Radioluminescence microscopy (RLM) is a newly developed paradigm for visualizing radiotracer uptake in live cells. High-resolution images are obtained by capturing the scintillation of individual beta particles emitted from cells using a low-light microscope and processing the resulting frames to estimate the location of the corresponding decay events. We have developed a novel reconstruction method that enhances the quality of RLM images in many aspects, including spatial resolution and sensitivity, with no hardware modifications. The proposed method requires minimal user interaction leading to more accurate and reproducible results. The major improvements include automated segmentation of the ionization tracks with no user input required and enhanced ionization sourcepoint localization. These operations are designed to robustly handle image artifacts and a wide variety of ionization track shapes to guarantee high performance across datasets collected under varying conditions. Implementation of this method on real and synthetic datasets is shown to provide cell reconstructions with 14% higher signal-to-noise ratio (SNR), 18% better spatial resolution, and 20% higher track segmentation sensitivity compared to the currently used algorithm. The performance measures are consistent for various datasets confirming the method’s reliability.