This letter investigates the feasibility of applying contrast agents for lesion classification in ultra wideband (UWB) breast imaging. Previous study has focused on distinguishing benign from malignant masses by exploiting their morphology-dependent backscatter signature via the complex natural resonances of the late-time target response. The tissue differentiation capability, however, deteriorates severely if the intrinsic contrast between the dielectric properties of dysplastic and normal tissues are small. A possible solution to this problem is proposed in this letter via the use of microwave contrast agents, where the damping factors of the differential backscatter responses before and after the infusion of contrast agents to a dysplastic inclusion are used to correlate with the anomaly shapes. The feasibility of this approach for lesion classification is demonstrated through comprehensive simulation studies using realistic numerical breast models.