Lock-In Thermography is an established non-destructively operating method for the analysis of failures in microelectronic devices. In recent years a major improvement was achieved allowing the acquisition of the time-resolved temperature responses of weak thermal spots that enhances defect localization in 3D stacked semiconductor architectures. The assessment of a defect's depth based on the numerical estimation of the delay between excitation and thermal response by analyzing the value of the lock-in phase is often prone to thermal noise and parasitic effects. In sample structures that contain partial or full transparence for the infrared signal between the origin and the sample surface, the interference of the direct (radiated) and the conducted signal component largely falsifies the phase value on which the classical depth estimation relies. In the present study blind source separation based on independent component analysis of the thermal signals was successfully applied to separate interfering signal components arising from direct thermal radiation and conduction for a precise estimation of the defect depth.