Abstract P4-03-08: A new real-time image fusion technique, a coordinated sonography and MRI using magnetic position tracking system, improves the sonographic identification of enhancing lesions in breast MRI

Abstract

Abstract Breast magnetic resonance imaging (MRI) is a method that can detect clinically and mammographically occult cancer foci. For lesions suspected to be malignant on the basis of breast MRI, targeted sonography can help determine whether the lesion is amenable to tissue biopsy using sonographic guidance. However, the reproducibility of sonography as well as interexaminer reliability in targeted sonography is still a clinical issue. Recently, we have developed real-time virtual sonography (RVS) that can coordinate a sonographic image and a MRI multiplanar reconstruction (MRI-MPR) of the same section in real time. Although RVS has been reported to be useful for breast cancers, only a few studies have investigated the size and positioning error between a real-time sonographic image and a MRI-MPR image. The purpose of this study was to evaluate the accuracy of RVS to sonographically identify enhancing lesions by breast MRI. Between December 2008 and May 2009, RVS was performed in 51 consecutive patients with 63 enhancing lesions. MRI was performed with the patients in the supine position using a 1.5-T imager with a body surface coil to achieve the same position as with sonography. To assess the accuracy of the RVS, the following three issues were analyzed: (i) The sonographic detection rate of enhancing lesions, (ii) the comparison of the tumor size measured by sonography and the MRI-MPR and (iii) the positioning errors as the distance from the actual sonographic position to the expected MRI position in 3-D. Among the 63 enhancing lesions, 42 (67%) lesions were identified by conventional B-mode, whereas the remaining 21 (33%) initial conventional B-mode occult lesions were identified by RVS alone. The sonographic size of the lesions detected by RVS alone was significantly smaller than that of lesions detected by conventional B-mode (p < 0.001). The mean tumor size provided by RVS was 12.3 mm for real-time sonography and 14.1 mm for MRI-MPR (r = 0.848, p < 0.001). The mean positioning errors for the transverse and sagittal planes and the depth from the skin were 7.7, 6.9 and 2.8 mm, respectively. The overall mean 3D positioning error was 12.0 mm. Our results suggest that RVS has good targeting accuracy to directly compare a sonographic image with MRI results without operator dependence. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P4-03-08.