Clinical applications of machine learning in neuroimaging
A Satellite Symposium at BIH 2015
On Sunday 30th August 2015, Dr James Cole from C3NL chaired a symposium at the Brain Informatics & Health conference, hosted by Imperial College London and held at the Royal Geographical Society, in South Kensington, London. For further details please visit the official conference website. The slides from the presentations will be made available on this page, shortly after the conference,
Symposium organising committee
Co-charing the symposium with Dr Cole are:
- Dr Ben Glocker, Biomedical Image Analysis Unit, Imperial College London
- Dr Pete Hellyer, Centre for Neuroimaging Science, King’s College London
- Dr Paul Aljabar, Biomedical Engineering Department, King’s College London
|Ana Namburete||Automated discovery of neurodevelopmental landmarks from 3D ultrasound images of the fetal brain||Institute of Biomedical Engineering, University of Oxford|
|Gareth Ball||Combining multimodal image analysis and clinical information to predict neurodevelopmental outcome in preterm infants||Centre for the Developing Brain, King’s College London||Ball_BIH_2015|
|Viktor Wottschel||Prediction of conversion to Multiple Sclerosis – an information fusion approach using Random Forests||Queen Square MS Centre, University College London||BIH2015_Wottschel|
|Kostas Kamnitsas||Deep Artificial Neural Networks for Lesion Segmentation in Multi-Sequence Brain MRI||Biomedical Image Analysis group, Imperial College London||Kamnitsas slides|
|Cyrus Eierud||Mild Traumatic Brain Injury Associations between Radiological Findings and Neurobehavioral Symptoms using Permutation Test and Support Vector Regression||National Intrepid Centre of Excellence (NICoE), Bethesda, USA||CyrusEierud_NICoE|
|Romy Lorenz & Rob Leech||The Automatic Neuroscientist: towards optimising patient-tailored cognitive rehabilitation therapy||Computational, Clinical & Clinical Neuroimaging Laboratory, Imperial College London|
Neuroimaging has given great insights into the biological processes underlying normal cognition and affected by disease, however, the field is yet to achieve significant impact at a clinical level. The inherently predictive framework of machine learning makes these techniques ideal to augment clinical decision-making, particularly as they can be used to make inference at an individual level, a key part of the developing trend towards personalised medicine. Nevertheless, the translation of cutting-edge machine learning techniques into clinical settings has been limited. This symposium aims to address this by encouraging researchers from both computer science and applied clinical or medical research to work more closely together and develop novel approaches to applying machine learning techniques. This is essential if we are to fully realise the potential of machine learning to translate neuroimaging data into tools that can directly aid clinicians and lead to improvements in the care and quality of life of patients with a range of neurological and psychiatric diseases.
The session will focus on how machine learning-based analysis of neuroimaging data can be applied in clinical settings. This will cover various techniques used to make classifications or predictions that lead to improvements in diagnosis, prognosis or clinical decision-making in neurological and psychiatric disorders. The potential medical applications of machine learning in neuroimaging range from across the lifespan.
Below are links to some of our published papers using machine learning in clinical neuroimaging
- Cole et al., (2015). Prediction of brain age suggests accelerated atropy after traumatic brain injury. Annals of Neurology
- Hellyer et al., (2013). Individual prediction of white matter injury following traumatic brain injury. Annals of Neurology
- Pandit, et al., (2014). Whole-brain mapping of structural connectivity in infants reveals altered connection strength associated with growth and preterm birth. Cerebral Cortex
- Gray et al., (2013). Manifold Forests for Multi-modality Classification of Alzheimer’s Disease. Decision Forests for Computer Vision and Medical Image Analysis.
- Zikic et al., (2014). Encoding Atlases by Randomized Classification Forests for Efficient Multi-Atlas Label Propagation. Medical Image Analysis.
- Konukoglu et al., (2013). Neighbourhood Approximation using Randomized Forests. Medical Image Analysis.