Content based medical image retrieval using local co-occurrence patterns

To index and retrieve the images from a medical image dataset, an algorithm is proposed in this paper which is framed using local Co-occurrence Patterns (LCoP). In the existing LDP is rotation variant and captures the information only in a particular direction. The proposed LCoP codes the occurrence among the 1-order differentiators in all possible directions and is a rotational invariant method. The impact of the algorithm is further made clear by integrating the same with Gabor transform. All results of experiments on OASIS-MRI shown a considerable improvement in terms of the performance metrics when compared to LBP, LTP, LDP even after taking the Gabor function.