Toshiba Medical Systems Corporation Research Articles

Catheter ablation of atrial fibrillation guided by registered computed tomographic image of the atrium

A 60-year-old man was referred to our institution for catheter ablation of symptomatic drug-refractory paroxysmal atrial fibrillation. Contrast-enhanced computed tomography (CT) was performed in the morning of the procedure using a 64-slice CT scanner (Aquillon, Toshiba Medical Systems Corporation, Tochigi, Japan). Axial slices of the chest were reconstructed and imported into an electroanatomic mapping system (Carto XP, Biosense Webster, Inc., Diamond Bar, CA, USA). Using a new software package (CartoMerge, Image Integration Module), the left atrium (LA) was segmented from other cardiac structures, and its endocardial shell was extracted for later image registration.

MO-D-I-611-04: Region-Of-Interest (ROI) Computed Tomography: Combining Dual Resolution XRII Images

Purpose: High‐resolution computed tomographic systems can require large radiation dose for full‐field‐of‐view (FFOV) reconstructions. We have developed a method for obtaining data inside an ROI at high‐dose and high‐resolution and combining it with low‐dose, low‐resolution data outside the ROI. The combined image is reconstructed to obtain improved image quality both inside and outside the ROI but at reduced integral dose. Method and Materials: Projection images of a head phantom were obtained using the 5″ and 12″ modes of an x‐ray image intensifier. The 5″ mode effectively defined an ROI and the 12″ mode encompassed the whole object (i.e., FFOV), but at 1/4 the dose of the 5″ mode. The 5″ and 12″ projection images were aligned and combined. Since the data outside the ROI was obtained at a lower dose, the intensity inside and outside the ROI in the combined image was equalized prior to backprojection. The mapping function was determined, using the pixels lying just inside and outside the ROI. A second approach was simulated by assigning the average pixel value in the ROI to all pixels outside the ROI. Both cases were reconstructed using a Feldkamp cone‐beam algorithm. Results: For both cases, we obtained reconstruction comparable to FFOV reconstruction inside the ROI. However, the combined image produced a reconstructed image comparable to the FFOV reconstruction outside the ROI as well. Moreover, the resolution of the region inside the ROI in the reconstructed image was determined by the 5″ mode as opposed to the 12″ mode. Conclusion: ROI CT using two sets of data provides a way to reconstruct an ROI with greatly reduced integral dose and artifacts and yet improved spatial resolution. Supported by: NIH R01NS38746, R01NS43924, R01HL52567, R01EB002916, and R01EB002873 and the Toshiba Medical Systems Corporation).

TH-C-I-611-03: Three-Dimensional Reconstruction of An Asymmetric Vessel Phantom Using Two X-Ray Projections

Purpose: Vascular lumen shape often predicts potential lumen occlusion and inadequate blood supply. Three‐dimensional (3D) lumen information can facilitate diagnosis and interventional therapies. We present a technique for 3D reconstruction of asymmetric vessel lumens using two x‐ray projections. Method and Materials:Images were acquired of a contrast‐filled vessel phantom with an asymmetric lesion. For two‐view reconstruction, lumen diameters, lesion geometric characteristics in each view, and correspondence between views were determined using previously described methods. Intensity profiles across the vessel lumen were extracted along epipolar lines. The profiles were fit using an elliptical model, and the deviation from ellipticity was taken to correspond to intruding plaque. The fitted profiles from the two views were used to generate elliptical cross sections, and the plaque information was used to deform these cross sections asymmetrically. The 3D lumen was created by combining individual 2D cross sections along the vessel length. To verify the method, 3D reconstructions were obtained in two ways from a set of digital subtraction angiography(DSA)images of the phantom obtained using a high‐resolution microangiographic detector (43 micron pixels): (a) using two orthogonal projections and our two‐view method, and (b) using 180 projections and a Feldkamp algorithm. The two reconstructed lumens were superimposed and the root‐mean‐square (RMS) distance between the lumen boundaries and percent overlap of the lumen volumes were calculated. Results: The RMS distance between the boundaries of the two‐view and the micro‐CT reconstructed lumens was less than 0.2 mm over the entire length of the vessel which had a diameter of 3.6 mm in non‐lesion segments; the volume overlap was 94%. Conclusion: Accurate 3D reconstruction of an asymmetrically stenosed vessel lumen was obtained using two radiographic projections as verified by comparison with microangiographic cone‐beam CTreconstruction. Supported by NIH EB002873, HL52567, NS43924 and Toshiba Medical Systems Corporation.

SU-FF-I-63: Reproducibility of Guidewire Positioning and Stent Path for Endovascular Interventions

Purpose: Clinicians often have difficulty passing guidewires and stents during endovascular interventions. As a first step toward facilitating guidewire and stent guidance, we have investigated the reproducibility of paths that guidewires and stents take as they pass through the access vessels. Method and Materials: An internal carotid vessel phantom having tortuousity similar to the carotid siphon was constructed from 3 mm tubing and encased in Sylgard elastomer. A guidewire was repeatedly passed through the phantom. Each time, images were acquired for several views (9′ II mode, 225 micron pixel size). In addition, an undeployed stent (length 30 mm, diameter 1.3 mm) was repeatedly passed over the guidewire while images were acquired at 10 fps again using different views, without repositioning of the guidewire. Pairs of views (0° LAO, 82° LAO) were selected for biplane reconstruction. The centerlines of the guidewire were manually indicated and fit with a spline in each image. The correspondence between the indicated centerlines is established using the epipolar constraints, and 3D centerline of the guidewire and stent path were reconstructed. The reproducibility of the guidewire and stent positions were then determined after aligning the images (to remove jitter) and the 3D centerlines. Results: The average repeated-indication difference for the guidewire was 0.4 ± 0.1mm (2D) and 0.5 ± 0.1mm (3D). The average repeated-positioning difference for the guidewire was 0.4 ± 0.2mm (2D) and 0.4 ± 0.1mm (3D). The average guidewire-versus-stent-path difference was 0.5 ± 0.2mm (2D) and 0.4 ± 0.04mm (3D). Conclusion: Guidewire positions are very stable, and stent passage appears to affect guidewire position near the vessel wall, indicating that physical/mechanical models may be used to determine guidewire and stent-pathways. Support from NIH R01 EB02916 and Toshiba Medical Systems Corporation.