Deformable Gabor Feature Networks for Biomedical Image Classification
This paper tackles the problem of improving the representation of complex geometric structures in medical images for medical image analysis, which is an incremental improvement for researchers in medical imaging.
The authors address the problem of insufficient representation of complex geometric structures in medical images by current deep learning models. They introduce a deformable Gabor convolution (DGConv) to improve interpretability and handle complex spatial variations, applying it to multi-instance multi-label classification on mammograms and pulmonary x-ray images.
In recent years, deep learning has dominated progress in the field of medical image analysis. We find however, that the ability of current deep learning approaches to represent the complex geometric structures of many medical images is insufficient. One limitation is that deep learning models require a tremendous amount of data, and it is very difficult to obtain a sufficient amount with the necessary detail. A second limitation is that there are underlying features of these medical images that are well established, but the black-box nature of existing convolutional neural networks (CNNs) do not allow us to exploit them. In this paper, we revisit Gabor filters and introduce a deformable Gabor convolution (DGConv) to expand deep networks interpretability and enable complex spatial variations. The features are learned at deformable sampling locations with adaptive Gabor convolutions to improve representativeness and robustness to complex objects. The DGConv replaces standard convolutional layers and is easily trained end-to-end, resulting in deformable Gabor feature network (DGFN) with few additional parameters and minimal additional training cost. We introduce DGFN for addressing deep multi-instance multi-label classification on the INbreast dataset for mammograms and on the ChestX-ray14 dataset for pulmonary x-ray images.