Spatiotemporal variations in the magnitude of completeness Mc make it challenging to confidently assess seismic hazard or even to simply compare earthquake rates between regions. In this study, we introduce new techniques to correct for heterogeneous Mc in a treatment of the eastern and Atlantic Canada earthquake catalog (1985--2022). We first introduce new methodology to predict Mc(x,t) based on the distribution of seismometers. Second, we introduce a modified maximum-likelihood estimator (MLE) for b (the b-value) that accounts for spatiotemporal Mc variation, allowing the inclusion of more earthquakes. Third, we compute the ratio of detected/predicted M>1 earthquakes as a function of Mc and apply it to create a calibrated M>1 event-rate map. The resulting map has advantages over a moment-rate map, which is effectively sensitive only to the very largest earthquakes in the dataset. The new MLE results in a modestly more precise b when applied to the Charlevoix Seismic Zone, and a substantial increase in precision when applied to the full Atlantic Canada region. It may prove useful in future hazard assessments, particularly of regions with highly heterogeneous Mc and relatively sparse catalogs.