Machine learning of neuroimaging to diagnose cognitive impairment and dementia: a systematic review and comparative analysis
Authors:
Enrico Pellegrini,
Lucia Ballerini,
Maria del C. Valdes Hernandez,
Francesca M. Chappell,
Victor González-Castro,
Devasuda Anblagan,
Samuel Danso,
Susana Muñoz Maniega,
Dominic Job,
Cyril Pernet,
Grant Mair,
Tom MacGillivray,
Emanuele Trucco,
Joanna Wardlaw
Abstract:
INTRODUCTION: Advanced machine learning methods might help to identify dementia risk from neuroimaging, but their accuracy to date is unclear.
METHODS: We systematically reviewed the literature, 2006 to late 2016, for machine learning studies differentiating healthy ageing through to dementia of various types, assessing study quality, and comparing accuracy at different disease boundaries.
RES…
▽ More
INTRODUCTION: Advanced machine learning methods might help to identify dementia risk from neuroimaging, but their accuracy to date is unclear.
METHODS: We systematically reviewed the literature, 2006 to late 2016, for machine learning studies differentiating healthy ageing through to dementia of various types, assessing study quality, and comparing accuracy at different disease boundaries.
RESULTS: Of 111 relevant studies, most assessed Alzheimer's disease (AD) vs healthy controls, used ADNI data, support vector machines and only T1-weighted sequences. Accuracy was highest for differentiating AD from healthy controls, and poor for differentiating healthy controls vs MCI vs AD, or MCI converters vs non-converters. Accuracy increased using combined data types, but not by data source, sample size or machine learning method.
DISCUSSION: Machine learning does not differentiate clinically-relevant disease categories yet. More diverse datasets, combinations of different types of data, and close clinical integration of machine learning would help to advance the field.
△ Less
Submitted 11 April, 2018; v1 submitted 5 April, 2018;
originally announced April 2018.
