Microwave imaging has seen an increasing amount of clinical trials in the last two years, including ongoing and planned commercial development. The rapidly increasing number of studies with patient imaging is encouraging researchers to focus on the challenges of translating microwave imaging algorithms to clinical use. A large number of imaging algorithms have been proposed for radar-based imaging, however few detailed comparisons of these algorithms using realistic experimental or patient data have been published. This paper looks at two leading radar-based algorithms in experimental phantoms with and without tumours present. It shows that although some algorithms improve image quality when the tumour is present, the algorithms can also increase false positives in breast phantoms containing glandular structures.