Medical Image Classification with Weighted Latent Semantic Tensors and Deep Convolutional Neural Networks

This paper proposes a novel approach for identifying the modality of medical images combining Latent Semantic Analysis (LSA) with Convolutional Neural Networks (CNN). In particular, we aim in investigating the potential of Neural Networks when images are represented by compact descriptors. To this end, an optimized latent semantic space is constructed that captures the affinity of images to each modality using a pre-trained network. The images are represented by a Weighted Latent Semantic Tensor in a lower space and they are used to train a deep CNN that makes the final classification. The evaluation of the proposed algorithm was based on the datasets from the ImageCLEF Medical Subfigure classification contest. Experimental results demonstrate the effectiveness and the efficiency of our framework in terms of classification accuracy, achieving comparable results to current state-of-the-art approaches on the aforementioned datasets.