Optical aberrations significantly impair microscopic image quality across various domains, including cell biology and histopathology diagnostics. Traditional adaptive optics techniques, such as wavefront shaping and guide star utilization, face challenges, especially in imaging biological tissues. Here, we introduce a computational adaptive optics approach tailored for optically thick samples. Utilizing the tilt-tilt correlation from the optical memory effect, our method detects phase differences in aberrations caused by small tilts in the incident waves. Experimental validation demonstrates our technique's capacity to enhance imaging of thick human tissues under substantial aberration conditions using a transmission-mode holotomography setup. Remarkably, our approach works robustly against sample movement, which is essential for enhanced imaging accuracy in critical biomedical applications.