Search

The results from a single patient setup with simulated reconstruction results are presented as a proof of concept. For this patient it was decided that a roughly hemi-thoracic ROI would be ideal for providing sufficient information for patient setup and sparing the contralateral breast. AP aligned parallel chords were chosen for the reconstruction geometry. We found that by shifting approximately 7 cm both anteriorly and to the patient's left, we are able to image over approximately a 190 degree arc. The collision prediction model is shown in Figure 1. In Figure 3 the full data sinogram is shown with the data ranges required for a traditional full-fan/half-scan, a half-fan/full-scan and the proposed conformal ROI technique. A summary of results is given in Table I. Note that, with this patient setup, a typical half-fan/full-scan geometry would cause a patient collision. With a symmetric detector full-fan/half-scan geometry the collision could be avoided but would require an 82 cm wide detector, more then double what is on the current system.

MR signals can be acquired with high spatial and spectral resolution so that each pixel is represented by a high-resolution proton spectrum of water and fat. The spectra can be analyzed to produce images with improved contrast, anatomic accuracy, and sensitivity to contrast media uptake, compared to conventional MR imaging. Analysis of high spectral and spatial resolution (HiSS) MR datasets must take the complex structure of water and fat proton signals in vivo into account. This complexity can greatly increase the information content of HiSS MR images when appropriate data analysis methods are used. We evaluate two methods that effectively increase spectral resolution of HiSS data and, thus, improve image contrast, signal-to-noise ratio, and edge delineation. We describe methods that reduce artifacts due to truncation of the free induction decay, identify maxima of complex proton signals associated with each image voxel, and generate images derived from pure absorption spectra of the water resonance. We demonstrate improvements in image quality obtained with these methods and use them to distinguish between metastatic and non-metastatic rodent tumors based on numerical measures of image texture. Related approaches may significantly improve sensitivity and specificity of clinical MRI.