Philips and UKE develop computer-aided diagnosis of PET and MRI brain scans

11 July 2007

Hamburg, Germany. The University Medical Center Hamburg-Eppendorf (UKE) and Royal Philips Electronics have developed software for computer-aided diagnosis (CAD) of PET and MRI imaging of the brain to aid the identification of neurodegenerative diseases.

The CAD software automatically interprets PET (positron emission tomography) brain scans of patients suspected of having a neurodegenerative disease that leads to dementia, and combines them with MRI (magnetic resonance imaging) scans for accurate differential diagnosis.

The development of such a system will ultimately mean a better quality of life for patients by enabling earlier prescription of drugs that delay progression of the disease, and hence delay the worst effects of dementia. It will also provide pharmaceutical companies and clinicians with a valuable tool for the development and testing of new, potentially curative drugs for neurodegenerative diseases such as Alzheimer’s.

“In the not too distant future there is going to be much greater demand for the accurate early diagnosis of neurodegenerative disease and not everyone will have access to the clinical expertise of a university hospital to obtain it,” says Dr. Ralph Buchert of the Department of Nuclear Medicine at UKE. “The availability of an automated system will help less experienced physicians to achieve the same high level of accuracy in their diagnoses.”

Dementia is a debilitating condition that already affects more than 25 million people worldwide*, the commonest form being Alzheimer’s disease. As the demographics of world populations increasingly shift towards older age groups, dementia is widely expected to reach epidemic proportions unless effective treatments are found for it.

“Building on our expertise in multi-modal diagnostic imaging, we’ve combined functional and structural brain-scan information into a fully integrated and easy to use system for diagnosing the principal neurodegenerative diseases that cause dementia,” says Dr. Lothar Spies, Head of the Digital Imaging Department at Philips Research. “Ultimately, it will enable early treatment and highly personalized therapies.”

The software accurately overlays anatomical images of the brain obtained from MRI scans with PET scans that display brain activity — specifically the uptake of glucose that fuels brain activity. By using advanced image processing and computer learning techniques in combination with a database of reference brain-scans, the system then analyses the images automatically and displays anomalous brain patterns in a concise way. Based on these patterns, it then suggests a diagnosis. As a result, the system will help less experienced doctors to achieve the same diagnostic accuracy as highly trained specialists.

The new diagnostic technique, which has already proven its accuracy using historical image data and known patient outcomes, is about to undergo clinical evaluation at UKE.

The clinical evaluation will run the CAD system alongside UKE’s existing dementia diagnosis procedures with the aim of fine-tuning the system’s ability to detect and differentiate the three most common types of neurodegenerative disease — Alzheimer’s Disease, Lewy-body Dementia and Frontotemporal Dementia.

To top