University of Surrey

Test tubes in the lab Research in the ATI Dance Research

A Kalman based approach with EM optimization for respiratory motion modelling in medical imaging

Smith, Rhodri L, Rahni, Ashrani Aizzudin Abd and Wells, Kevin (2018) A Kalman based approach with EM optimization for respiratory motion modelling in medical imaging IEEE Transactions on Radiation and Plasma Medical Sciences.

[img]
Preview
Text
Rhodri_Kalman_IEEE_paper.pdf - Accepted version Manuscript

Download (1MB) | Preview

Abstract

Respiratory motion degrades quantitative and qualitative analysis of medical images. Estimation and hence correction of motion commonly uses static correspondence models between an external surrogate signal and internal motion. This work presents a patient specific respiratory motion model with the ability to adapt in the presence of irregular motion via a Kalman filter with Expectation Maximisation for parameter estimation. The adaptive approach introduces generalizability allowing the model to account for a broader variety of motion. This may be required in the presence of irregular breathing and with different sensors monitoring the external surrogate signal. The motion model framework utilizing an adaptive Kalman filter approach is tested on dynamic MRI data of nine volunteers and compared to a state-of-the-art static total least squares approach. Results demonstrate the framework is capable of reducing motion to the order of < 3mm and is significantly (p < 0:001) more effective in the presence of irregular motion, assessed using the F test for model comparison. Utilizing the total sum of squares of estimated vector field error from the calculated ground truth, we observe approximately a fifty percent reduction in root mean square error and thirty percent reduction in standard deviation utilizing the Kalman model (EKF) in comparison to a static counterpart.

Item Type: Article
Divisions : Faculty of Engineering and Physical Sciences > Electronic Engineering > Centre for Vision Speech and Signal Processing
Authors :
NameEmailORCID
Smith, Rhodri L
Rahni, Ashrani Aizzudin Abd
Wells, KevinK.Wells@surrey.ac.uk
Date : 2 November 2018
DOI : 10.1109/TRPMS.2018.2879441
Copyright Disclaimer : 2469-7311 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Uncontrolled Keywords : Machine learning; Respiratory motion correction; Bayesian inference; Kalman filtering; Optimization
Depositing User : Diane Maxfield
Date Deposited : 22 Feb 2019 17:01
Last Modified : 28 Mar 2019 11:55
URI: http://epubs.surrey.ac.uk/id/eprint/850556

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year


Information about this web site

© The University of Surrey, Guildford, Surrey, GU2 7XH, United Kingdom.
+44 (0)1483 300800