Abstract
Convolutional neural network (CNN) models obtain state of the art performance on image classification, localization, and segmentation tasks. Limitations in computer hardware, most notably memory size in deep learning accelerator cards, prevent relatively large images, such as those from medical and satellite imaging, from being processed as a whole in their original resolution. A fully convolutional topology, such as U-Net, is typically trained on down-sampled images and inferred on images of their original size and resolution, by simply dividing the larger image into smaller (typically overlapping) tiles, making predictions on these tiles, and stitching them back together as the prediction for the whole image. In this study, we show that this tiling technique combined with translationally-invariant nature of CNNs causes small, but relevant differences during inference that can be detrimental in the performance of the model. Here we quantify these variations in both medical (i.e., BraTS) and non-medical (i.e., satellite) images and show that training a 2D U-Net model on the whole image substantially improves the overall model performance. Finally, we compare 2D and 3D semantic segmentation models to show that providing CNN models with a wider context of the image in all three dimensions leads to more accurate and consistent predictions. Our results suggest that tiling the input to CNN models—while perhaps necessary to overcome the memory limitations in computer hardware—may lead to undesirable and unpredictable errors in the model's output that can only be adequately mitigated by increasing the input of the model to the largest possible tile size.
Highlights
Since their resurgence in 2012 convolutional neural networks (CNN) have rapidly proved to be the state-of-the-art method for computer-aided diagnosis in medical imaging, and have led to improved accuracy in classification, localization, and segmentation tasks (Krizhevsky et al, 2012; Chen et al, 2016; Greenspan et al, 2016)
Convolutional networks are a natural fit for tiling methods, as they can be trained on images of one size and perform inference on images of a larger size by breaking the large image into smaller, overlapping tiles (Ronneberger et al, 2015; Çiçek et al, 2016; Roth et al, 2018)
Brain Tumor Segmentation (BraTS) The medical data used for our evaluations reflect the publiclyavailable training dataset of the International Brain Tumor Segmentation (BraTS) challenge 20191 (Figure 1) (Menze et al, 2014; Bakas et al, 2017a,b,c; Bakas et al, 2018)
Summary
Since their resurgence in 2012 convolutional neural networks (CNN) have rapidly proved to be the state-of-the-art method for computer-aided diagnosis in medical imaging, and have led to improved accuracy in classification, localization, and segmentation tasks (Krizhevsky et al, 2012; Chen et al, 2016; Greenspan et al, 2016). Tiling is often applied when using large images due to the memory limitations of the hardware (Roth et al, 2018). In CNN models, the activation maps of the intermediate layers use several times the memory footprint of the original input image. These activation maps can increase the allocated memory to hundreds of gigabytes. Convolutional networks are a natural fit for tiling methods, as they can be trained on images of one size and perform inference on images of a larger size by breaking the large image into smaller, overlapping tiles (Ronneberger et al, 2015; Çiçek et al, 2016; Roth et al, 2018). To perform the overlapping tiling at inference time, varying N × N (or in the 3D case, N × N × N) tiles are cropped from the whole image at uniformly spaced offsets along the image dimensions
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.