Semi-automatic motion compensation of contrast-enhanced ultrasound images from abdominal organs for perfusion analysis Sebastian Schäfer a,n , Kim Nylund b , Fredrik Sævik c,b , Trond Engjom d,b , Martin Mézl e , Radovan Jiřík f,g , Georg Dimcevski d,b , Odd Helge Gilja b,c , Klaus Tönnies a a Department of Simulation and Graphics, University of Magdeburg, Universitätsplatz 2, 39106 Magdeburg, Germany b National Centre for Ultrasound in Gastroenterology, Haukeland University Hospital, Bergen, Norway c Department of Clinical Medicine, University of Bergen, Norway d Section of Gastroenterology, Department of Medicine, Haukeland University Hospital, Bergen, Norway e Department of Biomedical Engineering, Brno University of Technology, Czech Republic f Institute of Scientific Instruments of the Academy of Sciences of the Czech Republic, Brno, Czech Republic g International Clinical Research Center – Center of Biomedical Engineering, St. Anne's University Hospital Brno, Czech Republic article info Article history: Received 7 April 2014 Accepted 30 September 2014 Keywords: Ultrasonography Motion analysis Motion compensation Registration CEUS Contrast-enhanced ultrasound Perfusion Perfusion modeling abstract This paper presents a system for correcting motion influences in time-dependent 2D contrast-enhanced ultrasound (CEUS) images to assess tissue perfusion characteristics. The system consists of a semi- automatic frame selection method to find images with out-of-plane motion as well as a method for automatic motion compensation. Translational and non-rigid motion compensation is applied by introducing a temporal continuity assumption. A study consisting of 40 clinical datasets was conducted to compare the perfusion with simulated perfusion using pharmacokinetic modeling. Overall, the proposed approach decreased the mean average difference between the measured perfusion and the pharmacokinetic model estimation. It was non-inferior for three out of four patient cohorts to a manual approach and reduced the analysis time by 41% compared to manual processing. & 2014 Elsevier Ltd. All rights reserved. 1. Introduction Ultrasound (US) imaging is one of the most commonly available medical imaging techniques. It is cost-efficient compared to other imaging modalities, radiation free and portable. In the last decade, significant research has been conducted in the field of US contrast agents (CA) and associated acquisition protocols [1,2]. Contrast-enhanced ultrasound (CEUS) is commonly used to assess perfusion in various organs such as the heart, liver, kidney, spleen, pancreas or bowel [3–5]. Gas-filled microbubbles serve as true intra-vascular tracers as they do not leak into extra-vascular space and their non-linear echo can be separated from the tissue echo (see Fig. 1). Thus, they enable the measurement of absolute perfusion parameters such as blood volume, mean transit time and blood flow in an organ per unit time [6,7]. It is crucial to know the actual perfusion of an organ in many processes such as assessing ischemia, characterizing tumors and separating inflammatory from fibrotic processes [8–10]. Absolute perfusion can be quantified using pharmacokinetic modeling where the CA-concentration time curve is approximated by an appropriate model [6,11]. CEUS image sequences are subject to limitations interfering with correct analysis of the data. The most noticeable interference in addition to noise and speckle artifacts is motion [12]. Motion originates from various sources and can be subdivided into intrinsic and extrinsic sources [13]. Intrinsic sources are breathing, perfusion or digestive activity resulting in moving organs or tissue. Extrinsic sources include the tilting or moving of the US probe. This happens unintentionally due to breathing or patient motion. However, this also occurs intentionally when the examiner tries to adapt the field of view of the transducer to image a region of interest (ROI). This adaption is often necessary at the time of CA arrival when enhancement becomes visible. While breathing or tilting of the probe mostly causes linear motion shifts, organ motion such as aorta pulsation or the heartbeat results in non-linear deformation effects. Valid spatial correspondences of tissue over time cannot be established if the images of the sequences are influenced by such motion [14]. A specific problem of 2D image acquisition is that motion can emerge in directions perpendicular to the imaging plane (3D motion) so that frames are acquired out-of-plane. Whereas motion occurring within the imaging plane can be compensated by Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/cbm Computers in Biology and Medicine http://dx.doi.org/10.1016/j.compbiomed.2014.09.014 0010-4825/& 2014 Elsevier Ltd. All rights reserved. n Corresponding author. Tel.: þ49 391 67 51441. E-mail address: sebastian.schaefer@ovgu.de (S. Schäfer). Computers in Biology and Medicine 63 (2015) 229–237