Development and Optimization of a Dedicated Dual-Modality SPECT-CT System for Improved Breast Lesion Diagnosis
X-ray mammography is the most widely used breast cancer imaging technique. However, over 400,000 women newly diagnosed each year are misdiagnosed and undergo a biopsy. Current mammography techniques are limited by: (1) low image contrast, especially in women with dense breasts; (2) difficulty in diagnosing and detecting lesions close to the chest wall and in women with radiographically dense breasts; (3) structural overlap onto a two-dimensional (2D) image plane; and (4) patient discomfort due to breast compression. Therefore, three-dimensional (3D) tomographic breast imaging approaches for pendant, uncompressed breasts have been explored to overcome these limitations and improve the detection of breast lesions. The goal of this thesis is to characterize and implement a dual-modality SPECT-CT dedicated breast imaging system that can overcome these limitations and integrate both metabolic and anatomical information to further improve the visual quality and quantitative accuracy over independent systems alone.
Initial work on this thesis started out with characterizing the modulation transfer function (MTF) in 3D for the independent dedicated SPECT and CT systems. Using a novel phantom to measure the MTF at different locations in a 3D reconstructed volume, results show that acquiring images with a step-and-shoot mode and with trajectories that meet the sampling criteria, uniform resolution throughout a 3D reconstructed volume is obtained.
The effects of sampling and system geometry on the reconstructed CT images are investigated. As expected, constraining the x-ray source and detector to a circular tilt yields insufficiently sampled reconstructed images, which contain geometric distortions, reconstruction inaccuracies, and cupping artifacts. Although beam hardening and scatter are considered to be the main causes of cupping artifacts in the reconstructed CT images, this study suggests that insufficient sampling might be a third cause to cupping artifacts in the reconstructed images. An additional finding in this study is that despite the insufficient sampling in the reconstructed CT images, high frequency objects (small size) are preserved more than low frequency objects (large size).
Using a lateral offset geometry (i.e. the entire system shifted such that the central ray of the cone-beam is at an offset with respect to the COR) in CT has also been shown previously to introduce circular and cylindrical artifacts in the reconstructed coronal and sagittal CT slices, respectively. Monte Carlo studies show that these artifacts are due to mechanical detector misalignment. However, cropping the projections, such that there is less of an overlap between conjugate projections, or placing the system in a centered geometry can eliminate these artifacts.
Next, the dual-modality SPECT-CT scanner is designed and built. The performance of this scanner is evaluated with geometric and anthropomorphic phantoms. Despite only nearly complete sampling from both systems, results illustrate that SPECT and CT images can be registered and fused with minimal error.
The feasibility of using the reconstructed CT images to quantify different tissue components is also investigated by using different materials (acrylic, delrin, polyethylene, and fat-equivalent and glandular-equivalent plastics) and a cadaver human breast. By implementing scatter correction using the beam stop approach, scatter corrected reconstructed images yield attenuation coefficient values to within 11% of their actual values.
Finally, few clinical studies are done to evaluate the effectiveness of the dual-modality scanner. Although the CT is currently limited in the amount of breast volume that can be imaged, reconstructed images appear to have minimal distortion and reconstruction inaccuracy. Fused SPECT-CT images also show the significance of using functional information from SPECT to help localize the lesion in the anatomical CT images.
The dual-modality SPECT-CT scanner has successfully demonstrated its capability to uniformly sample an uncompressed breast with 3D complex trajectories that meet the sampling criteria and provide tissue quantification and localization information. This system will be a clinically useful imaging tool in detecting cancer, especially in women with high risk of breast cancer, monitoring treatment therapies, and improving surgical biopsy guidance.
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License.
Rights for Collection: Duke Dissertations
Works are deposited here by their authors, and represent their research and opinions, not that of Duke University. Some materials and descriptions may include offensive content. More info