Concerns over fuel scarcity and environmental degradation largely drive the increasing popularity of electric vehicles (EVs). Lithium-ion batteries (LIBs), known for their high energy and power densities, are the favored power source for EVs. Over the past few decades, research has been concentrated on ensuring these batteries operate efficiently, safely, and reliably. A key issue impacting the safety of Li-ion batteries is thermal runaway (TR), which can lead to hazardous battery fires. Internal short circuits (ISCs) are often the primary cause of these TR incidents, making the early detection of spontaneous ISC formation a pivotal diagnostic task. This research introduces an innovative ISC detection technique for cylindrical Li-ion battery cells. This technique is based on the augmentation of the model state vector in an extended Kalman filter (EKF), combining both classical voltage measurements to surface temperature observations. This framework enables real-time estimation of the internal ISC state while maintaining computational efficiency. The proposed method is tested numerically considering a high-fidelity numerical plant cycled using charge-depleting tests that mimic a practical battery cell working cycle at various C rates and at different ambient temperatures to account for both load and environmental uncertainties. The results demonstrate the robustness and effectiveness of the method. In addition, the method has been proven to be computationally efficient, demonstrating the feasibility of its real-time implementation.