Introduction: Microwave thermal abla4on (MTA) is an effec4ve and minimally invasive technique for cancer treatment. A percutaneous antenna is inserted into the tumor and fed with microwave frequency power for a few minutes, heating the surrounding tissue through electromagnetic energy deposition. Temperature monitoring is essential to ensure complete tumor coverage while avoiding damage to healthy tissue. Methods: Since tissue electrical properties change with temperature, Microwave Imaging (MWI) has recently been proposed as a non-invasive, real-time monitoring approach: antennas on the body surface probe the treated region and the backscattered signals are processed to retrieve the electrical properties as a proxy for temperature distribution. However, MWI is limited by scarce available data and signal attenuation. We propose a treatment-specificm approach exploiting a priori knowledge of the scenario and its expected evolution. Studies have shown that MTA typically produces elliptically shaped variations of the heated tissue; accordingly, we build a custom basis consis4ng of N=4 concentric elliptical annuli and we use it to process simulated data in the 0.75–1.7 GHz frequency range through a truncated singular value decomposi4on (TSVD) scheme. Results: Comparing the results obtained with a standard TSVD scheme and the proposed treatment-specific one, it is evident that the latter allows better discrimination of the transi4on from the initial to the intermediate and to the final ablation stage. The improved results are achievable thanks to the sub-centimetre spatial discrimination of about 8 mm enabled by elliptical annuli basis, whereas the standard TSVD-based approach only allows a smooth discrimination of the treated tissue boundaries. Conclusions: Preliminary numerical validation on a 2D anthropomorphic phantom demonstrates the feasibility of this custom-basis approach, paving the way for further assessment in realistic 3D scenarios.