doi:10.1016/j.ultrasmedbio.2008.01.018 ● Original Contribution ADAPTIVE CLUTTER REJECTION FOR 3D COLOR DOPPLER IMAGING: PRELIMINARY CLINICAL STUDY YANG MO YOO,* SIDDHARTHA SIKDAR,* KEREM KARADAYI, † ORPHEUS KOLOKYTHAS, ‡ and YONGMIN KIM* †‡ *Departments of Bioengineering, † Electrical Engineering and ‡ Radiology, University of Washington, Seattle, WA, USA (Received 16 August 2007; revised 11 December 2007; in final form 28 January 2008) Abstract—In three-dimensional (3D) ultrasound color Doppler imaging (CDI), effective rejection of flash artifacts caused by tissue motion (clutter) is important for improving sensitivity in visualizing blood flow in vessels. Since clutter characteristics can vary significantly during volume acquisition, a clutter rejection tech- nique that can adapt to the underlying clutter conditions is desirable for 3D CDI. We have previously developed an adaptive clutter rejection (ACR) method, in which an optimum filter is dynamically selected from a set of predesigned clutter filters based on the measured clutter characteristics. In this article, we evaluated the ACR method with 3D in vivo data acquired from 37 kidney transplant patients clinically indicated for a duplex ultrasound examination. We compared ACR against a conventional clutter rejection method, down-mixing (DM), using a commonly-used flow signal-to-clutter ratio (SCR) and a new metric called fractional residual clutter area (FRCA). The ACR method was more effective in removing the flash artifacts while providing higher sensitivity in detecting blood flow in the arcuate arteries and veins in the parenchyma of transplanted kidneys. ACR provided 3.4 dB improvement in SCR over the DM method (11.4  1.6 dB versus 8.0  2.0 dB, p < 0.001) and had lower average FRCA values compared with the DM method (0.006  0.003 versus 0.036  0.022, p < 0.001) for all study subjects. These results indicate that the new ACR method is useful for removing nonstationary tissue motion while improving the image quality for visualizing 3D vascular structure in 3D CDI. (E-mail: ykim@u.washington.edu) © 2008 Published by Elsevier Inc. on behalf of World Federation for Ultrasound in Medicine & Biology. Key Words: 3D ultrasound color Doppler imaging, Adaptive clutter rejection, Kidney transplant, Flow signal- to-clutter ratio, Fractional residual clutter area. INTRODUCTION Three-dimensional (3D) color Doppler imaging (CDI) has potential to be a useful tool for noninvasively eval- uating complex vascular morphology (Ohto et al. 2005; Alcazar 2006; Baxter 2003; Rajiah et al. 2006). How- ever, before 3D CDI can be widely accepted in clinical practice, several limitations need to be overcome, espe- cially low image quality due to flash artifacts caused by the clutter originating from surrounding tissues and slowly-moving vessel walls (Wu et al. 1998; Nelson et al. 2001; Bailey et al. 2001; Rygh et al. 2006). These artifacts not only lower image quality but also negatively impact clinical productivity. For example, Raine-Fen- ning et al. (2004) found that the volume acquisition process during 3D CDI examinations had to be repeated in 22.5% of cases because of excessive flash artifacts. In CDI, a static clutter rejection method is typically used for suppressing flash artifacts. Various clutter fil- ters, such as finite impulse response (FIR), infinite im- pulse response (IIR) and regression, have been proposed for clutter rejection (Hoeks et al. 1991; Chornoboy 1992; Kadi and Loupas 1995). However, since the clutter char- acteristics in CDI can vary significantly due to nonsta- tionary tissue motion, usually caused by cardiac pulsa- tion, respiration and transducer/patient movement, a static rejection method cannot effectively suppress all the clutter. For effectively removing the nonstationary clutter, several adaptive clutter rejection methods have been proposed (e.g., the down-mixing (DM) and eigenvector regression filtering). In the DM method (Thomas and Hall 1994), the mean clutter velocity (or Doppler fre- Address correspondence to: Yongmin Kim, Professor, Department of Bioengineering, University of Washington, Box 355061, Seattle, WA 98195-5061, USA. E-mail: ykim@u.washington.edu Ultrasound in Med. & Biol., Vol. 34, No. 8, pp. 1221–1231, 2008 © 2008 Published by Elsevier Inc. on behalf of World Federation for Ultrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/08/$–see front matter 1221