Simplicity Versus Realism in Medical VR
We are so caught up in our desire for ever greater realism and ever greater accuracy when using VR and modelling tools to render the human body for diagnosis in whole or in part, that we often forget the benefits of simplicity.
Virtual reality is a wonderful tool if used right, capable of showing almost any item imaginable from practically any perspective we can think of. Those of us who work with it all the time, often see it like a hyperactive puppy, eagerly awaiting the next trick we throw its way.
A danger we often miss in our eagerness to impress, is that sometimes the most impressive display method is not the most practical. We miss the real strength of VR it's ability to render something in ways you never physically see in our headlong rush to outdo ourselves.
A prime example of the kind of thing we often miss, is the work of a team of computer scientists, physicists, and physicians at Harvard university. One year ago, they developed a paradigm shift in artery diagnosis. Rather than rendering the arteries as they appear in nature, as a 3D network of twists and turns, visually not unlike tree branches, then ry and provide visual tools for the doctor to work out from this confusing mess where the problem area is, they turned the whole thing on its head.
They used VR visualisation methods to transform the original 3D mess, into a flat 2D diagram, with minimum confusion and maximum comprehension. By removing clutter and presenting the data in the most straight-forward method possible, it greatly assists the physician in determining exactly where the problem lays.
The new VR system's name is HemoVis, and ties directly into DICOM data sources. It still uses the same complex fluid dynamics calculations common to 3D VR systems it <i>is</i> a 3D VR system. The difference is, when it returns data to the end-user it does so in a 2D format, with the system itself, doing the work of scanning through the branching coronary system, looking for the faults the physician would be scanning for usually shear stress points in the artery, that can only be found via a moving fluid simulation in 3D. The results, with the bulk of the work already done, are then handed to the physician in a format designed for maximum information in minimum time.
In clinical settings at the moment, 3D models, automatically rendered from CT data are the standard method presented to doctors, in order to portray the shape and spatial arrangement of all blood vessels in the target area. That display is useful, but not useful enough.
In a small trial of HemoVis, which started after the program was first presented at conference in 2011 (the IEEE Information Visualization Conference), the researchers found that doctors tsting the system out were able to increase their diagnostic accuracy from 39% to 91%. A significant jump,l with the only change being the way the data was presented, and a decrease in the interactivity options to allow the doctor to focus on the results.
Our goal was to design a visual representation of the data that was as accurate and efficient for patient diagnosis as possible, lead author of the study Michelle Borkin stated. "What we found is that the prettiest, most popular visualization is not always the most effective."
"In the 3D case, the more complex and branched the arteries were, the longer it took to complete the patient diagnosis, and the lower the accuracy was. In the 2D representation, it didn't matter how many branches we had or how complex they werewe got consistently fast, accurate results. We werent expecting that.
The researchers relied on continual input from physicians and others with clinical or laboratory imaging experience throughout the process. Through extensive surveys and interviews, they identified the most popular options for display, accurate layout, and colouring of these arterial projections.
Doctors initially showed some resistance when asked about the benefits of an alternative colour scheme, citing familiarity, chromatic vibrancy, and aesthetic appeal in support of the rainbow scheme that is currently most common in scientific visual representations. However, there are problems with that display. Most notably known conflicts between rainbow colouring and the human visual system's ability to understand the significance of data.
As Borkin put it: For years, visualization, computer science, and psychology researchers have identified that colour is critical for conveying the value of data, but that the rainbow colouring is not well-attuned to the human visual system.
Accordingly, HemoVis departs from the traditional practice of rainbow colour-coding in favour of a graded single-colour scheme (red to black) that can represent placement along a continuum.
In tests, diagnostic accuracy, as measured by the proportion of diseased areas identified, increased dramatically with the new colour scheme.