Personal tools

You are here: Home / Research / Biomedical Image Analysis / Professor Alison Noble / Machine learning in medical imaging

Machine learning in medical imaging

Professor Noble's group has an interest in how machine learning (and lots of medical images) can be used for quantitative analysis of ultrasound and microscopy image analysis. Both imaging modalities are hard to analyse using conventional image analysis techniques which are frequently based on assumptions  that imaging artefacts are weak (or can be corrected prior to analysis) , and objects of interest are well represented by geometric constraints, and classic edge, region, texture features. These assumptions often do not well-represent the appearance of clinical ultrasound and microscopy images.

The group's research ranges from the design and evaluation of state-of-art machine learning key structure detection, segmentation and quantification methods which embed knowledge and constraints on ultrasound acquisition and features, to the design of new tools to empower biologists to answer biological questions that they haven't been able to consider before due to a limit of quality of microscopy images, or non-scalability of manual analysis techniques.

Representative Publications (see also laboratory full publication list):

C Arteta,  V Lempitsky, JA Noble, A Zisserman
Learning to Detect Partially Overlapping Instances.
Proceedings of the 26th IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2013). June 25-27, 2013, Portland, USA. To appear.

B Rahmatullah, A Papageorghiou, JA Noble.
Integration of local and global features for anatomical object detection in ultrasound.

Proceedings of the Medical Imaging and Computer Assisted Interventions (MICCAI), 1-5 October 2012, Nice, France.

C Arteta, V Lempitsky, JA Noble, A Zisserman.
Learning Non-overlapping Extremal Regions To Detect Cells

Proceedings of the Medical Imaging and Computer Assisted Interventions (MICCAI), 1-5 October 2012, Nice, France.

 

RV Stebbing, JE McManigle, JA Noble.
Interpreting edge information for improved endocardium delineation in echocardiograms.
Proceedings of the International Symposium on Biomedical Imaging (ISBI) - IEEE, 2 - 5 May 2012, Barcelona, Spain, pp. 238-241.

M Yaqub, R Napolitano, C Ioannou, AT Papageorghiou, JA Noble.
Automatic detection of local fetal brain structures in ultrasound images.

Proceedings of the International Symposium on Biomedical Imaging (ISBI) - IEEE, 2 - 5 May 2012, Barcelona, Spain. pp. 1555-1558.

G Flaccavento, V Lempitsky, I Pope, PR Barber, A Zisserman, JA Noble, B Vojnovic. 
Learning to Count Cells: applications to lens-free imaging of large fields.
Proc. International Workshop on Microscopic image analysis with applications in Biology – Heidelberg, 2011

M Yaqub, M Verhoek, J Mcmanigle, JA Noble.
Learning optical flow propagation strategies using random forests for fast segmentation in dynamic 2D and 3D echocardiography
.
Proc. International Workshop on Machine Learning in Medical Imaging (MLMI), Toronto, 18 Sept 2011.

B Rahmatullah, A Papageorghiou, JA Noble.
Automated selection of standardized planes from ultrasound volumes
.
Proc. International Workshop on Machine Learning in Medical Imaging (MLMI), Toronto, 18 Sept 2011.

M Yaqub, K Javaid, C Cooper, JA Noble.
Improving the classification accuracy of the classic RF method by intelligent feature selection and weighted voting of trees with application to medical image segmentation
.
Proc. International Workshop on Machine Learning in Medical Imaging (MLMI ), Toronto, 18 Sept 2011.

V Lempitsky, M Verhoek, JA Noble, A Blake,
Random forest classification for automatic delineation of myocardium in real-time 3D echocardiography,

Functional Imaging and Modelling of the Heart 2009, Nice, France.