Imperial College London
Alternate

ProfessorDanielRueckert

Faculty of Engineering, Department of Computing

Professor of Visual Information Processing
 
 
 
//

Contact

 

+44 (0)20 7594 8333d.rueckert Website

 
 
//

Location

 

568Huxley BuildingSouth Kensington Campus

//

Summary

 

Publications

Citation

BibTex format

@inproceedings{Li:2018:10.1007/978-3-030-00928-1_45,
author = {Li, Y and Khanal, B and Hou, B and Alansary, A and Cerrolaza, J and Sinclair, M and Matthew, J and Gupta, C and Knight, C and Kainz, B and Rueckert, D},
doi = {10.1007/978-3-030-00928-1_45},
pages = {392--400},
publisher = {Springer Verlag},
title = {Standard plane detection in 3D fetal ultrasound using an iterative transformation network},
url = {http://dx.doi.org/10.1007/978-3-030-00928-1_45},
year = {2018}
}
Download

RIS format (EndNote, RefMan)

TY  - CPAPER
AB  - Standard scan plane detection in fetal brain ultrasound (US) forms a crucialstep in the assessment of fetal development. In clinical settings, this is doneby manually manoeuvring a 2D probe to the desired scan plane. With the adventof 3D US, the entire fetal brain volume containing these standard planes can beeasily acquired. However, manual standard plane identification in 3D volume islabour-intensive and requires expert knowledge of fetal anatomy. We propose anew Iterative Transformation Network (ITN) for the automatic detection ofstandard planes in 3D volumes. ITN uses a convolutional neural network to learnthe relationship between a 2D plane image and the transformation parametersrequired to move that plane towards the location/orientation of the standardplane in the 3D volume. During inference, the current plane image is passediteratively to the network until it converges to the standard plane location.We explore the effect of using different transformation representations asregression outputs of ITN. Under a multi-task learning framework, we introduceadditional classification probability outputs to the network to act asconfidence measures for the regressed transformation parameters in order tofurther improve the localisation accuracy. When evaluated on 72 US volumes offetal brain, our method achieves an error of 3.83mm/12.7 degrees and3.80mm/12.6 degrees for the transventricular and transcerebellar planesrespectively and takes 0.46s per plane.
AU  - Li,Y
AU  - Khanal,B
AU  - Hou,B
AU  - Alansary,A
AU  - Cerrolaza,J
AU  - Sinclair,M
AU  - Matthew,J
AU  - Gupta,C
AU  - Knight,C
AU  - Kainz,B
AU  - Rueckert,D
DO  - 10.1007/978-3-030-00928-1_45
EP  - 400
PB  - Springer Verlag
PY  - 2018///
SN  - 0302-9743
SP  - 392
TI  - Standard plane detection in 3D fetal ultrasound using an iterative transformation network
UR  - http://dx.doi.org/10.1007/978-3-030-00928-1_45
UR  - https://link.springer.com/chapter/10.1007%2F978-3-030-00928-1_45
UR  - http://hdl.handle.net/10044/1/60751
ER  -
Download