During medical investigations of the head, the presence of skull bone constitutes a major challenge in generating accurate diagnostics. Photoacoustic imaging technology, with its functional imaging capabilities, has demonstrated the potential for brain imaging at low cost and with low maintenance requirements. While photoacoustic signal generation in deep tissue and through the skull has been demonstrated, an effective method of aberration correction for transcranial photoacoustic imaging has not yet been developed. In this study, we present a method based on enfolded deep learning algorithms that accurately compensates for acoustic aberrations caused by the head layers, allowing hemorrhage detection. Using a realistic simulated framework, a large quantity of aberrated images is acquired, reconstructed, and corrected.