right, note the use of so-called physical navigation (i.e. walking!) to view the slides.
Can't see the videos? You'll need Flash version 10 - get it here.

Methods

We assessed the Powerwall for diagnostic use by directly comparing it with the conventional microscope. Custom-made software was written to render virtual slides on a 50 megapixel virtual reality powerwall. A controlled experiment was performed with 8 pathologist subjects performing 4 standard tasks to compare the efficiency and acceptability of the two modalities. Users had no previous experience of using a powerwall.

Results

Efficiency and diagnostic confidence were equivalent between microscope and Powerwall - mean time to complete diagnostic tasks was 86s for the conventional microscope and 88s for the powerwall (see figure below); self-reported diagnostic confidence was over 90% for both modalities.  

Mean time to perform tasks was equivalent between Conventional Microscope and Powerwall

Conclusions

This pilot study suggests that a powerwall has the potential to be as efficient as the conventional microscope, and despite the lack of familiarity of subjects with the system their diagnostic confidence was maintained using this new interface.
The device is currently a prototype but funding has been obtained from the UK National Institute for Health Research (NIHR) to develop it further and install two Powerwalls on site at St. James University Hospital Leeds.

Reference

Virtual reality Powerwall versus conventional microscope for viewing pathology slides: an experimental comparison

Treanor D (1), Jordan-Owers N (1), Hodrien J (2), Wood J (2), Quirke P (1), Ruddle R (2)

1. Pathology and Tumour Biology, Leeds Institute of Molecular Medicine, University of Leeds
2. Visualization and Virtual Reality Group, School of Computing, University of Leeds

Accepted for Publication, Histopathology, 2009

Acknowlegements

V V R group, University of Leeds

JISC/ vizNET, Prototype Powerwall funding

Pathological Society of Great Britain and Ireland, Pilot Grant Scheme 2007