Galen Durant Reed1, John Kurhanewicz1, Daniel B. Vigneron1, Susan M. Noworolski1
1Department of Radiology and Biomedical Imaging, University of California San Francisco, San Francisco, CA, USA
Automated detection of prostate cancer using the metabolite peak integrals of MRSI data is notoriously difficult due to the spatial inhomogeneity of the endorectal surface coil. A semi-automatic correction algorithm which normalizes MRSI data to the coils analytic field map was applied to three phantoms and 18 prostate cancer patients 3T MRSI data. Spectral peaks showed increased spatial homogeneity: 39% and 30% for suppressed water and citrate (phantoms) and 30% for suppressed water (patients). Additionally, maps of coil-corrected choline integrals showed potential in identifying anterior prostate carcinomas which were not clearly identifiable on un-normalized metabolite maps or metabolite ratio maps.