EVALUATION OF THE DETECTABILITY OF BREAST-CANCER LESIONS USING A MODIFIED ANTHROPOMORPHIC PHANTOM

During the development and characterization of imaging technology or n ew imaging protocols, it is usually instructive to perform phantom exp eriments. Often, very simplified forms of the realistic patient anatom y are used that may be acceptable under certain conditions; however, t he implications for patient studies can be misleading. This is particu larly true in breast and axillary node imaging. The complexities prese nted by the anatomy, variable object scatter, attenuation and inhomoge neous distribution of activity in this upper thoracic region provide a significant challenge to the imaging task. Methods: A tissue-equivale nt anthropomorphic phantom of the thorax (Radiology Support Devices, I nc., Long Beach, CA) containing fillable cavities and organs was modif ied for the studies, The phantom was filled with realistic levels of F DG activity and scanned on a Siemens ECAT HR+ whole-body PET scanner. Breast attachments containing 2.0- and 2.55-cc lesions with lesion-to- background ratios of 5:1 and 7:1, respectively, were imaged. Scatter a nd attenuation effects were analyzed with various experimental setups. A lymph node experiment and a multibed position whole-phantom scan al so were performed to illustrate the extent to which the phantom repres ents the human thorax. Results: Regions of interest were drawn on the lesions as well as the background breast tissue in all studies. It was found that the signal-to-noise ratio decreased 65% when a more realis tic phantom (lesions plus breasts plus thorax, all containing activity ) was used, as compared to a simple phantom (lesions plus breasts cont aining activity; no thorax), due to the effects of increased scatter a nd attenuation. A 23% decrease in the contrast also was seen from the scan of the more realistic phantom due to surrounding activity from ne arby organs such as the heart, as well as an increase in the volume of attenuating media. Conclusion: This new phantom allows us to more rea listically model the conditions for breast and lymph node imaging, lea ding to preclinical testing that will produce results that better appr oximate those that will be found in vivo, The phantom will be a valuab le tool in comparing different imaging technologies, data collection s trategies and image reconstruction algorithms for applications in brea st cancer using PET, SPECT or scintimammography systems.