Electron Paramagnetic Resonance Imaging (EPRI) and spectroscopy (EPR) are noninvasive techniques used to detect and quantify paramagnetic species. EPRI is particularly promising for in-vivo oximetry, which is crucial for managing tumors and cardiac diseases. Since direct observation of oxygen is not feasible in preclinical imagers, oxygen levels are instead detected indirectly through interactions with EPR probes and oxygen concentration maps are recovered using a source separation framework. This work introduces a Total Variation-based variational method for EPRI source separation, specifically when multiple paramagnetic species are present in the acquisition device. This method significantly reduces acquisition time and is validated through complex in vitro experiments. By considering EPR probes under different oxygen conditions as distinct paramagnetic species, the source separation framework can be adapted to reconstruct oxygen concentration maps without relying on extensive Spectral-Spatial Imaging (SSI) acquisitions. This approach is validated through synthetic experiments and is supported by theoretical guarantees for accurate oxygen recovery.