End-to-End Adversarial Retinal Image Synthesis

A Tour of Unsupervised Deep Learning for Medical Image Analysis

BACKGROUND

Interpretation of medical images for the diagnosis and treatment of complex diseases from high-dimensional and heterogeneous data remains a key challenge in transforming healthcare. In the last few years, both supervised and unsupervised deep learning achieved promising results in the area of medical image analysis. Several reviews on supervised deep learning are published, but hardly any rigorous review on unsupervised deep learning for medical image analysis is available.

OBJECTIVES

The objective of this review is to systematically present various unsupervised deep learning models, tools, and benchmark datasets applied to medical image analysis. Some of the discussed models are autoencoders and its other variants, Restricted Boltzmann machines (RBM), Deep belief networks (DBN), Deep Boltzmann machine (DBM), and Generative adversarial network (GAN). Further, future research opportunities and challenges of unsupervised deep learning techniques for medical image analysis are also discussed.

CONCLUSION

Currently, interpretation of medical images for diagnostic purposes is usually performed by human experts that may be replaced by computer-aided diagnosis due to advancement in machine learning techniques, including deep learning, and the availability of cheap computing infrastructure through cloud computing. Both supervised and unsupervised machine learning approaches are widely applied in medical image analysis, each of them having certain pros and cons. Since human supervisions are not always available or inadequate or biased, therefore, unsupervised learning algorithms give a big hope with lots of advantages for biomedical image analysis.

Applications of deep learning in fundus images: A review

The use of fundus images for the early screening of eye diseases is of great clinical importance. Due to its powerful performance, deep learning is becoming more and more popular in related applications, such as lesion segmentation, biomarkers segmentation, disease diagnosis and image synthesis. Therefore, it is very necessary to summarize the recent developments in deep learning for fundus images with a review paper. In this review, we introduce 143 application papers with a carefully designed hierarchy. Moreover, 33 publicly available datasets are presented. Summaries and analyses are provided for each task. Finally, limitations common to all tasks are revealed and possible solutions are given. We will also release and regularly update the state-of-the-art results and newly-released datasets at https://github.com/nkicsl/Fundus_Review to adapt to the rapid development of this field.

DR-GAN: Conditional Generative Adversarial Network for Fine-Grained Lesion Synthesis on Diabetic Retinopathy Images

Diabetic retinopathy (DR) is a complication of diabetes that severely affects eyes. It can be graded into five levels of severity according to international protocol. However, optimizing a grading model to have strong generalizability requires a large amount of balanced training data, which is difficult to collect, particularly for the high severity levels. Typical data augmentation methods, including random flipping and rotation, cannot generate data with high diversity. In this paper, we propose a diabetic retinopathy generative adversarial network (DR-GAN) to synthesize high-resolution fundus images which can be manipulated with arbitrary grading and lesion information. Thus, large-scale generated data can be used for more meaningful augmentation to train a DR grading and lesion segmentation model. The proposed retina generator is conditioned on the structural and lesion masks, as well as adaptive grading vectors sampled from the latent grading space, which can be adopted to control the synthesized grading severity. Moreover, a multi-scale spatial and channel attention module is devised to improve the generation ability to synthesize small details. Multi-scale discriminators are designed to operate from large to small receptive fields, and joint adversarial losses are adopted to optimize the whole network in an end-to-end manner. With extensive experiments evaluated on the EyePACS dataset connected to Kaggle, as well as the FGADR dataset, we validate the effectiveness of our method, which can both synthesize highly realistic (1280 × 1280) controllable fundus images and contribute to the DR grading task.

Semi-supervised task-driven data augmentation for medical image segmentation

Supervised learning-based segmentation methods typically require a large number of annotated training data to generalize well at test time. In medical applications, curating such datasets is not a favourable option because acquiring a large number of annotated samples from experts is time-consuming and expensive. Consequently, numerous methods have been proposed in the literature for learning with limited annotated examples. Unfortunately, the proposed approaches in the literature have not yet yielded significant gains over random data augmentation for image segmentation, where random augmentations themselves do not yield high accuracy. In this work, we propose a novel task-driven data augmentation method for learning with limited labeled data where the synthetic data generator, is optimized for the segmentation task. The generator of the proposed method models intensity and shape variations using two sets of transformations, as additive intensity transformations and deformation fields. Both transformations are optimized using labeled as well as unlabeled examples in a semi-supervised framework. Our experiments on three medical datasets, namely cardiac, prostate and pancreas, show that the proposed approach significantly outperforms standard augmentation and semi-supervised approaches for image segmentation in the limited annotation setting. The code is made publicly available at https://github.com/krishnabits001/task_driven_data_augmentation.

EpilepsyGAN: Synthetic Epileptic Brain Activities With Privacy Preservation

Epilepsy is a chronic neurological disorder affecting more than 65 million people worldwide and manifested by recurrent unprovoked seizures. The unpredictability of seizures not only degrades the quality of life of the patients, but it can also be life-threatening. Modern systems monitoring electroencephalography (EEG) signals are being currently developed with the view to detect epileptic seizures in order to alert caregivers and reduce the impact of seizures on patients' quality of life. Such seizure detection systems employ state-of-the-art machine learning algorithms that require a large amount of labeled personal data for training. However, acquiring EEG signals during epileptic seizures is a costly and time-consuming process for medical experts and patients. Furthermore, this data often contains sensitive personal information, presenting privacy concerns. In this work, we generate synthetic seizure-like brain electrical activities, i.e., EEG signals, that can be used to train seizure detection algorithms, alleviating the need for sensitive recorded data. Our experiments show that the synthetic seizure data generated with our GAN model succeeds at preserving the privacy of the patients without producing any degradation in performance during seizure monitoring.

Machine Learning Techniques for Ophthalmic Data Processing: A Review

Machine learning and especially deep learning techniques are dominating medical image and data analysis. This article reviews machine learning approaches proposed for diagnosing ophthalmic diseases during the last four years. Three diseases are addressed in this survey, namely diabetic retinopathy, age-related macular degeneration, and glaucoma. The review covers over 60 publications and 25 public datasets and challenges related to the detection, grading, and lesion segmentation of the three considered diseases. Each section provides a summary of the public datasets and challenges related to each pathology and the current methods that have been applied to the problem. Furthermore, the recent machine learning approaches used for retinal vessels segmentation, and methods of retinal layers and fluid segmentation are reviewed. Two main imaging modalities are considered in this survey, namely color fundus imaging, and optical coherence tomography. Machine learning approaches that use eye measurements and visual field data for glaucoma detection are also included in the survey. Finally, the authors provide their views, expectations and the limitations of the future of these techniques in the clinical practice.

Generative models in pathology: synthesis of diagnostic quality pathology images†

Within artificial intelligence and machine learning, a generative model is a powerful tool for learning any kind of data distribution. With the advent of deep learning and its success in image recognition, the field of deep generative models has clearly emerged as one of the promising fields for medical imaging. In a recent issue of The Journal of Pathology, Levine, Peng et al demonstrate the ability of generative models to synthesize high-quality pathology images. They suggested that generative models can serve as an unlimited source of images either for educating freshman pathologists or training machine learning models for diverse image analysis tasks, especially in scarce cases, while resolving patients' privacy and confidentiality concerns. © 2020 The Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Self-Supervised Feature Learning via Exploiting Multi-Modal Data for Retinal Disease Diagnosis

The automatic diagnosis of various retinal diseases from fundus images is important to support clinical decision-making. However, developing such automatic solutions is challenging due to the requirement of a large amount of human-annotated data. Recently, unsupervised/self-supervised feature learning techniques receive a lot of attention, as they do not need massive annotations. Most of the current self-supervised methods are analyzed with single imaging modality and there is no method currently utilize multi-modal images for better results. Considering that the diagnostics of various vitreoretinal diseases can greatly benefit from another imaging modality, e.g., FFA, this paper presents a novel self-supervised feature learning method by effectively exploiting multi-modal data for retinal disease diagnosis. To achieve this, we first synthesize the corresponding FFA modality and then formulate a patient feature-based softmax embedding objective. Our objective learns both modality-invariant features and patient-similarity features. Through this mechanism, the neural network captures the semantically shared information across different modalities and the apparent visual similarity between patients. We evaluate our method on two public benchmark datasets for retinal disease diagnosis. The experimental results demonstrate that our method clearly outperforms other self-supervised feature learning methods and is comparable to the supervised baseline. Our code is available at GitHub.

Self-Supervised Ultrasound to MRI Fetal Brain Image Synthesis

Fetal brain magnetic resonance imaging (MRI) offers exquisite images of the developing brain but is not suitable for second-trimester anomaly screening, for which ultrasound (US) is employed. Although expert sonographers are adept at reading US images, MR images which closely resemble anatomical images are much easier for non-experts to interpret. Thus in this article we propose to generate MR-like images directly from clinical US images. In medical image analysis such a capability is potentially useful as well, for instance for automatic US-MRI registration and fusion. The proposed model is end-to-end trainable and self-supervised without any external annotations. Specifically, based on an assumption that the US and MRI data share a similar anatomical latent space, we first utilise a network to extract the shared latent features, which are then used for MRI synthesis. Since paired data is unavailable for our study (and rare in practice), pixel-level constraints are infeasible to apply. We instead propose to enforce the distributions to be statistically indistinguishable, by adversarial learning in both the image domain and feature space. To regularise the anatomical structures between US and MRI during synthesis, we further propose an adversarial structural constraint. A new cross-modal attention technique is proposed to utilise non-local spatial information, by encouraging multi-modal knowledge fusion and propagation. We extend the approach to consider the case where 3D auxiliary information (e.g., 3D neighbours and a 3D location index) from volumetric data is also available, and show that this improves image synthesis. The proposed approach is evaluated quantitatively and qualitatively with comparison to real fetal MR images and other approaches to synthesis, demonstrating its feasibility of synthesising realistic MR images.

TTTS-STgan: Stacked Generative Adversarial Networks for TTTS Fetal Surgery Planning Based on 3D Ultrasound

Twin-to-twin transfusion syndrome (TTTS) is characterized by an unbalanced blood transfer through placental abnormal vascular connections. Prenatal ultrasound (US) is the imaging technique to monitor monochorionic pregnancies and diagnose TTTS. Fetoscopic laser photocoagulation is an elective treatment to coagulate placental communications between both twins. To locate the anomalous connections ahead of surgery, preoperative planning is crucial. In this context, we propose a novel multi-task stacked generative adversarial framework to jointly learn synthetic fetal US generation, multi-class segmentation of the placenta, its inner acoustic shadows and peripheral vasculature, and placenta shadowing removal. Specifically, the designed architecture is able to learn anatomical relationships and global US image characteristics. In addition, we also extract for the first time the umbilical cord insertion on the placenta surface from 3D HD-flow US images. The database consisted of 70 US volumes including singleton, mono- and dichorionic twins at 17-37 gestational weeks. Our experiments show that 71.8% of the synthesized US slices were categorized as realistic by clinicians, and that the multi-class segmentation achieved Dice scores of 0.82 ± 0.13, 0.71 ± 0.09, and 0.72 ± 0.09, for placenta, acoustic shadows, and vasculature, respectively. Moreover, fetal surgeons classified 70.2% of our completed placenta shadows as satisfactory texture reconstructions. The umbilical cord was successfully detected on 85.45% of the volumes. The framework developed could be implemented in a TTTS fetal surgery planning software to improve the intrauterine scene understanding and facilitate the location of the optimum fetoscope entry point.

Adversarial Confidence Learning for Medical Image Segmentation and Synthesis

Generative adversarial networks (GAN) are widely used in medical image analysis tasks, such as medical image segmentation and synthesis. In these works, adversarial learning is directly applied to the original supervised segmentation (synthesis) networks. The usage of adversarial learning is effective in improving visual perception performance since adversarial learning works as realistic regularization for supervised generators. However, the quantitative performance often cannot improve as much as the qualitative performance, and it can even become worse in some cases. In this paper, we explore how we can take better advantage of adversarial learning in supervised segmentation (synthesis) models and propose an adversarial confidence learning framework to better model these problems. We analyze the roles of discriminator in the classic GANs and compare them with those in supervised adversarial systems. Based on this analysis, we propose adversarial confidence learning, i.e., besides the adversarial learning for emphasizing visual perception, we use the confidence information provided by the adversarial network to enhance the design of supervised segmentation (synthesis) network. In particular, we propose using a fully convolutional adversarial network for confidence learning to provide voxel-wise and region-wise confidence information for the segmentation (synthesis) network. With these settings, we propose a difficulty-aware attention mechanism to properly handle hard samples or regions by taking structural information into consideration so that we can better deal with the irregular distribution of medical data. Furthermore, we investigate the loss functions of various GANs and propose using the binary cross entropy loss to train the proposed adversarial system so that we can retain the unlimited modeling capacity of the discriminator. Experimental results on clinical and challenge datasets show that our proposed network can achieve state-of-the-art segmentation (synthesis) accuracy. Further analysis also indicates that adversarial confidence learning can both improve the visual perception performance and the quantitative performance.

SHIFT: speedy histological-to-immunofluorescent translation of a tumor signature enabled by deep learning

Spatially-resolved molecular profiling by immunostaining tissue sections is a key feature in cancer diagnosis, subtyping, and treatment, where it complements routine histopathological evaluation by clarifying tumor phenotypes. In this work, we present a deep learning-based method called speedy histological-to-immunofluorescent translation (SHIFT) which takes histologic images of hematoxylin and eosin (H&E)-stained tissue as input, then in near-real time returns inferred virtual immunofluorescence (IF) images that estimate the underlying distribution of the tumor cell marker pan-cytokeratin (panCK). To build a dataset suitable for learning this task, we developed a serial staining protocol which allows IF and H&E images from the same tissue to be spatially registered. We show that deep learning-extracted morphological feature representations of histological images can guide representative sample selection, which improved SHIFT generalizability in a small but heterogenous set of human pancreatic cancer samples. With validation in larger cohorts, SHIFT could serve as an efficient preliminary, auxiliary, or substitute for panCK IF by delivering virtual panCK IF images for a fraction of the cost and in a fraction of the time required by traditional IF.

Written in Blood: Applying Shape Grammars to Retinal Vasculatures

Purpose

Blood vessel networks within the retina are crucial for maintaining tissue perfusion and therefore good vision. Their complexity and unique patterns often require a steep learning curve for humans to identify trends and changes in the shape and topology of the networks, even though there exists much information important to identifying disease within them.

Methods

Through image processing, the vasculature is isolated from other features of the fundus images, forcing the viewer to focus on the complex vascular feature. This article explores an approach using a grammar based on shape to describe retinal vasculature and to generate realistic and increasingly unrealistic artificial vascular networks that are then reviewed by ophthalmologists via digital survey. The ophthalmologists are asked whether these artificial vascular networks appeared realistic or unrealistic.

Results

With only three rules (initiate, branch, and curve), the grammar accomplishes these goals. Networks are generated by adding noise to rule parameters present in existing networks. Via the survey of synthetic networks generated with different noise parameters, a correlation between noise in the branch rule and realistic association is revealed.

Conclusions

By creating a language to describe retinal vasculature, this article allows for the potential of new insight into such an important but less understood feature of the retina, which in the future may play a role in diagnosing or helping to predict types of ocular disease.

Translational Relevance

Applying shape grammar to describe retinal vasculature permits new understanding, which in turn provides the potential for new diagnostic tools.

Predicting PET-derived myelin content from multisequence MRI for individual longitudinal analysis in multiple sclerosis

Multiple sclerosis (MS) is a demyelinating and inflammatory disease of the central nervous system (CNS). The demyelination process can be repaired by the generation of a new sheath of myelin around the axon, a process termed remyelination. In MS patients, the demyelination-remyelination cycles are highly dynamic. Over the years, magnetic resonance imaging (MRI) has been increasingly used in the diagnosis of MS and it is currently the most useful paraclinical tool to assess this diagnosis. However, conventional MRI pulse sequences are not specific for pathological mechanisms such as demyelination and remyelination. Recently, positron emission tomography (PET) with radiotracer [¹¹C]PIB has become a promising tool to measure in-vivo myelin content changes which is essential to push forward our understanding of mechanisms involved in the pathology of MS, and to monitor individual patients in the context of clinical trials focused on repair therapies. However, PET imaging is invasive due to the injection of a radioactive tracer. Moreover, it is an expensive imaging test and not offered in the majority of medical centers in the world. In this work, by using multisequence MRI, we thus propose a method to predict the parametric map of [¹¹C]PIB PET, from which we derived the myelin content changes in a longitudinal analysis of patients with MS. The method is based on the proposed conditional flexible self-attention GAN (CF-SAGAN) which is specifically adjusted for high-dimensional medical images and able to capture the relationships between the spatially separated lesional regions during the image synthesis process. Jointly applying the sketch-refinement process and the proposed attention regularization that focuses on the MS lesions, our approach is shown to outperform the state-of-the-art methods qualitatively and quantitatively. Specifically, our method demonstrated a superior performance for the prediction of myelin content at voxel-wise level. More important, our method for the prediction of myelin content changes in patients with MS shows similar clinical correlations to the PET-derived gold standard indicating the potential for clinical management of patients with MS.

Hi-Net: Hybrid-Fusion Network for Multi-Modal MR Image Synthesis

Magnetic resonance imaging (MRI) is a widely used neuroimaging technique that can provide images of different contrasts (i.e., modalities). Fusing this multi-modal data has proven particularly effective for boosting model performance in many tasks. However, due to poor data quality and frequent patient dropout, collecting all modalities for every patient remains a challenge. Medical image synthesis has been proposed as an effective solution, where any missing modalities are synthesized from the existing ones. In this paper, we propose a novel Hybrid-fusion Network (Hi-Net) for multi-modal MR image synthesis, which learns a mapping from multi-modal source images (i.e., existing modalities) to target images (i.e., missing modalities). In our Hi-Net, a modality-specific network is utilized to learn representations for each individual modality, and a fusion network is employed to learn the common latent representation of multi-modal data. Then, a multi-modal synthesis network is designed to densely combine the latent representation with hierarchical features from each modality, acting as a generator to synthesize the target images. Moreover, a layer-wise multi-modal fusion strategy effectively exploits the correlations among multiple modalities, where a Mixed Fusion Block (MFB) is proposed to adaptively weight different fusion strategies. Extensive experiments demonstrate the proposed model outperforms other state-of-the-art medical image synthesis methods.

An Adversarial Learning Approach to Medical Image Synthesis for Lesion Detection

The identification of lesion within medical image data is necessary for diagnosis, treatment and prognosis. Segmentation and classification approaches are mainly based on supervised learning with well-paired image-level or voxel-level labels. However, labeling the lesion in medical images is laborious requiring highly specialized knowledge. We propose a medical image synthesis model named abnormal-to-normal translation generative adversarial network (ANT-GAN) to generate a normal-looking medical image based on its abnormal-looking counterpart without the need for paired training data. Unlike typical GANs, whose aim is to generate realistic samples with variations, our more restrictive model aims at producing a normal-looking image corresponding to one containing lesions, and thus requires a special design. Being able to provide a "normal" counterpart to a medical image can provide useful side information for medical imaging tasks like lesion segmentation or classification validated by our experiments. In the other aspect, the ANT-GAN model is also capable of producing highly realistic lesion-containing image corresponding to the healthy one, which shows the potential in data augmentation verified in our experiments.

Synthetic image augmentation with generative adversarial network for enhanced performance in protein classification

Proteins are complex macromolecules accountable for the biological processes in the cell. In biomedical research, the images of protein are extensively used in medicine. The rate at which these images are produced makes it difficult to evaluate them manually and hence there exists a need to automate the system. The quality of images is still a major issue. In this paper, we present the use of different image enhancement techniques that improves the contrast of these images. Besides the quality of images, the challenge of gathering such datasets in the field of medicine persists. We use generative adversarial networks for generating synthetic samples to ameliorate the results of CNN. The performance of the synthetic data augmentation was compared with the classic data augmentation on the classification task, an increase of 2.7% in Macro F1 and 2.64% in Micro F1 score was observed. Our best results were obtained by the pretrained Inception V4 model that gave a fivefold cross-validated macro F1 of 0.603. The achieved results are contrasted with the existing work and comparisons show that the proposed method outperformed.

Melanoma detection using adversarial training and deep transfer learning

Skin lesion datasets consist predominantly of normal samples with only a small percentage of abnormal ones, giving rise to the class imbalance problem. Also, skin lesion images are largely similar in overall appearance owing to the low inter-class variability. In this paper, we propose a two-stage framework for automatic classification of skin lesion images using adversarial training and transfer learning toward melanoma detection. In the first stage, we leverage the inter-class variation of the data distribution for the task of conditional image synthesis by learning the inter-class mapping and synthesizing under-represented class samples from the over-represented ones using unpaired image-to-image translation. In the second stage, we train a deep convolutional neural network for skin lesion classification using the original training set combined with the newly synthesized under-represented class samples. The training of this classifier is carried out by minimizing the focal loss function, which assists the model in learning from hard examples, while down-weighting the easy ones. Experiments conducted on a dermatology image benchmark demonstrate the superiority of our proposed approach over several standard baseline methods, achieving significant performance improvements. Interestingly, we show through feature visualization and analysis that our method leads to context based lesion assessment that can reach an expert dermatologist level.

A visual framework to create photorealistic retinal vessels for diagnosis purposes

The methods developed in recent years for synthesising an ocular fundus can be been divided into two main categories. The first category of methods involves the development of an anatomical model of the eye, where artificial images are generated using appropriate parameters for modelling the vascular networks and fundus. The second type of method has been made possible by the development of deep learning techniques and improvements in the performance of hardware (especially graphics cards equipped with a large number of cores). The methodology proposed here to produce high-resolution synthetic fundus images is intended to be an alternative to the increasingly widespread use of generative adversarial networks to overcome the problems that arise in producing slightly modified versions of the same real images. This will allow the simulation of pathologies and the prediction of eye-related diseases. The proposed approach is based on the principle of least action and correctly places the vessels on the simulated eye fundus without using real morphometric information. An a posteriori analysis of the average characteristics such as the size, length, bifurcations, and endpoint positioning confirmed the substantial accuracy of the proposed approach compared to real data. A graphical user interface allows the user to make any changes in real time by controlling the positions of control points.

Deep learning-based multi-view fusion model for screening 2019 novel coronavirus pneumonia: A multicentre study

PURPOSE

To develop a deep learning-based method to assist radiologists to fast and accurately identify patients with COVID-19 by CT images.

METHODS

We retrospectively collected chest CT images of 495 patients from three hospitals in China. 495 datasets were randomly divided into 395 cases (80%, 294 of COVID-19, 101 of other pneumonia) of the training set, 50 cases (10%, 37 of COVID-19, 13 of other pneumonia) of the validation set and 50 cases (10%, 37 of COVID-19, 13 of other pneumonia) of the testing set. We trained a multi-view fusion model using deep learning network to screen patients with COVID-19 using CT images with the maximum lung regions in axial, coronal and sagittal views. The performance of the proposed model was evaluated by both the validation and testing sets.

RESULTS

The multi-view deep learning fusion model achieved the area under the receiver-operating characteristics curve (AUC) of 0.732, accuracy of 0.700, sensitivity of 0.730 and specificity of 0.615 in validation set. In the testing set, we can achieve AUC, accuracy, sensitivity and specificity of 0.819, 0.760, 0.811 and 0.615 respectively.

CONCLUSIONS

Based on deep learning method, the proposed diagnosis model trained on multi-view images of chest CT images showed great potential to improve the efficacy of diagnosis and mitigate the heavy workload of radiologists for the initial screening of COVID-19 pneumonia.

Sample-Adaptive GANs: Linking Global and Local Mappings for Cross-Modality MR Image Synthesis

Generative adversarial network (GAN) has been widely explored for cross-modality medical image synthesis. The existing GAN models usually adversarially learn a global sample space mapping from the source-modality to the target-modality and then indiscriminately apply this mapping to all samples in the whole space for prediction. However, due to the scarcity of training samples in contrast to the complicated nature of medical image synthesis, learning a single global sample space mapping that is "optimal" to all samples is very challenging, if not intractable. To address this issue, this paper proposes sample-adaptive GAN models, which not only cater for the global sample space mapping between the source- and the target-modalities but also explore the local space around each given sample to extract its unique characteristic. Specifically, the proposed sample-adaptive GANs decompose the entire learning model into two cooperative paths. The baseline path learns a common GAN model by fitting all the training samples as usual for the global sample space mapping. The new sample-adaptive path additionally models each sample by learning its relationship with its neighboring training samples and using the target-modality features of these training samples as auxiliary information for synthesis. Enhanced by this sample-adaptive path, the proposed sample-adaptive GANs are able to flexibly adjust themselves to different samples, and therefore optimize the synthesis performance. Our models have been verified on three cross-modality MR image synthesis tasks from two public datasets, and they significantly outperform the state-of-the-art methods in comparison. Moreover, the experiment also indicates that our sample-adaptive strategy could be utilized to improve various backbone GAN models. It complements the existing GANs models and can be readily integrated when needed.

Retinal Drug Delivery: Rethinking Outcomes for the Efficient Replication of Retinal Behavior

The retina is a highly organized structure that is considered to be "an approachable part of the brain." It is attracting the interest of development scientists, as it provides a model neurovascular system. Over the last few years, we have been witnessing significant development in the knowledge of the mechanisms that induce the shape of the retinal vascular system, as well as knowledge of disease processes that lead to retina degeneration. Knowledge and understanding of how our vision works are crucial to creating a hardware-adaptive computational model that can replicate retinal behavior. The neuronal system is nonlinear and very intricate. It is thus instrumental to have a clear view of the neurophysiological and neuroanatomic processes and to take into account the underlying principles that govern the process of hardware transformation to produce an appropriate model that can be mapped to a physical device. The mechanistic and integrated computational models have enormous potential toward helping to understand disease mechanisms and to explain the associations identified in large model-free data sets. The approach used is modulated and based on different models of drug administration, including the geometry of the eye. This work aimed to review the recently used mathematical models to map a directed retinal network.

Synthesis of fracture radiographs with deep neural networks

Purpose

We describe a machine learning system for converting diagrams of fractures into realistic X-ray images. We further present a method for iterative, human-guided refinement of the generated images and show that the resulting synthetic images can be used during training to increase the accuracy of deep classifiers on clinically meaningful subsets of fracture X-rays.

Methods

A neural network was trained to reconstruct images from programmatically created line drawings of those images. The images were then further refined with an optimization-based technique. Ten physicians were recruited into a study to assess the realism of synthetic radiographs created by the neural network. They were presented with mixed sets of real and synthetic images and asked to identify which images were synthetic. Two classifiers were trained to detect humeral shaft fractures: one only on true fracture images, and one on both true and synthetic images.

Results

Physicians were 49.63% accurate in identifying whether images were synthetic or real. This is close to what would be expected by pure chance (i.e. random guessing). A classifier trained only on real images detected fractures with 67.21% sensitivity when no fracture fixation hardware was present. A classifier trained on both real images and synthetic images was 75.54% sensitive.

Conclusion

Our method generates X-rays realistic enough to be indistinguishable from real X-rays. We also show that synthetic images generated using this method can be used to increase the accuracy of deep classifiers on clinically meaningful subsets of fracture X-rays.

Intelligent Image Synthesis for Accurate Retinal Diagnosis

Ophthalmology is a core medical field that is of interest to many. Retinal examination is a commonly performed diagnostic procedure that can be used to inspect the interior of the eye and screen for any pathological symptoms. Although various types of eye examinations exist, there are many cases where it is difficult to identify the retinal condition of the patient accurately because the test image resolution is very low because of the utilization of simple methods. In this paper, we propose an image synthetic approach that reconstructs the vessel image based on past retinal image data using the multilayer perceptron concept with artificial neural networks. The approach proposed in this study can convert vessel images to vessel-centered images with clearer identification, even for low-resolution retinal images. To verify the proposed approach, we determined whether high-resolution vessel images could be extracted from low-resolution images through a statistical analysis using high- and low-resolution images extracted from the same patient.

GAN-based synthetic brain PET image generation

In recent days, deep learning technologies have achieved tremendous success in computer vision-related tasks with the help of large-scale annotated dataset. Obtaining such dataset for medical image analysis is very challenging. Working with the limited dataset and small amount of annotated samples makes it difficult to develop a robust automated disease diagnosis model. We propose a novel approach to generate synthetic medical images using generative adversarial networks (GANs). Our proposed model can create brain PET images for three different stages of Alzheimer’s disease—normal control (NC), mild cognitive impairment (MCI), and Alzheimer’s disease (AD).

Detection of Early Signs of Diabetic Retinopathy Based on Textural and Morphological Information in Fundus Images

Estimated blind people in the world will exceed 40 million by 2025. To develop novel algorithms based on fundus image descriptors that allow the automatic classification of retinal tissue into healthy and pathological in early stages is necessary. In this paper, we focus on one of the most common pathologies in the current society: diabetic retinopathy. The proposed method avoids the necessity of lesion segmentation or candidate map generation before the classification stage. Local binary patterns and granulometric profiles are locally computed to extract texture and morphological information from retinal images. Different combinations of this information feed classification algorithms to optimally discriminate bright and dark lesions from healthy tissues. Through several experiments, the ability of the proposed system to identify diabetic retinopathy signs is validated using different public databases with a large degree of variability and without image exclusion.

Medical Image Synthesis via Deep Learning

Medical images have been widely used in clinics, providing visual representations of under-skin tissues in human body. By applying different imaging protocols, diverse modalities of medical images with unique characteristics of visualization can be produced. Considering the cost of scanning high-quality single modality images or homogeneous multiple modalities of images, medical image synthesis methods have been extensively explored for clinical applications. Among them, deep learning approaches, especially convolutional neural networks (CNNs) and generative adversarial networks (GANs), have rapidly become dominating for medical image synthesis in recent years. In this chapter, based on a general review of the medical image synthesis methods, we will focus on introducing typical CNNs and GANs models for medical image synthesis. Especially, we will elaborate our recent work about low-dose to high-dose PET image synthesis, and cross-modality MR image synthesis, using these models.

Geographic atrophy segmentation in SD-OCT images using synthesized fundus autofluorescence imaging

BACKGROUND AND OBJECTIVE

Accurate assessment of geographic atrophy (GA) is critical for diagnosis and therapy of non-exudative age-related macular degeneration (AMD). Herein, we propose a novel GA segmentation framework for spectral-domain optical coherence tomography (SD-OCT) images that employs synthesized fundus autofluorescence (FAF) images.

METHODS

An en-face OCT image is created via the restricted sub-volume projection of three-dimensional OCT data. A GA region-aware conditional generative adversarial network is employed to generate a plausible FAF image from the en-face OCT image. The network balances the consistency between the entire synthesize FAF image and the lesion. We use a fully convolutional deep network architecture to segment the GA region using the multimodal images, where the features of the en-face OCT and synthesized FAF images are fused on the front-end of the network.

RESULTS

Experimental results for 56 SD-OCT scans with GA indicate that our synthesis algorithm can generate high-quality synthesized FAF images and that the proposed segmentation network achieves a dice similarity coefficient, an overlap ratio, and an absolute area difference of 87.2%, 77.9%, and 11.0%, respectively.

CONCLUSION

We report an automatic GA segmentation method utilizing synthesized FAF images.

SIGNIFICANCE

Our method is effective for multimodal segmentation of the GA region and can improve AMD treatment.

Assessment of Deep Generative Models for High-Resolution Synthetic Retinal Image Generation of Age-Related Macular Degeneration

Importance

Deep learning (DL) used for discriminative tasks in ophthalmology, such as diagnosing diabetic retinopathy or age-related macular degeneration (AMD), requires large image data sets graded by human experts to train deep convolutional neural networks (DCNNs). In contrast, generative DL techniques could synthesize large new data sets of artificial retina images with different stages of AMD. Such images could enhance existing data sets of common and rare ophthalmic diseases without concern for personally identifying information to assist medical education of students, residents, and retinal specialists, as well as for training new DL diagnostic models for which extensive data sets from large clinical trials of expertly graded images may not exist.

Objective

To develop DL techniques for synthesizing high-resolution realistic fundus images serving as proxy data sets for use by retinal specialists and DL machines.

Design, Setting, and Participants

Generative adversarial networks were trained on 133 821 color fundus images from 4613 study participants from the Age-Related Eye Disease Study (AREDS), generating synthetic fundus images with and without AMD. We compared retinal specialists' ability to diagnose AMD on both real and synthetic images, asking them to assess image gradability and testing their ability to discern real from synthetic images. The performance of AMD diagnostic DCNNs (referable vs not referable AMD) trained on either all-real vs all-synthetic data sets was compared.

Main Outcomes and Measures

Accuracy of 2 retinal specialists (T.Y.A.L. and K.D.P.) for diagnosing and distinguishing AMD on real vs synthetic images and diagnostic performance (area under the curve) of DL algorithms trained on synthetic vs real images.

Results

The diagnostic accuracy of 2 retinal specialists on real vs synthetic images was similar. The accuracy of diagnosis as referable vs nonreferable AMD compared with certified human graders for retinal specialist 1 was 84.54% (error margin, 4.06%) on real images vs 84.12% (error margin, 4.16%) on synthetic images and for retinal specialist 2 was 89.47% (error margin, 3.45%) on real images vs 89.19% (error margin, 3.54%) on synthetic images. Retinal specialists could not distinguish real from synthetic images, with an accuracy of 59.50% (error margin, 3.93%) for retinal specialist 1 and 53.67% (error margin, 3.99%) for retinal specialist 2. The DCNNs trained on real data showed an area under the curve of 0.9706 (error margin, 0.0029), and those trained on synthetic data showed an area under the curve of 0.9235 (error margin, 0.0045).

Conclusions and Relevance

Deep learning-synthesized images appeared to be realistic to retinal specialists, and DCNNs achieved diagnostic performance on synthetic data close to that for real images, suggesting that DL generative techniques hold promise for training humans and machines.

Learning from adversarial medical images for X-ray breast mass segmentation

BACKGROUND AND OBJECTIVE

Simulation of diverse lesions in images is proposed and applied to overcome the scarcity of labeled data, which has hindered the application of deep learning in medical imaging. However, most of current studies focus on generating samples with class labels for classification and detection rather than segmentation, because generating images with precise masks remains a challenge. Therefore, we aim to generate realistic medical images with precise masks for improving lesion segmentation in mammagrams.

METHODS

In this paper, we propose a new framework for improving X-ray breast mass segmentation performance aided by generated adversarial lesion images with precise masks. Firstly, we introduce a conditional generative adversarial network (cGAN) to learn the distribution of real mass images as well as a mapping between images and corresponding segmentation masks. Subsequently, a number of lesion images are generated from various binary input masks using the generator in the trained cGAN. Then the generated adversarial samples are concatenated with original samples to produce a dataset with increased diversity. Furthermore, we introduce an improved U-net and train it on the previous augmented dataset for breast mass segmentation.

RESULTS

To demonstrate the effectiveness of our proposed method, we conduct experiments on publicly available mammogram database of INbreast and a private database provided by Nanfang Hospital in China. Experimental results show that an improvement up to 7% in Jaccard index can be achieved over the same model trained on original real lesion images.

CONCLUSIONS

Our proposed method can be viewed as one of the first steps toward generating realistic X-ray breast mass images with masks for precise segmentation.

Arterial Spin Labeling Images Synthesis From sMRI Using Unbalanced Deep Discriminant Learning

Adequate medical images are often indispensable in contemporary deep learning-based medical imaging studies, although the acquisition of certain image modalities may be limited due to several issues including high costs and patients issues. However, thanks to recent advances in deep learning techniques, the above tough problem can be substantially alleviated by medical images synthesis, by which various modalities including T1/T2/DTI MRI images, PET images, cardiac ultrasound images, retinal images, and so on, have already been synthesized. Unfortunately, the arterial spin labeling (ASL) image, which is an important fMRI indicator in dementia diseases diagnosis nowadays, has never been comprehensively investigated for the synthesis purpose yet. In this paper, ASL images have been successfully synthesized from structural magnetic resonance images for the first time. Technically, a novel unbalanced deep discriminant learning-based model equipped with new ResNet sub-structures is proposed to realize the synthesis of ASL images from structural magnetic resonance images. The extensive experiments have been conducted. Comprehensive statistical analyses reveal that: 1) this newly introduced model is capable to synthesize ASL images that are similar towards real ones acquired by actual scanning; 2) synthesized ASL images obtained by the new model have demonstrated outstanding performance when undergoing rigorous tests of region-based and voxel-based corrections of partial volume effects, which are essential in ASL images processing; and 3) it is also promising that the diagnosis performance of dementia diseases can be significantly improved with the help of synthesized ASL images obtained by the new model, based on a multi-modal MRI dataset containing 355 demented patients in this paper.

Image Synthesis in Multi-Contrast MRI With Conditional Generative Adversarial Networks

Acquiring images of the same anatomy with multiple different contrasts increases the diversity of diagnostic information available in an MR exam. Yet, the scan time limitations may prohibit the acquisition of certain contrasts, and some contrasts may be corrupted by noise and artifacts. In such cases, the ability to synthesize unacquired or corrupted contrasts can improve diagnostic utility. For multi-contrast synthesis, the current methods learn a nonlinear intensity transformation between the source and target images, either via nonlinear regression or deterministic neural networks. These methods can, in turn, suffer from the loss of structural details in synthesized images. Here, in this paper, we propose a new approach for multi-contrast MRI synthesis based on conditional generative adversarial networks. The proposed approach preserves intermediate-to-high frequency details via an adversarial loss, and it offers enhanced synthesis performance via pixel-wise and perceptual losses for registered multi-contrast images and a cycle-consistency loss for unregistered images. Information from neighboring cross-sections are utilized to further improve synthesis quality. Demonstrations on T₁- and T₂- weighted images from healthy subjects and patients clearly indicate the superior performance of the proposed approach compared to the previous state-of-the-art methods. Our synthesis approach can help improve the quality and versatility of the multi-contrast MRI exams without the need for prolonged or repeated examinations.

Retinal Image Synthesis and Semi-Supervised Learning for Glaucoma Assessment

Recent works show that generative adversarial networks (GANs) can be successfully applied to image synthesis and semi-supervised learning, where, given a small labeled database and a large unlabeled database, the goal is to train a powerful classifier. In this paper, we trained a retinal image synthesizer and a semi-supervised learning method for automatic glaucoma assessment using an adversarial model on a small glaucoma-labeled database and a large unlabeled database. Various studies have shown that glaucoma can be monitored by analyzing the optic disc and its surroundings, and for that reason, the images used in this paper were automatically cropped around the optic disc. The novelty of this paper is to propose a new retinal image synthesizer and a semi-supervised learning method for glaucoma assessment based on the deep convolutional GANs. In addition, and to the best of our knowledge, this system is trained on an unprecedented number of publicly available images (86926 images). This system, hence, is not only able to generate images synthetically but to provide labels automatically. Synthetic images were qualitatively evaluated using t-SNE plots of features associated with the images and their anatomical consistency was estimated by measuring the proportion of pixels corresponding to the anatomical structures around the optic disc. The resulting image synthesizer is able to generate realistic (cropped) retinal images, and subsequently, the glaucoma classifier is able to classify them into glaucomatous and normal with high accuracy (AUC = 0.9017). The obtained retinal image synthesizer and the glaucoma classifier could then be used to generate an unlimited number of cropped retinal images with glaucoma labels.

CycleGAN for style transfer in X-ray angiography

PURPOSE

We aim to perform generation of angiograms for various vascular structures as a mean of data augmentation in learning tasks. The task is to enhance the realism of vessels images generated from an anatomically realistic cardiorespiratory simulator to make them look like real angiographies.

METHODS

The enhancement is performed by applying the CycleGAN deep network for transferring the style of real angiograms acquired during percutaneous interventions into a data set composed of realistically simulated arteries.

RESULTS

The cycle consistency was evaluated by comparing an input simulated image with the one obtained after two cycles of image translation. An average structural similarity (SSIM) of 0.948 on our data sets has been obtained. The vessel preservation was measured by comparing segmentations of an input image and its corresponding enhanced image using Dice coefficient.

CONCLUSIONS

We proposed an application of the CycleGAN deep network for enhancing the artificial data as an alternative to classical data augmentation techniques for medical applications, particularly focused on angiogram generation. We discussed success and failure cases, explaining conditions for the realistic data augmentation which respects both the complex physiology of arteries and the various patterns and textures generated by X-ray angiography.

Polyp Segmentation using Generative Adversarial Network

Colorectal cancer is one of the highest causes of cancer-related death and the patient's survival rate depends on the stage at which polyps are detected. Polyp segmentation is a challenging research task due to variations in the size and shape of polyps leading to necessitate robust approaches for diagnosis. This paper studies the deep generative convolutional framework for the task of polyp segmentation. Here, the analysis of polyp segmentation has been explored with the pix2pix conditional generative adversarial network. On CVC- Clinic dataset, the proposed network achieves Jaccard index of 81.27% and Dice index of 88.48%.

Ea-GANs: Edge-Aware Generative Adversarial Networks for Cross-Modality MR Image Synthesis

Magnetic resonance (MR) imaging is a widely used medical imaging protocol that can be configured to provide different contrasts between the tissues in human body. By setting different scanning parameters, each MR imaging modality reflects the unique visual characteristic of scanned body part, benefiting the subsequent analysis from multiple perspectives. To utilize the complementary information from multiple imaging modalities, cross-modality MR image synthesis has aroused increasing research interest recently. However, most existing methods only focus on minimizing pixel/voxel-wise intensity difference but ignore the textural details of image content structure, which affects the quality of synthesized images. In this paper, we propose edge-aware generative adversarial networks (Ea-GANs) for cross-modality MR image synthesis. Specifically, we integrate edge information, which reflects the textural structure of image content and depicts the boundaries of different objects in images, to reduce this gap. Corresponding to different learning strategies, two frameworks are proposed, i.e., a generator-induced Ea-GAN (gEa-GAN) and a discriminator-induced Ea-GAN (dEa-GAN). The gEa-GAN incorporates the edge information via its generator, while the dEa-GAN further does this from both the generator and the discriminator so that the edge similarity is also adversarially learned. In addition, the proposed Ea-GANs are 3D-based and utilize hierarchical features to capture contextual information. The experimental results demonstrate that the proposed Ea-GANs, especially the dEa-GAN, outperform multiple state-of-the-art methods for cross-modality MR image synthesis in both qualitative and quantitative measures. Moreover, the dEa-GAN also shows excellent generality to generic image synthesis tasks on benchmark datasets about facades, maps, and cityscapes.

Synthetic Retinal Images from Unconditional GANs

Synthesized retinal images are highly demanded in the development of automated eye applications since they can make machine learning algorithms more robust by increasing the size and heterogeneity of the training database. Recently, conditional Generative Adversarial Networks (cGANs) based synthesizers have been shown to be promising for generating retinal images. However, cGANs based synthesizers require segmented blood vessels (BV) along with RGB retinal images during training. The amount of such data (i.e., retinal images and their corresponding BV) available in public databases is very small. Therefore, for training cGANs, an extra system is necessary either for synthesizing BV or for segmenting BV from retinal images. In this paper, we show that by using unconditional GANs (uGANs) we can generate synthesized retinal images without using BV images.

Retinal image synthesis from multiple-landmarks input with generative adversarial networks

BACKGROUND

Medical datasets, especially medical images, are often imbalanced due to the different incidences of various diseases. To address this problem, many methods have been proposed to synthesize medical images using generative adversarial networks (GANs) to enlarge training datasets for facilitating medical image analysis. For instance, conventional methods such as image-to-image translation techniques are used to synthesize fundus images with their respective vessel trees in the field of fundus image.

METHODS

In order to improve the image quality and details of the synthetic images, three key aspects of the pipeline are mainly elaborated: the input mask, architecture of GANs, and the resolution of paired images. We propose a new preprocessing pipeline named multiple-channels-multiple-landmarks (MCML), aiming to synthesize color fundus images from a combination of vessel tree, optic disc, and optic cup images. We compared both single vessel mask input and MCML mask input on two public fundus image datasets (DRIVE and DRISHTI-GS) with different kinds of Pix2pix and Cycle-GAN architectures. A new Pix2pix structure with ResU-net generator is also designed, which has been compared with the other models.

RESULTS AND CONCLUSION

As shown in the results, the proposed MCML method outperforms the single vessel-based methods for each architecture of GANs. Furthermore, we find that our Pix2pix model with ResU-net generator achieves superior PSNR and SSIM performance than the other GANs. High-resolution paired images are also beneficial for improving the performance of each GAN in this work. Finally, a Pix2pix network with ResU-net generator using MCML and high-resolution paired images are able to generate good and realistic fundus images in this work, indicating that our MCML method has great potential in the field of glaucoma computer-aided diagnosis based on fundus image.

Synthesizing Chest X-Ray Pathology for Training Deep Convolutional Neural Networks

Medical datasets are often highly imbalanced with over-representation of prevalent conditions and poor representation of rare medical conditions. Due to privacy concerns, it is challenging to aggregate large datasets between health care institutions. We propose synthesizing pathology in medical images as a means to overcome these challenges. We implement a deep convolutional generative adversarial network (DCGAN) to create synthesized chest X-rays based upon a modest sized labeled dataset. We used a combination of real and synthesized images to train deep convolutional neural networks (DCNNs) to detect pathology across five classes of chest X-rays. The comparative study of DCNNs trained with the combination of real and synthesized images showed that these networks can outperform similar networks trained solely with real images in pathology classification. This improved performance is largely attributable to the balancing of the dataset using DCGAN synthesized images, where classes that are lacking in example images are preferentially augmented.

Transfer Recurrent Feature Learning for Endomicroscopy Image Recognition

Probe-based confocal laser endomicroscopy (pCLE) is an emerging tool for epithelial cancer diagnosis, which enables in-vivo microscopic imaging during endoscopic procedures and facilitates the development of automatic recognition algorithms to identify the status of tissues. In this paper, we propose a transfer recurrent feature learning framework for classification tasks on pCLE videos. At the first stage, the discriminative feature of single pCLE frame is learned via generative adversarial networks based on both pCLE and histology modalities. At the second stage, we use recurrent neural networks to handle the varying length and irregular shape of pCLE mosaics taking the frame-based features as input. The experiments on real pCLE data sets demonstrate that our approach outperforms, with statistical significance, state-of-the-art approaches. A binary classification accuracy of 84.1% has been achieved.