Back to Projects

Image Computing and Computational Modelling (pre-2019)

Machine learning for differential diagnosis of dementia from multi-modality MR and PET imaging

Project ID: 2014_315

Student: Christopher Bowles

1st supervisor: Daniel Rueckert, Imperial College London
2nd supervisor: Roger Gunn, Imperial College London

Dementia has recently been confirmed as a health priority both in Europe and in the USA. Dementia accounts for costs equivalent to about 1% of the global gross domestic product (GDP) – 461 billion euros annually. In 2010, 36 million people were living with dementia worldwide but the number is expected to increase to 115 million by 2050. Overcoming the huge human and economic challenge of dementia requires a solution to two major clinical challenges: 1) lack of efficient treatments and 2) lack of tools for early diagnosis to select patients that would benefit from such treatments. While our group and others have made progress regarding the early diagnosis of Alzheimer’s disease based on imaging, other domains have been little explored. Differential diagnosis, i.e. separation between different types dementias, such as Alzheimer’s disease, vascular dementia, fronto-temporal lobe dementia and Lewy-body dementia, is of paramount importance, but remains very challenging. In this project we will address the challenges of early diagnosis as well as differential diagnosis by exploiting rich, multi-modal imaging, including MR and PET images as well as advanced machine learning techniques, in particular deep learning and multiple instance learning techniques.

Back to Projects