Hybrid Modality Fusion of Planar Scintigrams and CT Topograms to Localize Sentinel Lymph Nodes in Breast Lymphoscintigraphy: Technical Description and Phantom Studies

Abstract

Lymphoscintigraphy is a nuclear medicine procedure that is used to detect sentinel lymph nodes (SLNs). This project sought to investigate fusion of planar scintigrams with CT topograms as a means of improving the anatomic reference for the SLN localization. Heretofore, the most common lymphoscintigraphy localization method has been backlighting with a (57)Co sheet source. Currently, the most precise method of localization through hybrid SPECT/CT increases the patient absorbed dose by a factor of 34 to 585 (depending on the specific CT technique factors) over the conventional (57)Co backlighting. The new approach described herein also uses a SPECT/CT scanner, which provides mechanically aligned planar scintigram and CT topogram data sets, but only increases the dose by a factor of two over that from (57)Co backlighting. Planar nuclear medicine image fusion with CT topograms has been proven feasible and offers a clinically suitable compromise between improved anatomic details and minimally increased radiation dose.

Figure 1

Traditional breast lymphoscintigraphy technique with ⁵⁷Co-backlighting for localization. The four images on the left (A.I–A.IV) are emission images for a bi-lateral lymphoscintigraphic study. The current protocol for localizing sentinel nodes is to backlight the patient using a ⁵⁷Co sheet source to create a silhouette of the patient. The anterior (B.I), left lateral (B.II), and right lateral emission + ⁵⁷Co transmission images (B.III) are shown on the right.

Figure 2

Diagram of object magnification and minification for objects in different coronal planes relative to isocenter. Topograms differ from projection radiographs because the topogram is calibrated such that a “virtual” detector lies at a plane through the isocenter. The gray arrow represents a real object that is projected to isocenter during topogram reconstruction. This reconstruction technique causes objects that are physically closer to the X-ray source, relative to the gantry isocenter, to appear larger in the image, and vice versa when the object is physically closer than isocenter to the detectors (unlike projection radiographs, where objects closest to the detector are least affected by distortion).

Figure 3

Phantom Setup for Breast Lymphoscintigraphy Studies. The micro spheres used to represent the lymph nodes are applied directly to the surface of the thorax phantom. 0.5 cm of thin superflab is placed over the nodes, then the chest overlay and breast phantoms are positioned. The 4.0 mL sphere used for the injection site is then embedded in a 2-3 cm thick piece of superflab that is placed on one of the breast phantoms.

Figure 4

Measured spatial distortion of CT topogram. The error in lateral distance from isocenter (measured-true) of a point object (mammography bead) varied linearly as a function of both the lateral and coronal distance from isocenter.

Figure 5

Accuracy of lateral distortion compensation algorithm for planar scintigram-CT topogram registration. Fused images of the original (uncompensated) scintigram and CT topogram of three button sources (a), and stretched (compensated) scintigram and CT topogram (b). The vertical shift in the original fused images is due to a registration offset in the table travel direction of the scanner equal to the distance between the edge of the field of view of the gamma camera detector and the edge of the image matrix. This offset is accounted for in the compensation algorithm so the scintigram and topogram accurately register in the longitudinal direction of the scanner. On the right, the scintigram is compensated to correct the lateral position of the right-most button source. The residual error in lateral position of a given button source after compensation was within ±3.44 mm.

Figure 6

Comparison of scintigram-CT topogram method to conventional lymphoscintigraphy. Anterior (a) and lateral (b) views of ten-minute emission images, conventional localization images, and planar scintigrams fused with CT topograms, of an anthropomorphic thorax phantom with a simulated internal mammary SLN. The current acquisition and processing workflow provides the physician with a visual comparison of the conventional and proposed method of localization. The anterior view of the scintigram-topogram fused image (a-right) provides location information for the internal mammary lymph node relative to the ribs and sternal notches.