Search

Subarachnoid hemorrhage (SAH) caused by the rupture of cerebral aneurysm is a serious life-threatening disease. Therefore, estimating the risk of cerebral aneurysm rupture is clinically important. In this paper, a semi-automatic method for estimating the risk of rupture of cerebral aneurysm was proposed. We applied a variety of methods to extract features of cerebral aneurysm images and 3D modeling, and used XGBoost and fully connected neural network for classification and analysis respectively. The method achieved an F2-score of 0.862 on the test set of CADA 2020.

Domain shift happens in cross-domain scenarios commonly because of the wide gaps between different domains: when applying a deep learning model well-trained in one domain to another target domain, the model usually performs poorly. To tackle this problem, unsupervised domain adaptation (UDA) techniques are proposed to bridge the gap between different domains, for the purpose of improving model performance without annotation in the target domain. Particularly, UDA has a great value for multimodal medical image analysis, where annotation difficulty is a practical concern. However, most existing UDA methods can only achieve satisfactory improvements in one adaptation direction (e.g., MRI to CT), but often perform poorly in the other (CT to MRI), limiting their practical usage. In this paper, we propose a bidirectional UDA (BiUDA) framework based on disentangled representation learning for equally competent two-way UDA performances. This framework employs a unified domain-aware pattern encoder which not only can adaptively encode images in different domains through a domain controller, but also improve model efficiency by eliminating redundant parameters. Furthermore, to avoid distortion of contents and patterns of input images during the adaptation process, a content-pattern consistency loss is introduced. Additionally, for better UDA segmentation performance, a label consistency strategy is proposed to provide extra supervision by recomposing target-domain-styled images and corresponding source-domain annotations. Comparison experiments and ablation studies conducted on two public datasets demonstrate the superiority of our BiUDA framework to current state-of-the-art UDA methods and the effectiveness of its novel designs. By successfully addressing two-way adaptations, our BiUDA framework offers a flexible solution of UDA techniques to the real-world scenario.

Automated surface segmentation of retinal layer is important and challenging in analyzing optical coherence tomography (OCT). Recently, many deep learning based methods have been developed for this task and yield remarkable performance. However, due to large spatial gap and potential mismatch between the B-scans of OCT data, all of them are based on 2D segmentation of individual B-scans, which may loss the continuity information across the B-scans. In addition, 3D surface of the retina layers can provide more diagnostic information, which is crucial in quantitative image analysis. In this study, a novel framework based on hybrid 2D-3D convolutional neural networks (CNNs) is proposed to obtain continuous 3D retinal layer surfaces from OCT. The 2D features of individual B-scans are extracted by an encoder consisting of 2D convolutions. These 2D features are then used to produce the alignment displacement field and layer segmentation by two 3D decoders, which are coupled via a spatial transformer module. The entire framework is trained end-to-end. To the best of our knowledge, this is the first study that attempts 3D retinal layer segmentation in volumetric OCT images based on CNNs. Experiments on a publicly available dataset show that our framework achieves superior results to state-of-the-art 2D methods in terms of both layer segmentation accuracy and cross-B-scan 3D continuity, thus offering more clinical values than previous works.

With the advancement and prevailing success of Transformer models in the natural language processing (NLP) field, an increasing number of research works have explored the applicability of Transformer for various vision tasks and reported superior performance compared with convolutional neural networks (CNNs). However, as the proper training of Transformer generally requires an extremely large quantity of data, it has rarely been explored for the medical imaging tasks. In this paper, we attempt to adopt the Vision Transformer for the retinal disease classification tasks, by pre-training the Transformer model on a large fundus image database and then fine-tuning on downstream retinal disease classification tasks. In addition, to fully exploit the feature representations extracted by individual image patches, we propose a multiple instance learning (MIL) based ‘MIL head’, which can be conveniently attached to the Vision Transformer in a plug-and-play manner and effectively enhances the model performance for the downstream fundus image classification tasks. The proposed MIL-VT framework achieves superior performance over CNN models on two publicly available datasets when being trained and tested under the same setup. The implementation code and pre-trained weights are released for public access (Code link: https://github.com/greentreeys/MIL-VT).

Cervical cancer causes the fourth most cancer-related deaths of women worldwide. One of the most commonly-used clinical tools for the diagnosis of cervical intraepithelial neoplasia (CIN) and cervical cancer is colposcopy examination. However, due to the challenging imaging conditions such as light reflection on the cervix surface, the clinical accuracy of colposcopy examination is relatively low. In this paper, we propose a computer-aided diagnosis (CAD) system to accurately segment the lesion areas (i.e., CIN and cancer) from colposcopic images, which can not only assist colposcopists for clinical decision, but also provide the guideline for the location of biopsy sites. In clinical practice, colposcopists often need to zoom in the potential lesion area for clearer observation. The colposcopic images with multi-scale views result in a difficulty for current straight-forward deep learning networks to process. To address the problem, we propose a novel attention mechanism, namely feature library, which treats the whole backbone network as a pool of features and extract the useful features on different scales from the pool to recalibrate the most informative representation. Furthermore, to well-train and evaluate our deep learning network, we collect a large-scale colposcopic image dataset for CervIcal lesioN sEgMentAtion (CINEMA), consisting of 34,337 images from 9,652 patients. The lesion areas in the colposcopic images are manually annotated by experienced colposcopists. Extensive experiments are conducted on the CINEMA dataset, which demonstrate the effectiveness of our feature library dealing with cervical lesions of varying sizes.

Learning by imitation is one of the most significant abilities of human beings and plays a vital role in human’s computational neural system. In medical image analysis, given several exemplars (anchors), experienced radiologist has the ability to delineate unfamiliar organs by imitating the reasoning process learned from existing types of organs. Inspired by this observation, we propose OrganNet which learns a generalized organ concept from a set of annotated organ classes and then transfer this concept to unseen classes. In this paper, we show that such process can be integrated into the one-shot segmentation task which is a very challenging but meaningful topic. We propose pyramid reasoning modules (PRMs) to model the anatomical correlation between anchor and target volumes. In practice, the proposed module first computes a correlation matrix between target and anchor computerized tomography (CT) volumes. Then, this matrix is used to transform the feature representations of both anchor volume and its segmentation mask. Finally, OrganNet learns to fuse the representations from various inputs and predicts segmentation results for target volume. Extensive experiments show that OrganNet can effectively resist the wide variations in organ morphology and produce state-of-the-art results in one-shot segmentation task. Moreover, even when compared with fully-supervised segmentation models, OrganNet is still able to produce satisfying segmentation results.

With the rapidly growing number of people affected by various retinal diseases, there is a strong clinical interest for fully automatic and accurate retinal disease recognition. The unique characteristics of how retinal diseases are manifested on the fundus images pose a major challenge for automatic recognition. In order to tackle the challenges, we propose a local-global dual perception (LGDP) based deep multiple instance learning (MIL) framework that integrates the instance contribution from both local scale and global scale. The major components of the proposed framework include a local pyramid perception module (LPPM) that emphasizes the key instances from the local scale, and a global perception module (GPM) that provides a spatial weight distribution from a global scale. Extensive experiments on three major retinal disease benchmarks demonstrate that the proposed framework outperforms many state-of-the-art deep MIL methods, especially for recognizing the pathological images. Last but not least, the proposed deep MIL framework can be conveniently embedded into any convolutional backbones via a plug-and-play manner and effectively boost the performance.

Rare diseases are characterized by low prevalence and are often chronically debilitating or life-threatening. Imaging-based classification of rare diseases is challenging due to the severe shortage in training examples. Few-shot learning (FSL) methods tackle this challenge by extracting generalizable prior knowledge from a large base dataset of common diseases and normal controls, and transferring the knowledge to rare diseases. Yet, most existing methods require the base dataset to be labeled and do not make full use of the precious examples of the rare diseases. To this end, we propose in this work a novel hybrid approach to rare disease classification, featuring two key novelties targeted at the above drawbacks. First, we adopt the unsupervised representation learning (URL) based on self-supervising contrastive loss, whereby to eliminate the overhead in labeling the base dataset. Second, we integrate the URL with pseudo-label supervised classification for effective self-distillation of the knowledge about the rare diseases, composing a hybrid approach taking advantages of both unsupervised and (pseudo-) supervised learning on the base dataset. Experimental results on classification of rare skin lesions show that our hybrid approach substantially outperforms existing FSL methods (including those using fully supervised base dataset) for rare disease classification via effective integration of the URL and pseudo-label driven self-distillation, thus establishing a new state of the art.

Most existing short text classification models suffer from poor performance because of the information sparsity of short texts and the polysemous class-bearing words. To alleviate these issues, we propose a context-sensitive topic memory network (cs-TMN) by learning context-sensitive text representations and content-aware extended topic knowledge. Different from TMN that utilizes context-independent word embedding and extended topic knowledge, we further employ context-sensitive word embedding, comprised of local context representation and global context representation to alleviate the polysemous issue. Besides, extended topic knowledge matched by context-sensitive word embedding is proven content-aware in comparison with previous works. Empirical results demonstrate the effectiveness of our cs-TMN, outperforming state-of-the-art models on short text classification on four public datasets.

Automated methods for Cobb angle estimation are of high demand for scoliosis assessment. Existing methods typically calculate the Cobb angle from landmark estimation, or simply combine the low-level task (e.g., landmark detection and spine segmentation) with the Cobb angle regression task, without fully exploring the benefits from each other. In this study, we propose a novel multi-task framework, named Seg4Reg+, which jointly optimizes the segmentation and regression networks. We thoroughly investigate both local and global consistency and knowledge transfer between each other. Specifically, we propose an attention regularization module leveraging class activation maps (CAMs) from image-segmentation pairs to discover additional supervision in the regression network, and the CAMs can serve as a region-of-interest enhancement gate to facilitate the segmentation task in turn. Meanwhile, we design a novel triangle consistency learning to train the two networks jointly for global optimization. The evaluations performed on the public AASCE Challenge dataset demonstrate the effectiveness of each module and superior performance of our model to the state-of-the-art methods.

Deep neural networks for medical images are extremely vulnerable to adversarial examples (AEs), which poses security concerns on clinical decision-making. Recent findings have shown that existing medical AEs are easy to detect in feature space. To better understand this phenomenon, we thoroughly investigate the characteristic of traditional medical AEs in feature space. Specifically, we first perform a stress test to reveal the vulnerability of medical images and compare them to natural images. Then, we theoretically prove that the existing adversarial attacks manipulate the prediction by continuously optimizing the vulnerable representations in a fixed direction, leading to outlier representations in feature space. Interestingly, we find this vulnerability is a double-edged sword that can be exploited to help hide AEs in the feature space. We propose a novel hierarchical feature constraint (HFC) as an add-on to existing white-box attacks, which encourages hiding the adversarial representation in the normal feature distribution. We evaluate the proposed method on two public medical image datasets, namely Fundoscopy and Chest X-Ray. Experimental results demonstrate the superiority of our HFC as it bypasses an array of state-of-the-art adversarial medical AEs detector more efficiently than competing adaptive attacks. Our code is available at https://github.com/qsyao/Hierarchical_Feature_Constraint.

Current deep learning based disease diagnosis systems usually fall short in catastrophic forgetting, i.e., directly fine-tuning the disease diagnosis model on new tasks usually leads to abrupt decay of performance on previous tasks. What is worse, the trained diagnosis system would be fixed once deployed but collecting training data that covers enough diseases is infeasible, which inspires us to develop a lifelong learning diagnosis system. In this work, we propose to adopt attention to combine medical entities and context, embedding episodic memory and consolidation to retain knowledge, such that the learned model is capable of adapting to sequential disease-diagnosis tasks. Moreover, we establish a new benchmark, named Jarvis-40, which contains clinical notes collected from various hospitals. Experiments show that the proposed method can achieve state-of-the-art performance on the proposed benchmark. Code is available at https://github.com/yifyang/LifelongLearningDiseaseDiagnosis.

Manually segmenting the hepatic vessels from Computer Tomography (CT) is far more expertise-demanding and laborious than other structures due to the low-contrast and complex morphology of vessels, resulting in the extreme lack of high-quality labeled data. Without sufficient high-quality annotations, the usual data-driven learning-based approaches struggle with deficient training. On the other hand, directly introducing additional data with low-quality annotations may confuse the network, leading to undesirable performance degradation. To address this issue, we propose a novel mean-teacher-assisted confident learning framework to robustly exploit the noisy labeled data for the challenging hepatic vessel segmentation task. Specifically, with the adapted confident learning assisted by a third party, i.e., the weight-averaged teacher model, the noisy labels in the additional low-quality dataset can be transformed from ‘encumbrance’ to ‘treasure’ via progressive pixel-wise soft-correction, thus providing productive guidance. Extensive experiments using two public datasets demonstrate the superiority of the proposed framework as well as the effectiveness of each component.

View planning for the acquisition of cardiac magnetic resonance imaging (CMR) requires acquaintance with the cardiac anatomy and remains a challenging task in clinical practice. Existing approaches to its automation relied either on an additional volumetric image not typically acquired in clinic routine, or on laborious manual annotations of cardiac structural landmarks. This work presents a clinic-compatible and annotation-free system for automatic CMR view planning. The system mines the spatial relationship—more specifically, locates and exploits the intersecting lines—between the source and target views, and trains deep networks to regress heatmaps defined by these intersecting lines. As the spatial relationship is self-contained in properly stored data, e.g., in the DICOM format, the need for manual annotation is eliminated. Then, a multi-view planning strategy is proposed to aggregate information from the predicted heatmaps for all the source views of a target view, for a globally optimal prescription. The multi-view aggregation mimics the similar strategy practiced by skilled human prescribers. Experimental results on 181 clinical CMR exams show that our system achieves superior accuracy to existing approaches including conventional atlas-based and newer deep learning based ones, in prescribing four standard CMR views. The mean angle difference and point-to-plane distance evaluated against the ground truth planes are 5.98\(^\circ \) and 3.48 mm, respectively.

In recent years, there has been increasing awareness of the occurrence of fatigue fractures. Athletes and soldiers, who engaged in unaccustomed, repetitive or vigorous activities, are potential victims of such a fracture. Due to the slow-growing process of fatigue fracture, the early detection can effectively protect athletes and soldiers from the material bone breakage, which may result in the catastrophe of career retirement. In this paper, we propose a triplet-branch network (TBN) for the accurate fatigue fracture grading, which enables physicians to promptly take appropriate treatments. Particularly, the proposed TBN consists of three branches for representation learning, classifier learning and grade-related prior-knowledge learning, respectively. The former two branches are responsible to tackle the problem of class-imbalanced training data, while the latter one is implemented to embed grade-related prior-knowledge into the framework via an auxiliary ranking task. Extensive experiments have been conducted on our fatigue fracture X-ray image dataset. The experimental results show that our TBN can effectively address the problem of class-imbalanced training samples and achieve a satisfactory accuracy for fatigue fracture grading.

Due to the inter-observer variation, the ground truth of lesion areas in pathological images is generated by majority-voting of annotations provided by different pathologists. Such a process is extremely laborious, since each pathologist needs to spend hours or even days for pixel-wise annotations. In this paper, we propose a reinforcement learning framework to automatically refine the set of annotations provided by a single pathologist based on several exemplars of ground truth. Particularly, we treat each pixel as an agent with a shared pixel-level action space. The multi-agent model observes several paired single-pathologist annotations and ground truth, and tries to customize the strategy to narrow down the gap between them with episodes of exploring. Furthermore, we integrate a discriminator to the multi-agent framework to evaluate the quality of annotation refinement. A quality reward is yielded by the discriminator to update the policy of agents. Experimental results on the publicly available Gleason 2019 dataset demonstrate the effectiveness of our reinforcement learning framework—the segmentation network trained with our refined single-pathologist annotations achieves a comparable accuracy to the one using majority-voting-based ground truth.

For the task of metal artifact reduction (MAR), although deep learning (DL)-based methods have achieved promising performances, most of them suffer from two problems: 1) the CT imaging geometry constraint is not fully embedded into the network during training, leaving room for further performance improvement; 2) the model interpretability is lack of sufficient consideration. Against these issues, we propose a novel interpretable dual domain network, termed as InDuDoNet, which combines the advantages of model-driven and data-driven methodologies. Specifically, we build a joint spatial and Radon domain reconstruction model and utilize the proximal gradient technique to design an iterative algorithm for solving it. The optimization algorithm only consists of simple computational operators, which facilitate us to correspondingly unfold iterative steps into network modules and thus improve the interpretablility of the framework. Extensive experiments on synthesized and clinical data show the superiority of our InDuDoNet. Code is available in https://github.com/hongwang01/InDuDoNet.

Segmentation of objects of interest is one of the central tasks in medical image analysis, which is indispensable for quantitative analysis. When developing machine-learning based methods for automated segmentation, manual annotations are usually used as the ground truth toward which the models learn to mimic. While the bulky parts of the segmentation targets are relatively easy to label, the peripheral areas are often difficult to handle due to ambiguous boundaries and the partial volume effect, etc., and are likely to be labeled with uncertainty. This uncertainty in labeling may, in turn, result in unsatisfactory performance of the trained models. In this paper, we propose superpixel-based label softening to tackle the above issue. Generated by unsupervised over-segmentation, each superpixel is expected to represent a locally homogeneous area. If a superpixel intersects with the annotation boundary, we consider a high probability of uncertain labeling within this area. Driven by this intuition, we soften labels in this area based on signed distances to the annotation boundary and assign probability values within [0, 1] to them, in comparison with the original “hard”, binary labels of either 0 or 1. The softened labels are then used to train the segmentation models together with the hard labels. Experimental results on a brain MRI dataset and an optical coherence tomography dataset demonstrate that this conceptually simple and implementation-wise easy method achieves overall superior segmentation performances to baseline and comparison methods for both 3D and 2D medical images.

In deep learning era, pretrained models play an important role in medical image analysis, in which ImageNet pretraining has been widely adopted as the best way. However, it is undeniable that there exists an obvious domain gap between natural images and medical images. To bridge this gap, we propose a new pretraining method which learns from 700k radiographs given no manual annotations. We call our method as Comparing to Learn (C2L) because it learns robust features by comparing different image representations. To verify the effectiveness of C2L, we conduct comprehensive ablation studies and evaluate it on different tasks and datasets. The experimental results on radiographs show that C2L can outperform ImageNet pretraining and previous state-of-the-art approaches significantly. Code and models are available at https://github.com/funnyzhou/C2L_MICCAI2020.

Medical image analysis benefits Computer Aided Diagnosis (CADx). A fundamental analyzing approach is the classification of medical images, which serves for skin lesion diagnosis, diabetic retinopathy grading, and cancer classification on histological images. When learning these discriminative classifiers, we observe that the convolutional neural networks (CNNs) are vulnerable to distractor interference. This is due to the similar sample appearances from different categories (i.e., small inter-class distance). Existing attempts select distractors from input images by empirically estimating their potential effects to the classifier. The essences of how these distractors affect CNN classification are not known. In this paper, we explore distractors from the CNN feature space via proposing a neuron intrinsic learning method. We formulate a novel distractor-aware loss that encourages large distance between the original image and its distractor in the feature space. The novel loss is combined with the original classification loss to update network parameters by back-propagation. Neuron intrinsic learning first explores distractors crucial to the deep classifier and then uses them to robustify CNN inherently. Extensive experiments on medical image benchmark datasets indicate that the proposed method performs favorably against the state-of-the-art approaches.

Deep learning highly relies on the quantity of annotated data. However, the annotations for 3D volumetric medical data require experienced physicians to spend hours or even days for investigation. Self-supervised learning is a potential solution to get rid of the strong requirement of training data by deeply exploiting raw data information. In this paper, we propose a novel self-supervised learning framework for volumetric medical images. Specifically, we propose a context restoration task, i.e., Rubik’s cube++, to pre-train 3D neural networks. Different from the existing context-restoration-based approaches, we adopt a volume-wise transformation for context permutation, which encourages network to better exploit the inherent 3D anatomical information of organs. Compared to the strategy of training from scratch, fine-tuning from the Rubik’s cube++ pre-trained weight can achieve better performance in various tasks such as pancreas segmentation and brain tissue segmentation. The experimental results show that our self-supervised learning method can significantly improve the accuracy of 3D deep learning networks on volumetric medical datasets without the use of extra data.

In this paper, we propose a new pipeline to perform accurate spinal curvature estimation. The framework, named as Seg4Reg, contains two deep neural networks focusing on segmentation and regression, respectively. Based on the results generated by the segmentation model, the regression network directly predicts the cobb angles from segmentation masks. To alleviate the domain shift problem appeared between training and testing sets, we also conduct a domain adaptation module into network structures. Finally, by ensembling the predictions of different models, our method achieves 21.71 SMAPE in the testing set.

The automatic grading of diabetic retinopathy (DR) facilitates medical diagnosis for both patients and physicians. Existing researches formulate DR grading as an image classification problem. As the stages/categories of DR correlate with each other, the relationship between different classes cannot be explicitly described via a one-hot label because it is empirically estimated by different physicians with different outcomes. This class correlation limits existing networks to achieve effective classification. In this paper, we propose a Graph REsidual rE-ranking Network (GREEN) to introduce a class dependency prior into the original image classification network. The class dependency prior is represented by a graph convolutional network with an adjacency matrix. This prior augments image classification pipeline by re-ranking classification results in a residual aggregation manner. Experiments on the standard benchmarks have shown that GREEN performs favorably against state-of-the-art approaches.

Edge detection is a fundamental problem in different computer vision tasks. Recently, edge detection algorithms achieve satisfying improvement built upon deep learning. Although most of them report favorable evaluation scores, they often fail to accurately localize edges and give thick and blurry boundaries. In addition, most of them focus on 2D images and the challenging 3D edge detection is still under-explored. In this work, we propose a novel logical refinement network for crisp edge detection, which is motivated by the logical relationship between segmentation and edge maps and can be applied to both 2D and 3D images. The network consists of a joint object and edge detection network and a crisp edge refinement network, which predicts more accurate, clearer and thinner high quality binary edge maps without any post-processing. Extensive experiments are conducted on the 2D nuclei images from Kaggle 2018 Data Science Bowl and a private 3D microscopy images of a monkey brain, which show outstanding performance compared with state-of-the-art methods.

Medical images are generally labeled by multiple experts before the final ground-truth labels are determined. Consensus or disagreement among experts regarding individual images reflects the gradeability and difficulty levels of the image. However, when being used for model training, only the final ground-truth label is utilized, while the critical information contained in the raw multi-rater gradings regarding the image being an easy/hard case is discarded. In this paper, we aim to take advantage of the raw multi-rater gradings to improve the deep learning model performance for the glaucoma classification task. Specifically, a multi-branch model structure is proposed to predict the most sensitive, most specifical and a balanced fused result for the input images. In order to encourage the sensitivity branch and specificity branch to generate consistent results for consensus labels and opposite results for disagreement labels, a consensus loss is proposed to constrain the output of the two branches. Meanwhile, the consistency/inconsistency between the prediction results of the two branches implies the image being an easy/hard case, which is further utilized to encourage the balanced fusion branch to concentrate more on the hard cases. Compared with models trained only with the final ground-truth labels, the proposed method using multi-rater consensus information has achieved superior performance, and it is also able to estimate the difficulty levels of individual input images when making the prediction.

Deep neural network-based medical image classifications often use “hard” labels for training, where the probability of the correct category is 1 and those of others are 0. However, these hard targets can drive the networks over-confident about their predictions and prone to overfit the training data, affecting model generalization and adaption. Studies have shown that label smoothing and softening can improve classification performance. Nevertheless, existing approaches are either non-data-driven or limited in applicability. In this paper, we present the Class-Correlation Learning Network (CCL-Net) to learn interclass visual correlations from given training data, and produce soft labels to help with classification tasks. Instead of letting the network directly learn the desired correlations, we propose to learn them implicitly via distance metric learning of class-specific embeddings with a lightweight plugin CCL block. An intuitive loss based on a geometrical explanation of correlation is designed for bolstering learning of the interclass correlations. We further present end-to-end training of the proposed CCL block as a plugin head together with the classification backbone while generating soft labels on the fly. Our experimental results on the International Skin Imaging Collaboration 2018 dataset demonstrate effective learning of the interclass correlations from training data, as well as consistent improvements in performance upon several widely used modern network structures with the CCL block.

Domain shift between medical images from multicentres is still an open question for the community, which degrades the generalization performance of deep learning models. Generative adversarial network (GAN), which synthesize plausible images, is one of the potential solutions to address the problem. However, the existing GAN-based approaches are prone to fail at preserving image-objects in image-to-image (I2I) translation, which reduces their practicality on domain adaptation tasks. In this paper, we propose a novel GAN (namely MI\(^2\)GAN) to maintain image-contents during cross-domain I2I translation. Particularly, we disentangle the content features from domain information for both the source and translated images, and then maximize the mutual information between the disentangled content features to preserve the image-objects. The proposed MI\(^2\)GAN is evaluated on two tasks—polyp segmentation using colonoscopic images and the segmentation of optic disc and cup in fundus images. The experimental results demonstrate that the proposed MI\(^2\)GAN can not only generate elegant translated images, but also significantly improve the generalization performance of widely used deep learning networks (e.g., U-Net).

Recently, deep learning has been adopted to the glaucoma classification task with performance comparable to that of human experts. However, a well trained deep learning model demands a large quantity of properly labeled data, which is relatively expensive since the accurate labeling of glaucoma requires years of specialist training. In order to alleviate this problem, we propose a glaucoma classification framework which takes advantage of not only the properly labeled images, but also undiagnosed images without glaucoma labels. To be more specific, the proposed framework is adapted from the teacher-student-learning paradigm. The teacher model encodes the wrapped information of undiagnosed images to a latent feature space, meanwhile the student model learns from the teacher through knowledge transfer to improve the glaucoma classification. For the model training procedure, we propose a novel training strategy that simulates the real-world teaching practice named as “Learning To Teach with Knowledge Transfer (L2T-KT)", and establish a“Quiz Pool" as the teacher’s optimization target. Experiments show that the proposed framework is able to utilize the undiagnosed data effectively to improve the glaucoma prediction performance.

Witnessing the success of deep learning neural networks in natural image processing, an increasing number of studies have been proposed to develop deep-learning-based frameworks for medical image segmentation. However, since the pixel-wise annotation of medical images is laborious and expensive, the amount of annotated data is usually deficient to well-train a neural network. In this paper, we propose a semi-supervised approach to train neural networks with limited labeled data and a large quantity of unlabeled images for medical image segmentation. A novel pseudo-label (namely self-loop uncertainty), generated by recurrently optimizing the neural network with a self-supervised task, is adopted as the ground-truth for the unlabeled images to augment the training set and boost the segmentation accuracy. The proposed self-loop uncertainty can be seen as an approximation of the uncertainty estimation yielded by ensembling multiple models with a significant reduction of inference time. Experimental results on two publicly available datasets demonstrate the effectiveness of our semi-supervised approach.

Active learning, reducing the cost and workload of annotations, attracts increasing attentions from the community. Current active learning approaches commonly adopted uncertainty-based acquisition functions for the data selection due to their effectiveness. However, data selection based on uncertainty suffers from the overlapping problem, i.e., the top-K samples ranked by the uncertainty are similar. In this paper, we investigate the overlapping problem of recent uncertainty-based approaches and propose to alleviate the issue by taking representativeness into consideration. In particular, we propose a dual adversarial network, namely DAAL, for this purpose. Different from previous hybrid active learning methods requiring multi-stage data selections i.e., step-by-step evaluating the uncertainty and representativeness using different acquisition functions, our DAAL learns to select the most uncertain and representative data points in one-stage. Extensive experiments conducted on three publicly available datasets, i.e., CIFAR10/100 and Cityscapes, demonstrate the effectiveness of our method—a new state-of-the-art accuracy is achieved.

Primary angle closure glaucoma (PACG) is the leading cause of irreversible blindness among Asian people. Early detection of PACG is essential, so as to provide timely treatment and minimize the vision loss. In the clinical practice, PACG is diagnosed by analyzing the angle between the cornea and iris with anterior segment optical coherence tomography (AS-OCT). The rapid development of deep learning technologies provides the feasibility of building a computer-aided system for the fast and accurate segmentation of cornea and iris tissues. However, the application of deep learning methods in the medical imaging field is still restricted by the lack of enough fully-annotated samples. In this paper, we propose a novel framework to segment the target tissues accurately for the AS-OCT images, by using the combination of weakly-annotated images (majority) and fully-annotated images (minority). The proposed framework consists of two models which provide reliable guidance for each other. In addition, uncertainty guided strategies are adopted to increase the accuracy and stability of the guidance. Detailed experiments on the publicly available AGE dataset demonstrate that the proposed framework outperforms the state-of-the-art semi-/weakly-supervised methods and has a comparable performance as the fully-supervised method. Therefore, the proposed method is demonstrated to be effective in exploiting information contained in the weakly-annotated images and has the capability to substantively relieve the annotation workload.

Recent generative adversarial network (GAN) based methods (e.g., CycleGAN) are prone to fail at preserving image-objects in image-to-image translation, which reduces their practicality on tasks such as domain adaptation. Some frameworks have been proposed to adopt a segmentation network as the auxiliary regularization to prevent the content distortion. However, all of them require extra pixel-wise annotations, which is difficult to fulfill in practical applications. In this paper, we propose a novel GAN (namely OP-GAN) to address the problem, which involves a self-supervised module to enforce the image content consistency during image-to-image translations without any extra annotations. We evaluate the proposed OP-GAN on three publicly available datasets. The experimental results demonstrate that our OP-GAN can yield visually plausible translated images and significantly improve the semantic segmentation accuracy in different domain adaptation scenarios with off-the-shelf deep learning networks such as PSPNet and U-Net.

Due to the wide existence and large morphological variances of nuclei, accurate nuclei instance segmentation is still one of the most challenging tasks in computational pathology. The annotating of nuclei instances, requiring experienced pathologists to manually draw the contours, is extremely laborious and expensive, which often results in the deficiency of annotated data. The deep learning based segmentation approaches, which highly rely on the quantity of training data, are difficult to fully demonstrate their capacity in this area. In this paper, we propose a novel self-supervised learning framework to deeply exploit the capacity of widely-used convolutional neural networks (CNNs) on the nuclei instance segmentation task. The proposed approach involves two sub-tasks (i.e., scale-wise triplet learning and count ranking), which enable neural networks to implicitly leverage the prior-knowledge of nuclei size and quantity, and accordingly mine the instance-aware feature representations from the raw data. Experimental results on the publicly available MoNuSeg dataset show that the proposed self-supervised learning approach can remarkably boost the segmentation accuracy of nuclei instance—a new state-of-the-art average Aggregated Jaccard Index (AJI) of 70.63%, is achieved by our self-supervised ResUNet-101. To our best knowledge, this is the first work focusing on the self-supervised learning for instance segmentation.

Retinal artery/vein (A/V) classification lays the foundation for the quantitative analysis of retinal vessels, which is associated with potential risks of various cardiovascular and cerebral diseases. The topological connection relationship, which has been proved effective in improving the A/V classification performance for the conventional graph based method, has not been exploited by the deep learning based method. In this paper, we propose a Topology Ranking Generative Adversarial Network (TR-GAN) to improve the topology connectivity of the segmented arteries and veins, and further to boost the A/V classification performance. A topology ranking discriminator based on ordinal regression is proposed to rank the topological connectivity level of the ground-truth, the generated A/V mask and the intentionally shuffled mask. The ranking loss is further back-propagated to the generator to generate better connected A/V masks. In addition, a topology preserving module with triplet loss is also proposed to extract the high-level topological features and further to narrow the feature distance between the predicted A/V mask and the ground-truth. The proposed framework effectively increases the topological connectivity of the predicted A/V masks and achieves state-of-the-art A/V classification performance on the publicly available AV-DRIVE dataset.

Deep learning models, such as the fully convolutional network (FCN), have been widely used in 3D biomedical segmentation and achieved state-of-the-art performance. Multiple modalities are often used for disease diagnosis and quantification. Two approaches are widely used in the literature to fuse multiple modalities in the segmentation networks: early-fusion (which stacks multiple modalities as different input channels) and late-fusion (which fuses the segmentation results from different modalities at the very end). These fusion methods easily suffer from the cross-modal interference caused by the input modalities which have wide variations. To address the problem, we propose a novel deep learning architecture, namely OctopusNet, to better leverage and fuse the information contained in multi-modalities. The proposed framework employs a separate encoder for each modality for feature extraction and exploits a hyper-fusion decoder to fuse the extracted features while avoiding feature explosion. We evaluate the proposed OctopusNet on two publicly available datasets, i.e. ISLES-2018 and MRBrainS-2013. The experimental results show that our framework outperforms the commonly-used feature fusion approaches and yields the state-of-the-art segmentation accuracy.

Accurate lesion detection in computer tomography (CT) slices benefits pathologic organ analysis in the medical diagnosis process. More recently, it has been tackled as an object detection problem using the Convolutional Neural Networks (CNNs). Despite the achievements from off-the-shelf CNN models, the current detection accuracy is limited by the inability of CNNs on lesions at vastly different scales. In this paper, we propose a Multi-Scale Booster (MSB) with channel and spatial attention integrated into the backbone Feature Pyramid Network (FPN). In each pyramid level, the proposed MSB captures fine-grained scale variations by using Hierarchically Dilated Convolutions (HDC). Meanwhile, the proposed channel and spatial attention modules increase the network’s capability of selecting relevant features response for lesion detection. Extensive experiments on the DeepLesion benchmark dataset demonstrate that the proposed method performs superiorly against state-of-the-art approaches.

Retinal artery/vein (A/V) classification plays a critical role in the clinical biomarker study of how various systemic and cardiovascular diseases affect the retinal vessels. Conventional methods of automated A/V classification are generally complicated and heavily depend on the accurate vessel segmentation. In this paper, we propose a multi-task deep neural network with spatial activation mechanism that is able to segment full retinal vessel, artery and vein simultaneously, without the pre-requirement of vessel segmentation. The input module of the network integrates the domain knowledge of widely used retinal preprocessing and vessel enhancement techniques. We specially customize the output block of the network with a spatial activation mechanism, which takes advantage of a relatively easier task of vessel segmentation and exploits it to boost the performance of A/V classification. In addition, deep supervision is introduced to the network to assist the low level layers to extract more semantic information. The proposed network achieves pixel-wise accuracy of 95.70% for vessel segmentation, and A/V classification accuracy of 94.50%, which is the state-of-the-art performance for both tasks on the AV-DRIVE dataset. Furthermore, we have also tested the model performance on INSPIRE-AVR dataset, which achieves a skeletal A/V classification accuracy of 91.6%.

Semantic segmentation has been popularly addressed using fully convolutional networks (FCNs) with impressive results if the training set is diverse and large enough. However, FCNs often fail to achieve satisfactory results due to a limited number of manually labelled samples in medical imaging. In this paper, we propose a conjugate fully convolutional network (CFCN) to address this challenging problem. CFCN is a novel framework where pairwise samples are input and synergistically segmented in the network for capturing a rich context representation. To avoid overfitting introduced by appearance and shape changes in a small number of training samples, a fusion module is designed to provide proxy supervision for the network training process. Quantitative evaluation shows that the proposed method has a significant performance improvement on pathological liver segmentation.

Skip connections in deep networks have improved both segmentation and classification performance by facilitating the training of deeper network architectures and reducing the risks for vanishing gradients. The skip connections equip encoder-decoder like networks with richer feature representations, but at the cost of higher memory usage, computation, and possibly resulting in transferring non-discriminative feature maps. In this paper, we focus on improving the skip connections used in segmentation networks. We propose light, learnable skip connections which learn to first select the most discriminative channels, and then aggregate the selected ones as single channel attending to the most discriminative regions of input. We evaluate the proposed method on 3 different 2D and volumetric datasets and demonstrate that the proposed skip connections can outperform the traditional heavy skip connections of 4 different models in terms of segmentation accuracy (2% Dice), memory usage (at least 50%), and the number of network parameters (up to 70%).

Accurately segmenting left atrium in MR volume can benefit the ablation procedure of atrial fibrillation. Traditional automated solutions often fail in relieving experts from the labor-intensive manual labeling. In this paper, we propose a deep neural network based solution for automated left atrium segmentation in gadolinium-enhanced MR volumes with promising performance. We firstly argue that, for this volumetric segmentation task, networks in 2D fashion can present great superiorities in time efficiency and segmentation accuracy than networks with 3D fashion. Considering the highly varying shape of atrium and the branchy structure of associated pulmonary veins, we propose to adopt a pyramid module to collect semantic cues in feature maps from multiple scales for fine-grained segmentation. Also, to promote our network in classifying the hard examples, we propose an Online Hard Negative Example Mining strategy to identify voxels in slices with low classification certainties and penalize the wrong predictions on them. Finally, we devise a competitive training scheme to further boost the generalization ability of networks. Extensively verified on 20 testing volumes, our proposed framework achieves an average Dice of \(92.83\%\) in segmenting the left atria and pulmonary veins.

Unlike a conventional background inpainting approach that infers a missing area from image patches similar to the background, face completion requires semantic knowledge about the target object for realistic outputs. Current image inpainting approaches utilize generative adversarial networks (GANs) to achieve such semantic understanding. However, in adversarial learning, the semantic knowledge is learned implicitly and hence good semantic understanding is not always guaranteed. In this work, we propose a collaborative adversarial learning approach to face completion to explicitly induce the training process. Our method is formulated under a novel generative framework called collaborative GAN (collaGAN), which allows better semantic understanding of a target object through collaborative learning of multiple tasks including face completion, landmark detection and semantic segmentation. Together with the collaGAN, we also introduce an inpainting concentrated scheme such that the model emphasizes more on inpainting instead of autoencoding. Extensive experiments show that the proposed designs are indeed effective and collaborative adversarial learning provides better feature representations of the faces. In comparison with other generative image inpainting models and single task learning methods, our solution produces superior performances on all tasks.

Witnessed the development of deep learning, increasing number of studies try to build computer aided diagnosis systems for 3D volumetric medical data. However, as the annotations of 3D medical data are difficult to acquire, the number of annotated 3D medical images is often not enough to well train the deep learning networks. The self-supervised learning deeply exploiting the information of raw data is one of the potential solutions to loose the requirement of training data. In this paper, we propose a self-supervised learning framework for the volumetric medical images. A novel proxy task, i.e., Rubik’s cube recovery, is formulated to pre-train 3D neural networks. The proxy task involves two operations, i.e., cube rearrangement and cube rotation, which enforce networks to learn translational and rotational invariant features from raw 3D data. Compared to the train-from-scratch strategy, fine-tuning from the pre-trained network leads to a better accuracy on various tasks, e.g., brain hemorrhage classification and brain tumor segmentation. We show that our self-supervised learning approach can substantially boost the accuracies of 3D deep learning networks on the volumetric medical datasets without using extra data. To our best knowledge, this is the first work focusing on the self-supervised learning of 3D neural networks.

Accurate vessel centerline tracing greatly benefits vessel centerline geometry assessment and facilitates precise measurements of vessel diameters and lengths. However, cursive and longitudinal geometries of vessels make centerline tracing a challenging task in volumetric images. Treating the problem with traditional feature handcrafting is often ad-hoc and time-consuming, resulting in suboptimal solutions. In this work, we propose a unified end-to-end deep reinforcement learning approach for robust vessel centerline tracing in multi-modality 3D medical volumes. Instead of time-consuming exhaustive search in 3D space, we propose to learn an artificial agent to interact with surrounding environment and collect rewards from the interaction. A deep neural network is integrated to the system to predict stepwise action value for every possible actions. With this mechanism, the agent is able to probe through an optimal navigation path to trace the vessel centerline. Our proposed approach is evaluated on a dataset of over 2,000 3D volumes with diverse imaging modalities, including contrasted CT, non-contrasted CT, C-arm CT and MR images. The experimental results show that the proposed approach can handle large variations from vessel shape to imaging characteristics, with a tracing error as low as 3.28 mm and detection time as fast as 1.71 s per volume.

Recently, deep learning has demonstrated great success in computer vision with the capability to learn powerful image features from a large training set. However, most of the published work has been confined to solving 2D problems, with a few limited exceptions that treated the 3D space as a composition of 2D orthogonal planes. The challenge of 3D deep learning is due to a much larger input vector, compared to 2D, which dramatically increases the computation time and the chance of over-fitting, especially when combined with limited training samples (hundreds to thousands), typical for medical imaging applications. To address this challenge, we propose an efficient and robust deep learning algorithm capable of full 3D detection in volumetric data. A two-step approach is exploited for efficient detection. A shallow network (with one hidden layer) is used for the initial testing of all voxels to obtain a small number of promising candidates, followed by more accurate classification with a deep network. In addition, we propose two approaches, i.e., separable filter decomposition and network sparsification, to speed up the evaluation of a network. To mitigate the over-fitting issue, thereby increasing detection robustness, we extract small 3D patches from a multi-resolution image pyramid. The deeply learned image features are further combined with Haar wavelet-like features to increase the detection accuracy. The proposed method has been quantitatively evaluated for carotid artery bifurcation detection on a head-neck CT dataset from 455 patients. Compared to the state of the art, the mean error is reduced by more than half, from 5.97 mm to 2.64 mm, with a detection speed of less than 1 s/volume.

Chronic kidney disease affects one of every ten adults in USA (over 20 million). Computed tomography (CT) is a widely used imaging modality for kidney disease diagnosis and quantification. However, automatic pathological kidney segmentation is still a challenging task due to large variations in contrast phase, scanning range, pathology, and position in the abdomen, etc. Methods based on global image context (e.g., atlas- or regression-based approaches) do not work well. In this work, we propose to combine deep learning and marginal space learning (MSL), both using local context, for robust kidney detection and segmentation. Here, deep learning is exploited to roughly estimate the kidney center. Instead of performing a whole axial slice classification (i.e., whether it contains a kidney), we detect local image patches containing a kidney. The detected patches are aggregated to generate an estimate of the kidney center. Afterwards, we apply MSL to further refine the pose estimate by constraining the position search to a neighborhood around the initial center. The kidney is then segmented using a discriminative active shape model. The proposed method has been trained on 370 CT scans and tested on 78 unseen cases. It achieves a mean segmentation error of 2.6 and 1.7 mm for the left and right kidney, respectively. Furthermore, it eliminates all gross failures (i.e., segmentation is totally off) in a direct application of MSL.

Computerized algorithms and solutions in processing and diagnosis mammography X-ray, cardiovascular CT/MRI scans, and microscopy image play an important role in disease detection and computer-aided decision-making. Machine learning techniques have powered many aspects in medical investigations and clinical practice. Recently, deep learning is emerging a leading machine learning tool in computer vision and begins attracting considerable attentions in medical imaging. In this chapter, we provide a snapshot of this fast growing field specifically for mammography, cardiovascular, and microscopy image analysis . We briefly explain the popular deep neural networks and summarize current deep learning achievements in various tasks such as detection, segmentation, and classification in these heterogeneous imaging modalities. In addition, we discuss the challenges and the potential future trends for ongoing work.