In this paper, we present UNet++, a new, more powerful architecture for medical image segmentation. Our architecture is essentially a deeply-supervised encoder-decoder network where the encoder and decoder sub-networks are connected through a series of nested, dense skip pathways. The re-designed skip pathways aim at reducing the semantic gap between the feature maps of the encoder and decoder sub-networks. We argue that the optimizer would deal with an easier learning task when the feature maps from the decoder and encoder networks are semantically similar. We have evaluated UNet++ in comparison with U-Net and wide U-Net architectures across multiple medical image segmentation tasks: nodule segmentation in the low-dose CT scans of chest, nuclei segmentation in the microscopy images, liver segmentation in abdominal CT scans, and polyp segmentation in colonoscopy videos. Our experiments demonstrate that UNet++ with deep supervision achieves an average IoU gain of 3.9 and 3.4 points over U-Net and wide U-Net, respectively.

UNet++: A Nested U-Net Architecture for Medical Image Segmentation

Unet++: A nested u-net architecture for medical image segmentation

Artificial intelligence (AI) algorithms, particularly deep learning, have demonstrated remarkable progress in image-recognition tasks. Methods ranging from convolutional neural networks to variational autoencoders have found myriad applications in the medical image analysis field, propelling it forward at a rapid pace. Historically, in radiology practice, trained physicians visually assessed medical images for the detection, characterization and monitoring of diseases. AI methods excel at automatically recognizing complex patterns in imaging data and providing quantitative, rather than qualitative, assessments of radiographic characteristics. In this Opinion article, we establish a general understanding of AI methods, particularly those pertaining to image-based tasks. We explore how these methods could impact multiple facets of radiology, with a general focus on applications in oncology, and demonstrate ways in which these methods are advancing the field. Finally, we discuss the challenges facing clinical implementation and provide our perspective on how the domain could be advanced.

/pdf/artificial-intelligence-in-radiology-2c2bdk1z41.pdf

Artificial intelligence in radiology

The state-of-the-art models for medical image segmentation are variants of U-Net and fully convolutional networks (FCN). Despite their success, these models have two limitations: (1) their optimal depth is apriori unknown, requiring extensive architecture search or inefficient ensemble of models of varying depths; and (2) their skip connections impose an unnecessarily restrictive fusion scheme, forcing aggregation only at the same-scale feature maps of the encoder and decoder sub-networks. To overcome these two limitations, we propose UNet++, a new neural architecture for semantic and instance segmentation, by (1) alleviating the unknown network depth with an efficient ensemble of U-Nets of varying depths, which partially share an encoder and co-learn simultaneously using deep supervision; (2) redesigning skip connections to aggregate features of varying semantic scales at the decoder sub-networks, leading to a highly flexible feature fusion scheme; and (3) devising a pruning scheme to accelerate the inference speed of UNet++. We have evaluated UNet++ using six different medical image segmentation datasets, covering multiple imaging modalities such as computed tomography (CT), magnetic resonance imaging (MRI), and electron microscopy (EM), and demonstrating that (1) UNet++ consistently outperforms the baseline models for the task of semantic segmentation across different datasets and backbone architectures; (2) UNet++ enhances segmentation quality of varying-size objects—an improvement over the fixed-depth U-Net; (3) Mask RCNN++ (Mask R-CNN with UNet++ design) outperforms the original Mask R-CNN for the task of instance segmentation; and (4) pruned UNet++ models achieve significant speedup while showing only modest performance degradation. Our implementation and pre-trained models are available at https://github.com/MrGiovanni/UNetPlusPlus .

UNet++: Redesigning Skip Connections to Exploit Multiscale Features in Image Segmentation

High-resolution representations are essential for position-sensitive vision problems, such as human pose estimation, semantic segmentation, and object detection. Existing state-of-the-art frameworks first encode the input image as a low-resolution representation through a subnetwork that is formed by connecting high-to-low resolution convolutions \emph{in series} (e.g., ResNet, VGGNet), and then recover the high-resolution representation from the encoded low-resolution representation. Instead, our proposed network, named as High-Resolution Network (HRNet), maintains high-resolution representations through the whole process. There are two key characteristics: (i) Connect the high-to-low resolution convolution streams \emph{in parallel}; (ii) Repeatedly exchange the information across resolutions. The benefit is that the resulting representation is semantically richer and spatially more precise. We show the superiority of the proposed HRNet in a wide range of applications, including human pose estimation, semantic segmentation, and object detection, suggesting that the HRNet is a stronger backbone for computer vision problems. All the codes are available at~{\url{this https URL}}.

Deep High-Resolution Representation Learning for Visual Recognition

Generative adversarial networks (GANs) have ushered in a revolution in image-to-image translation. The development and proliferation of GANs raises an interesting question: can we train a GAN to remove an object, if present, from an image while otherwise preserving the image? Specifically, can a GAN "virtually heal" anyone by turning his medical image, with an unknown health status (diseased or healthy), into a healthy one, so that diseased regions could be revealed by subtracting those two images? Such a task requires a GAN to identify a minimal subset of target pixels for domain translation, an ability that we call fixed-point translation, which no GAN is equipped with yet. Therefore, we propose a new GAN, called Fixed-Point GAN, trained by (1) supervising same-domain translation through a conditional identity loss, and (2) regularizing cross-domain translation through revised adversarial, domain classification, and cycle consistency loss. Based on fixed-point translation, we further derive a novel framework for disease detection and localization using only image-level annotation. Qualitative and quantitative evaluations demonstrate that the proposed method outperforms the state of the art in multi-domain image-to-image translation and that it surpasses predominant weakly-supervised localization methods in both disease detection and localization. Implementation is available at this https URL.

Learning Fixed Points in Generative Adversarial Networks: From Image-to-Image Translation to Disease Detection and Localization

This paper introduces a new concept called "transferable visual words" (TransVW), aiming to achieve annotation efficiency for deep learning in medical image analysis. Medical imaging--focusing on particular parts of the body for defined clinical purposes--generates images of great similarity in anatomy across patients and yields sophisticated anatomical patterns across images, which are associated with rich semantics about human anatomy and which are natural visual words. We show that these visual words can be automatically harvested according to anatomical consistency via self-discovery, and that the self-discovered visual words can serve as strong yet free supervision signals for deep models to learn semantics-enriched generic image representation via self-supervision (self-classification and self-restoration). Our extensive experiments demonstrate the annotation efficiency of TransVW by offering higher performance and faster convergence with reduced annotation cost in several applications. Our TransVW has several important advantages, including (1) TransVW is a fully autodidactic scheme, which exploits the semantics of visual words for self-supervised learning, requiring no expert annotation; (2) visual word learning is an add-on strategy, which complements existing self-supervised methods, boosting their performance; and (3) the learned image representation is semantics-enriched models, which have proven to be more robust and generalizable, saving annotation efforts for a variety of applications through transfer learning. Our code, pre-trained models, and curated visual words are available at this https URL.

Transferable Visual Words: Exploiting the Semantics of Anatomical Patterns for Self-supervised Learning

Methods, systems, and media for discriminating and generating translated images are provided. In some embodiments, the method comprises: identifying a set of training images, wherein each image is associated with at least one domain from a plurality of domains; training a generator network to generate: i) a first fake image that is associated with a first domain; and ii) a second fake image that is associated with a second domain; training a discriminator network, using as inputs to the discriminator network: i) an image from the set of training images; ii) the first fake image; and iii) the second fake image; and using the generator network to generate, for an image not included in the set of training images at least one of: i) a third fake image that is associated with the first domain; and ii) a fourth fake image that is associated with the second domain.

Methods, systems, and media for discriminating and generating translated images

Methods, systems, and media for segmenting images are provided. In some embodiments, the method comprises: generating an aggregate U-Net comprised of a plurality of U-Nets, wherein each U-Net in the plurality of U-Nets has a different depth, wherein each U-Net is comprised of a plurality of nodes Xi,j, wherein i indicates a down-sampling layer the U-Net, and wherein j indicates a convolution layer of the U-Net; training the aggregate U-Net by: for each training sample in a group of training samples, calculating, for each node in the plurality of nodes Xi,j, a feature map xi,j, wherein xi,j is based on a convolution operation performed on a down-sampling of an output from Xi−1,j when j=0, and wherein xi,j is based on a convolution operation performed on an up-sampling operation of an output from Xi+1,j−1 when j>0; and predicting a segmentation of a test image using the trained aggregate U-Net.

Methods, systems, and media for segmenting images

The success of deep learning relies heavily on large datasets with extensive labels, but we often only have access to several small, heterogeneous datasets associated with partial labels, particularly in the field of medical imaging. When learning from multiple datasets, existing challenges include incomparable, heterogeneous, or even conflicting labeling protocols across datasets. In this paper, we propose a new initiative--"data, assemble"--which aims to unleash the full potential of partially labeled data and enormous unlabeled data from an assembly of datasets. To accommodate the supervised learning paradigm to partial labels, we introduce a dynamic adapter that encodes multiple visual tasks and aggregates image features in a question-and-answer manner. Furthermore, we employ pseudo-labeling and consistency constraints to harness images with missing labels and to mitigate the domain gap across datasets. From proof-of-concept studies on three natural imaging datasets and rigorous evaluations on two large-scale thorax X-ray benchmarks, we discover that learning from "negative examples" facilitates both classification and segmentation of classes of interest. This sheds new light on the computer-aided diagnosis of rare diseases and emerging pandemics, wherein "positive examples" are hard to collect, yet "negative examples" are relatively easier to assemble. As a result, besides exceeding the prior art in the NIH ChestXray benchmark, our model is particularly strong in identifying diseases of minority classes, yielding over 3-point improvement on average. Remarkably, when using existing partial labels, our model performance is on-par (p>0.05) with that using a fully curated dataset with exhaustive labels, eliminating the need for additional 40% annotation costs.

Zongwei Zhou

Papers

Learning Fixed Points in Generative Adversarial Networks: From Image-to-Image Translation to Disease Detection and Localization

Transferable Visual Words: Exploiting the Semantics of Anatomical Patterns for Self-supervised Learning

Methods, systems, and media for discriminating and generating translated images

Methods, systems, and media for segmenting images

Data, Assemble: Leveraging Multiple Datasets with Heterogeneous and Partial Labels.