Antonio Criminisi

Antonio Criminisi

Auto-generated profile, based on publicly available information

Antonio Criminisi

Machine Learning and Perception, Microsoft Research, Cambridge, UK

Research Focus & Metrics

Academic Career Length

23Years (2004-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
13 August 2015 in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization
Verified
Journal Article
Closed access
JMJan Margeta
Jan Margeta
ACA. Criminisi
A. Criminisi
RLR. Cabrera Lozoya
R. Cabrera Lozoya
DLD.C. Lee
D.C. Lee
NAN. Ayache
N. Ayache
In this paper, we propose a convolutional neural network-based method to automatically retrieve missing or noisy cardiac acquisition plane information from magnetic resonance imaging and predict the five most common cardiac views. We fine-tune a convolutional neural network (CNN) initially trained on a large natural image recognition data-set (Imagenet ILSVRC2012) and transfer the learnt feature representations to cardiac view recognition. We contrast this approach with a previously introduced method using classification forests and an augmented set of image miniatures, with prediction using off the shelf CNN features, and with CNNs learnt from scratch. We validate this algorithm on two different cardiac studies with 200 patients and 15 healthy volunteers, respectively. We show that there is value in fine-tuning a model trained for natural images to transfer it to medical images. Our approach achieves an average F1 score of 97.66% and significantly improves the state-of-the-art of image-based cardiac view recognition. This is an important building block to organise and filter large collections of cardiac data prior to further analysis. It allows us to merge studies from multiple centres, to perform smarter image filtering, to select the most appropriate image processing algorithm, and to enhance visualisation of cardiac data-sets in content-based image retrieval.

24 June 2015 in JMIR Human Factors
Verified
Journal Article
Open access
CMCecily Morrison
Cecily Morrison
MDMarcus D'Souza
Marcus D'Souza
KHKit Huckvale
Kit Huckvale
JDJonas F Dorn
Jonas F Dorn
JBJessica Burggraaff
Jessica Burggraaff
ASAbigail Sellen
Abigail Sellen
JMIR Human Factors - a leading peer-reviewed ehealth journal

13 November 2004
Verified
Conference Paper
Closed access
ACA. Criminisi
A. Criminisi
ABA. Blake
A. Blake
This paper describes a novel algorithm for the efficient synthesis of high-quality virtual views from only two input images. The emphasis is on the recovery of continuity of objects boundaries (figural continuity) with faithful synthesis of transparency effects. The contribution of this paper is two-fold: (i) the split-patch search (SPS) technique is introduced for dense stereo which handles transparency effects by assigning multiple disparities to mixed pixels; (ii) an efficient extension of exemplar-based image synthesis to the case of two-camera stereo is proposed. Furthermore, this paper presents an approximate but effective solution to the challenging problem of layer estimation and compositing in the case of small image patches. The effectiveness of the proposed technique is demonstrated on a number of stereo image pairs taken from two camera video-conferencing setups, where the quality of the synthesized talking heads is of paramount importance. Moreover the improvement in the quality of image synthesis is quantified by comparing the output of the SPS algorithm with thirteen ground-truth images.