Measurement of thermally ablated lesions in sonoelastographic images using level set methods Benjamin Castaneda* a , Jose Gerardo Tamez-Pena b , Man Zhang a , Kenneth Hoyt a , Kevin Bylund c , Jared Christensen c , Wael Saad c , John Strang c , Deborah J. Rubens c , Kevin J. Parker a a Department of Electrical & Computer Eng., University of Rochester, Rochester, NY, USA 14627; b 20 Devonwood Ln, Pittsford, Rochester, NY, USA 14534; c Department of Imaging Sciences, University of Rochester Medical Center, Rochester, NY, USA 14642 ABSTRACT The capability of sonoelastography to detect lesions based on elasticity contrast can be applied to monitor the creation of thermally ablated lesion. Currently, segmentation of lesions depicted in sonoelastographic images is performed manually which can be a time consuming process and prone to significant intra- and inter-observer variability. This work presents a semi-automated segmentation algorithm for sonoelastographic data. The user starts by planting a seed in the perceived center of the lesion. Fast marching methods use this information to create an initial estimate of the lesion. Subsequently, level set methods refine its final shape by attaching the segmented contour to edges in the image while maintaining smoothness. The algorithm is applied to in vivo sonoelastographic images from twenty five thermal ablated lesions created in porcine livers. The estimated area is compared to results from manual segmentation and gross pathology images. Results show that the algorithm outperforms manual segmentation in accuracy, inter- and intra-observer variability. The processing time per image is significantly reduced. Keywords: Elasticity, sonoelastography, image processing, segmentation, region growing, fast marching methods, level set methods, Mumford-Shah functional 1. INTRODUCTION Thermal ablation techniques such as radiofrequency ablation (RFA) and high intensity focused ultrasound (HIFU) have attracted the interest of the research community for their capabilities to treat tumors as minimally invasive techniques [1,2]. In particular, promising results have been reported in early clinical trials for the treatment of hepatic tumors [3,4]. Imaging modalities that dynamically and precisely monitor the lesion during and after the treatment are crucial for the success of thermal ablation therapies. Ultrasound (US) is generally used for imaging guidance during the ablation procedures. It offers convenient real-time guidance of RFA needle placement, and it has the advantage of being cost-effective and readily available in most clinical sites. However, as an imaging modality to monitor the creation of the lesions, US did not exhibit good results [3]. Besides the low intrinsic contrast between treated and untreated tissues, artifacts due to the gas bubbles created during the treatment appear as hyper-echoic formations [5]. These formations do not represent accurately the extent of ablation. Gas bubbles resolve gradually, resulting in underestimation of the lesion size. On the other hand, MRI capabilities on resolving soft tissues can be used to discriminate thermally ablated from healthy tissue [6], but the procedure becomes expensive and time consuming. Contrast enhanced CT imaging has also been proposed [7]. In this modality, thermally ablated lesions are depicted as hypo-attenuating regions. However, it also presents disadvantages such as ionizing radiation exposure, CT contrast agents’ side effects, and extended time of the procedure. Thermally ablated lesions present an elevated elasticity modulus when compared to the surrounding tissue. Consequently, elasticity imaging modalities have been proposed as an alternative to monitor lesion creation and follow- up [8,9]. In particular, sonoelastography [10] is an imaging technique that estimates the peak displacement of the tissue under an externally induced mechanical harmonic excitation [11]. In a previous study, sonoelastography was used to detect and measure thermal lesions in vivo in exposed-liver experiments [5]. Manual measurement of these lesions in the sonoelastographic images is challenging due to diffuse boundary definition and artifacts formed by respiratory motion and perfusion. As a result, outlining and measuring the lesions becomes a time-consuming process with high variability. *castaned@ece.rochester.edu; phone 1 585 748-3744; fax 1 585 273-4919 Honorable Mention Poster Paper Medical Imaging 2008: Ultrasonic Imaging and Signal Processing, edited by Stephen A. McAleavey, Jan D'hooge, Proc. of SPIE Vol. 6920, 692018, (2008) · 1605-7422/08/$18 · doi: 10.1117/12.770811 Proc. of SPIE Vol. 6920 692018-1 Downloaded from SPIE Digital Library on 09 Dec 2010 to 128.151.164.33. Terms of Use: http://spiedl.org/terms