Automated Lipid-Removal for Baseline Correction of Prostate-Cancer MRSI Data using Prior Knowledge.

Alan James Wright¹, Arend Heerschap¹

MRSI is a promising technique for the detection and localization of tumours in patients with prostate cancer, the clinical relevance of this data may be realised by the application of Pattern recognition techniques. The performance of such algorithms is partly dependent on artifacts such as residual large lipid signals from the tissue surrounding the prostate; these signals are broad and can overlap with the citrate resonances at 2.6ppm that play a key role in cancer diagnosis by MRSI. We present an algorithm that uses prior knowledge to remove lipid signals from prostate MRSI data sets.