Extraction of liver parenchyma is an important step in the evaluation of R2*-based hepatic iron content. Traditionally this is performed by radiologists via thresholding, an iterative, time consuming process susceptible to interrater variability. Here, a new method based on Frangi vesselness filtering is presented that automatically removes hepatic vessels from liver tissue, thereby reducing potential bias and improving workflow. The method was evaluated in digital liver phantoms and 42 iron overloaded patients. Excellent agreement between manual and automated liver parenchyma segmentation was found which demonstrates accuracy and robustness of the proposed method.