Streaming convolutional neural networks for end-to-end learning with multi-megapixel images
This addresses a critical bottleneck in medical imaging and other domains requiring high-resolution images, enabling more accurate disease detection, though it is a novel method for a known bottleneck rather than a paradigm shift.
The paper tackles the problem of training convolutional neural networks on multi-megapixel images, which is limited by memory constraints, by proposing a streaming method that processes smaller tiles, enabling end-to-end learning on images up to 66 megapixels and saving about 50GB of memory per image. It demonstrates improved performance, such as increasing the AUC from 0.580 to 0.706 for breast cancer metastasis detection.
Due to memory constraints on current hardware, most convolution neural networks (CNN) are trained on sub-megapixel images. For example, most popular datasets in computer vision contain images much less than a megapixel in size (0.09MP for ImageNet and 0.001MP for CIFAR-10). In some domains such as medical imaging, multi-megapixel images are needed to identify the presence of disease accurately. We propose a novel method to directly train convolutional neural networks using any input image size end-to-end. This method exploits the locality of most operations in modern convolutional neural networks by performing the forward and backward pass on smaller tiles of the image. In this work, we show a proof of concept using images of up to 66-megapixels (8192x8192), saving approximately 50GB of memory per image. Using two public challenge datasets, we demonstrate that CNNs can learn to extract relevant information from these large images and benefit from increasing resolution. We improved the area under the receiver-operating characteristic curve from 0.580 (4MP) to 0.706 (66MP) for metastasis detection in breast cancer (CAMELYON17). We also obtained a Spearman correlation metric approaching state-of-the-art performance on the TUPAC16 dataset, from 0.485 (1MP) to 0.570 (16MP). Code to reproduce a subset of the experiments is available at https://github.com/DIAGNijmegen/StreamingCNN.