Convolutional Neural Networks for Subgure Classication

A major challenge for Medical Image Retrieval (MIR) is the discovery of relationships between low-level image features (inten- sity, gradient, texture, etc.) and high-level semantics such as modal- ity, anatomy or pathology. Convolutional Neural Networks (CNNs) have been shown to have an inherent ability to automatically extract hier- archical representations from raw data. Their successful application in a variety of generalised imaging tasks suggests great potential for MIR. However, a major hurdle to their deployment in the medical domain is the relative lack of robust training corpora when compared to general imaging benchmarks such as ImageNET and CIFAR. In this paper, we present the adaptation of CNNs to the subgure classication subtask of the medical classication task at ImageCLEF 2015.