In this paper, we propose an ant colony optimization (ACO) aided two-dimensional (2D) soft output Viterbi algorithm (SOVA) detection scheme for holographic data storage systems. The proposed detection scheme adopts the conventional 2D SOVA detection scheme which has two one-dimensional (1D) SOVA detectors for the horizontal and vertical directions, and employs an ACO detector with a reliability test. The proposed detection scheme operates in two stages. In the first stage, the data is detected by conducting the 2D SOVA detection scheme for the horizontal and vertical directions. In the second stage, the reliability test is taken to detect the less reliable bits in a data page by using the outputs of the horizontal and vertical 1D SOVA detectors from the first stage. If the less reliable bits are detected, the ACO detector is performed to improve the reliability of the unreliable bits. Otherwise, the sum of outputs from the horizontal and vertical 1D SOVA detectors is detected by a threshold detector. The bit error rate performances of the proposed detection scheme are compared to the conventional detection schemes. Simulation results show that the proposed detection scheme performing the ACO detector for only 0.3% of bits has more than 7 dB gains compared to the conventional 2D SOVA detection scheme when the grade of blur is 1.8.