Decision Support Software To Assist in the Differential Diagnosis of Dementia

A usability and feasibility study of a decision-support software tool for the automatic evaluation of dementia using positron emission tomography (PET) image data has just been successfully concluded.

The Philips Research (Best, The Netherlands) software has been evaluated in collaboration with the University Medical Center Hamburg-Eppendorf (Hamburg, Germany) and Austin Hospital (Melbourne, Australia).

Currently, one of the most effective ways of diagnosing dementia in its earlier stages is by performing a PET brain scan with the tracer fluorodeoxyglucose (FDG). FDG-PET is a molecular imaging technique that produces a three-dimensional image of functional processes in the body--in this case, the uptake of glucose that fuels brain activity. However, the interpretation of PET brain-scan images requires a great amount of skill, especially in the early stages of neurodegenerative disease.

The Philips decision support software has shown positive results in retrospective studies using libraries of FDG-PET brain-scan images that had already been independently interpreted by an expert. In a study based on a University Medical Center Hamburg-Eppendorf library of FDG-PET scans from 83 patients, the software achieved better than 98% correspondence with the expert physician's interpretation when programmed to differentiate between brain scans showing no signs of dementia, brain scans characteristic of Alzheimer's disease and those characteristic of frontotemporal dementia.

In a similar 48-patient study using FDG-PET images provided by Prof. Christopher Rowe, director of Nuclear Medicine & Center for PET at the Austin Hospital, the software achieved better than 80% accuracy in differentiating between the scans of nondiseased patients, those suffering from Alzheimer's, frontotemporal dementia, and Lewy body dementia. This so-called four-class differential diagnosis is more difficult because indications of Alzheimer's and Lewy body dementia occur in similar areas of the brain, while indications of frontotemporal dementia appear in a separate area of the brain.

Dementia currently affects well over 25 million people worldwide. It is the result of a number of progressive degenerative diseases of the brain. The most common dementia causing diseases are Alzheimer's disease, Lewy-body dementia, and frontotemporal dementia--accounting for around 60%, 15%, and 10% of all dementia cases, respectively. Early differential diagnosis of the type of dementia that a patient is suffering from is essential to maximize the benefit of available drugs that can prevent rapid deterioration and subsequent loss in quality of life.

The Philips software analyses PET brain-scan images, and combines them with magnetic resonance imaging (MRI) scans to detect the characteristic patterns of brain diseases such as Alzheimer's, frontotemporal dementia, and Lewy Body dementia, and then quantifies the degree to which they resemble reference images of known dementia sufferers. The University Medical Center Hamburg-Eppendorf, codeveloper of the system, has been using the software alongside its existing diagnostic procedures for the last year to evaluate its feasibility and usability in a clinical setting.

"The results of the study have been truly excellent,” commented Dr. Ralph Buchert of the department of nuclear medicine at the University Medical Center Hamburg-Eppendorf. "With most steps performed automatically, operating the software is very straightforward and only adds a couple of minutes to the procedure time. When we made it available to referring physicians who were not skilled in reading FDG-PET and MRI images, they were able to analyze images and interpret the results within a few minutes.”

"Computer-based clinical decision support is regarded as one of the best ways of enabling physicians to optimally use available data and knowledge so that they can consistently deliver evidence-based medicine to their patients in a timely manner,” stated Dr. Henk van Houten, senior vice president Philips Research and head of the Healthcare research program. "Ultimately, such systems will be judged by the people who use them, and that is why it is important to develop this decision support software to aid the diagnosis of dementia in close collaboration with leading dementia research institutes.”

Following up on from the encouraging results of these studies, Philips Research is conducting additional studies involving the software in collaboration with the University of Washington (Seattle, USA).

Related Links:
Philips Research
University Medical Center Hamburg-Eppendorf