Real-time severe weather algorithms that are used to identify various storm attributes can be adversely affected by the presence of meteorological and non-meteorological contaminants such as anomalous propagation (AP), ground clutter (GC), clear-air return or biological scatters in the radar reflectivity data. We examine the Quality Control Neural Network, a new algorithm which classifies precipitation and non-precipitation returns from radar data and provides reflectivity tilts where the majority of contaminants are removed. We demonstrate that using the reflectivity tilts from the QCNN rather than the unedited reflectivity data improves the skill of the NSSL Mesocyclone Detection Algorithm (MDA). In order to determine a positive effect at classifying radar echoes, the MDA is run both without and with the QCNN filtering the original data. Results using 15 nationwide storm events show that the application of the QCNN effectively removes false MDA detection in clear air return while essentially not impacting the ability to detect mesocyclones in precipitation and storm regions.