Two brain images. Left is a healthy patient, right is suffering Alzheimer's Disease.

Typical brain images of 70 year-old subjects: a healthy subject (left) and a patient with Alzheimer's. The colour codes the thickness of the cortex (the more blue, the thinner).

Alzheimer's Disease: image processing for early diagnosis

CSIRO software aids early detection and diagnosis of this debilitating condition.

  • 14 July 2011 | Updated 14 October 2011

Alzheimer’s disease is a neurodegenerative condition which takes several decades to develop, gradually robbing people of basic cognitive functions, their memories and eventually their lives.

By the time a diagnosis of Alzheimer’s disease can be made using current techniques, the patient is likely to be experiencing significant loss of brain function.

Finding objective evidence of Alzheimer’s disease before symptoms occur will help ensure treatment and care aimed at slowing or preventing development of the disease.

Current activities

Through CSIRO’s Preventative Health Flagship, staff at the Australian e-Health Research Centre (a joint venture between CSIRO and the Queensland government) are developing computerised enhancement and analysis of Positron Emission Tomography (PET) and Magnetic Resonance Imaging (MRI) scans to detect and monitor disease progression.

Our software should assist treatment planning for individual patients as well as tracking the effectiveness of new therapies.

Alzheimer’s disease is associated with the build up of a protein called amyloid beta. A relatively new PET imaging agent, Pittsburgh Compound-B (PIB) acts like a dye, revealing the build-up of amyloid deposits in the brain.

While PET scanners pick up PIB, the raw intensity data of the scan needs to be processed before the amount of amyloid can be calculated and to allow comparison between patients. This is currently a lengthy, manual task. We are developing software to automate the process.

We are also building ‘average scans’ of different patient populations, such as age groups, using novel registration and modelling techniques. Using our custom-built templates for comparison studies or statistical analyses avoids the bias associated with the using ordinary templates or the scan of one individual.

MRI reveals the different tissues layers of the brain in 3D: thinning of the cortex, in particular, is associated with the progression of Alzheimer’s Disease. Unfortunately manual measurement of cortical thickness from this 3D dataset is difficult and prone to error.

We are developing algorithms that rapidly and accurately identifies the different regions of the brain and automatically calculates cortical thickness.

Outcomes

Our aim is to provide clinicians with software that would take a patient’s brain scan, make several measurements difficult or impossible for humans to obtain and benchmark that patient against a typical person of the same age.

This automated solution should provide improved reliability and accuracy, compared with manual image analysis.

Our software should assist treatment planning for individual patients as well as tracking the effectiveness of new therapies.

Achievements

Our imaging technologies have enabled accurate mapping of the early stages of Alzheimer’s disease: MRI revealing brain shrinkage and PET with PIB tracking the build-up of amyloid deposits.

Partners

This work is part of the Australian Imaging, Biomarker and Lifestyle (AIBL) study.

We are collaborating with research organisations worldwide on evaluation of our software.

Read more technical details of this work in Alzheimer’s Disease [external link].

AEHRC logo. The Australian e-Health Research Centre is a joint venture between CSIRO and the Queensland Government.