Over the past ten years, Ultra Wideband (UWB) Radar has been widely investigated as a biomedical imaging modality, used to detect early-stage breast cancer and to continuously monitor vital signs using both wearable and contactless devices. The advantages of the technology in terms of low-power requirements and non-ionising radiation are well recognised, with the technology being applied to a range of non-invasive medical applications, from respiration to heart monitoring. Across all these applications, there is a strong necessity to efficiently manage the large quantities of UWB data which will be captured. For wearable devices in particular, the efficient compression of UWB data allows the monitoring system to conserve limited resources such as memory and battery capacity, by reducing data storage and in some cases transmission requirements. In contrast to lossless compression techniques, lossy compression algorithms can achieve higher compression ratios and consequently greater power savings, at the expense of a marginal degradation of the reconstructed signal. This paper compares the lossy JPEG2000 and Set Partitioning In Hierarchical Trees (SPIHT) algorithms for UWB signal compression. This study examines the effects of lossy signal compression on an UWB breast cancer classification algorithm. This particular application was chosen because the classification algorithm relies heavily on shape and surface texture detail embedded in the Radar Target Signature (RTS) of the tumour, and therefore will provide both a robust and easily quantifiable test platform for the compression algorithms. The study will evaluate the performance of the classification algorithm as a function of Compression Ratio (CR) and Percentage Root-mean-square Difference (PRD) between the original and reconstructed UWB signals.