ICANCERRESEARCH(SUPPL.)55. 5759s-5763s, December 1, 1995] Fusion of Immunoscintigraphy Single Photon Emission Computed Tomography (SPECT) with CT of the Chest in Patients with Non-Small Cell Lung Cancer1 Sanjeev Katyal, Elissa Lipcon Kramer,2 Marilyn E. Noz, Dorothy McCauley, Abraham Chachoua, and Alan Steinfeld New York University' School of Medicine, [S. K.], Division of Nuclear Medicine fE. L K., M. E. N.]; New York University Kaplan Cancer Center, Department of Radiology (E. L K., A. C., A. S.]. Division of Oncology, Department of Medicine [A. C.]; and Division of Radiation Oncology, Department of Radiology [A. S.], New York University Medical Center/Bellevue Hospital Center, New York, New York 10016 Abstract quently are used to detect and localize abnormal masses or enlarged lymph nodes. However, they cannot differentiate reliably tumors from In non-small cell lung cancer (NSCLC), accurate staging is critical in benign masses or malignant lymphadenopathy from hyperplastic ad deciding between potentially curative surgery and paffiative treatment. enopathy seen in NSCLC. Also, when the normal anatomy of the Image registration, or fusion, combines the unique functional information patient is altered, i.e., in a postsurgical patient, detection of recurrent provided by SPECT imaging with the excellent anatomic detail offered by cancers may be difficult using only CT or MRI imaging. computed tomography (CT) or magnetic resonance hnaging to better characterize the imformation provided by each separate modality. In this In NSCLC, accurate, nomnvasive staging is important in deciding study, we explored the role of fusion ofimmunoscintigraphy SPECT with between potentially curative surgical resection and palliative treat CT in the staging of NSCLC. We fused chest CT with @Tc-labeled ment. This may also be important to the surgeon in planning surgical IMMU-4 anti-carcineembryomc antigen Fab' antibody fragment SPECT treatment or exploration. The combination of tissue-specific informa in 14 patients with NSCLC using a landmark-basedalgorithm.The tion with anatomic information becomes useful in the staging of algorithm's accuracy was a measure from the center-to-center distance NSCLC to better characterize the information provided on each sep and the percentage overlap of two regions of interest: one drawn on CT arate modality. Image registration, or fusion, precisely matches the and warped onto SPECT, the other drawn directly on the SPECT. We data from different imaging modalities and enhances the information found that the average center-to-center distance was 1.3 ±0.8 pixels. provided by both types of modalities. In the past, we have used this Average overlap was 46 ±20%. CT-SPECT fusion helped differentiate image registration algorithm to analyze radiolabeled anti-carcinoem tumor from normal blood pool, necrotic areas within viable tumor, tumor bryonic antigen antibody scans in patients with colorectal cancer (9). recurrence from scar, and malignant lymphadenopathy from hyperplasia. We conclude that fusion of CT and SPECT augments the information Using fusion, it is now possible to simultaneously examine functional provided by each separate modality. Future clinical applications of fusion and structural details of cross-sectional images in the chest (10—12). in NSCLC staging using immunoscintigraphy appear promising. In this study, we have fused CT with 99mTc@labeledIMMU-4 Fab' antibody fragment SPECT images for 14 patients with NSCLC to Introduction explore the potential role of image registration in the diagnosis and noninvasive staging of NSCLC. The diagnostic and staging evaluation of a patient with NSCLC3 relies on a variety of imaging modalities. SPECT imaging techniques Materials and Methods combined with immunoscintigraphy is a new technique that can provide unique functional information, i.e., the detection of abnormal Subjects concentrations of tumor antigen. Immunoscintigraphy has seen only Fourteen patients (7 men and 7 women) with biopsy-proven NSCLC were preliminary use in NSCLC to date (1—5).Immunoscintigraphy offers studied. The mean age of the patients was 69 years, with a range of 38—92.The improved sensitivity in detection of tumors such as colorectal carci histological types included poorly to well-differentiated bronchogenic adeno noma when used in conjunction with conventional imaging modalities carcinoma, broncheoalveolar carcinoma, mixed adenocarcinoma and squa (6). SPECT contributes to the sensitivity of this technique by enhanc mous carcinoma, and mixed adenocarcinoma and carcinoid. The clinical stage ing contrast and by providing more exact localization of scan abnor ranged from stage llb to stage IV. All patients were further evaluated by malities in colorectal cancer (7, 8) and in NSCLC (5). However, the medical history, physical examination, bone scan, and chest, brain, and ab dominal CT. Written informed consent was obtained according to the guide SPECT scan often is limited by poor spatial resolution with made lines of the Institutional Review Board. quate anatomic detail, i.e., although an abnormal area of increased uptake is noted, the precise location and adjacent structures frequently are not well visualized. Imaging Techniques One area of difficulty in the interpretation of SPECT images is in the differentiation of abnormal antibody uptake from normal blood Patients received 20—30mCi of @mTc@labeledIMMU-4anti-CEA Fab' antibody fragment (Immunomedics, Morris Plains, NJ). The fragment was containing structures (5). This is particularly true in the mediastinum, prepared by reduction of the F(ab')2 fragment of the IMMU-4 antibody, an where the prominent normal vascular activity may compromise the IgG,.c isotype specific for the Mr 200,000 epitope of CEA. The lyophilized utility of immunoscintigraphy SPECT in patients with NSCLC. On fragment was provided in kit form for radiolabeling. Radiolabeling was per occasion, other nonspecific localizations can be difficult to distinguish formed by mixing @“Tc-labeledpertechnetatein saline with 1.25 mg lyoph from abnormal areas of uptake. ilized fragments for 5 rain. Instant TLC in saline was performed to determine In contrast to SPECT images, structural imaging modalities such as the radiolabeling efficiency. This averaged 96.8 ±3.7%. Immunoreactivity CT or MRI provide excellent anatomic detail. These modalities fre was determined using a CEA affinity column and was found to average 61 ± 8.1%. The antibody was administered iv. over 20—30mm. For the SPECT of the chest, 57Co markers were placed at the coracoid I Presented at the “Fifth Conference on Radioimmunodetection and Radioimmuno therapy of Cancer,―October 6—8,1994, Princeton, NJ. Supported by grants from Immu processes, sternal notch, and xyphoid process. SPECT images were acquired nomedics, Inc., Morris Plains, NJ, and Sun Microsystems, Inc., Mountain View, CA. using a triple-headed rotating SPECT camera (Triad; Trionix Corp), fitted with 2 To whom requests for reprints should be addressed, at Division of Nuclear Medicine, low-energy high resolution collimators, and interfaced to a SUN SPARC 390 New York University Medical Center, 560 First Avenue, New York, NY 10016. (SUN Microsystems). Planar and SPECT imaging were performed either at 3 The abbreviations used are: NSCLC, nonsmall cell lung cancer; SPEC'F, tingle photon emission computed tomography; CT, computed tomography; MRI, magnetic 5—8and22—24hor at 16—18hafter iv. administration of the radiolabeled resonance imaging; CEA, carcinoembryonic antigen; ROl, region of interest. antibody. Acquisitions were performed with dual energy windows: a 20% 5759s Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 1995 American Association for Cancer Research. FUSION OF CHEST SPECI' AND CT IN NSCLC energy window centered at 140 KeV for the antibody images (@“Tc)anda 4% window centered at 122 KeV for the markers (57Co). One hundred twenty projections were acquired at 25—30s/view, over a 360°interval, into either a 64 X 64 or 128 x 128 matrix. This yielded two simultaneously acquired sets of images: one showing the radioimmunoconjugate distribution and the other, the radioactive markers. The projections were reconstructed into 3.5-mm thick ( 5, transaxial sections by using filtered back projection with a Hamming filter, 0.7 I 4 cyc/cm Nyquist and 0.4—0.55cyc/cm high cut frequency, and Chang's method of attenuation correction. )1 All CT studies of the chest were performed after the administration of oral and iv. contrast. The iodinated iv. contrast material was administered as two bolus infusions for dynamic incremental CT scanning. A total of 200 ml of 43% contrast material was delivered by a power injector (Angiomat CT injector: Liebel-Flarsheim, Cincinnati, OH) in a 50-mi bolus (2.5 ml/s), fol lowed by a 150-ml infusion (1 ml/s). Scanning, using either a GE 9800, High-Light Advantage or a High Speed (Helical) CT, commenced approxi mately 15 5 after initiation of the infusion phase. Contiguous axial 5—10-mm thick CT sections of the chest were obtained in a 512 X 512-pixel matrix. ( * ‘5' ‘S SPECT-CT Image Registration 4 The transaxial CT, SPECT antibody, and the SPECT markers slices were transferred offline, via Ethernet, to a Sun SPARC 670MP and converted to a standard AAPM/interfile format. Qsh, a hardware independent image display and handling toolkit composed of several software modules, was used to handle the standardization of image data from the contrasting modalities (SPECT and CT). To match pixel dimensions within the plane of reconstruc tion, the images were interpolated to a corresponding matrix size. Using Fig. 1. Landmarks are placed on the edge of activity on the SPECT slice representing the lungs (arrowheads), the body walls (open arrows), or an “intermediate―surfacethat standard bilinear interpolation, the 64 X 64 SPECT images were enlarged to corresponds to chest wall muscle on CT (closed arrows). a 128 X 128 format, and the 512 X 512 CT images were reduced to a 128 X 128 matrix format. The SPECT and CT slices were simultaneously displayed side-by-side on the same monitor with individualized color scales chosen for reference to the warped ROIs.