Unsupervised Reverse Domain Adaptation for Synthetic Medical Images via Adversarial Training

Simulation-to-real domain adaptation with teacher-student learning for endoscopic instrument segmentation

PURPOSE

Segmentation of surgical instruments in endoscopic video streams is essential for automated surgical scene understanding and process modeling. However, relying on fully supervised deep learning for this task is challenging because manual annotation occupies valuable time of the clinical experts.

METHODS

We introduce a teacher-student learning approach that learns jointly from annotated simulation data and unlabeled real data to tackle the challenges in simulation-to-real unsupervised domain adaptation for endoscopic image segmentation.

RESULTS

Empirical results on three datasets highlight the effectiveness of the proposed framework over current approaches for the endoscopic instrument segmentation task. Additionally, we provide analysis of major factors affecting the performance on all datasets to highlight the strengths and failure modes of our approach.

CONCLUSIONS

We show that our proposed approach can successfully exploit the unlabeled real endoscopic video frames and improve generalization performance over pure simulation-based training and the previous state-of-the-art. This takes us one step closer to effective segmentation of surgical instrument in the annotation scarce setting.

Disentangle domain features for cross-modality cardiac image segmentation

Unsupervised domain adaptation (UDA) generally learns a mapping to align the distribution of the source domain and target domain. The learned mapping can boost the performance of the model on the target data, of which the labels are unavailable for model training. Previous UDA methods mainly focus on domain-invariant features (DIFs) without considering the domain-specific features (DSFs), which could be used as complementary information to constrain the model. In this work, we propose a new UDA framework for cross-modality image segmentation. The framework first disentangles each domain into the DIFs and DSFs. To enhance the representation of DIFs, self-attention modules are used in the encoder which allows attention-driven, long-range dependency modeling for image generation tasks. Furthermore, a zero loss is minimized to enforce the information of target (source) DSFs, contained in the source (target) images, to be as close to zero as possible. These features are then iteratively decoded and encoded twice to maintain the consistency of the anatomical structure. To improve the quality of the generated images and segmentation results, several discriminators are introduced for adversarial learning. Finally, with the source data and their DIFs, we train a segmentation network, which can be applicable to target images. We validated the proposed framework for cross-modality cardiac segmentation using two public datasets, and the results showed our method delivered promising performance and compared favorably to state-of-the-art approaches in terms of segmentation accuracies. The source code of this work will be released via https://zmiclab.github.io/projects.html, once this manuscript is accepted for publication.

EndoSLAM dataset and an unsupervised monocular visual odometry and depth estimation approach for endoscopic videos

Deep learning techniques hold promise to develop dense topography reconstruction and pose estimation methods for endoscopic videos. However, currently available datasets do not support effective quantitative benchmarking. In this paper, we introduce a comprehensive endoscopic SLAM dataset consisting of 3D point cloud data for six porcine organs, capsule and standard endoscopy recordings, synthetically generated data as well as clinically in use conventional endoscope recording of the phantom colon with computed tomography(CT) scan ground truth. A Panda robotic arm, two commercially available capsule endoscopes, three conventional endoscopes with different camera properties, two high precision 3D scanners, and a CT scanner were employed to collect data from eight ex-vivo porcine gastrointestinal (GI)-tract organs and a silicone colon phantom model. In total, 35 sub-datasets are provided with 6D pose ground truth for the ex-vivo part: 18 sub-datasets for colon, 12 sub-datasets for stomach, and 5 sub-datasets for small intestine, while four of these contain polyp-mimicking elevations carried out by an expert gastroenterologist. To verify the applicability of this data for use with real clinical systems, we recorded a video sequence with a state-of-the-art colonoscope from a full representation silicon colon phantom. Synthetic capsule endoscopy frames from stomach, colon, and small intestine with both depth and pose annotations are included to facilitate the study of simulation-to-real transfer learning algorithms. Additionally, we propound Endo-SfMLearner, an unsupervised monocular depth and pose estimation method that combines residual networks with a spatial attention module in order to dictate the network to focus on distinguishable and highly textured tissue regions. The proposed approach makes use of a brightness-aware photometric loss to improve the robustness under fast frame-to-frame illumination changes that are commonly seen in endoscopic videos. To exemplify the use-case of the EndoSLAM dataset, the performance of Endo-SfMLearner is extensively compared with the state-of-the-art: SC-SfMLearner, Monodepth2, and SfMLearner. The codes and the link for the dataset are publicly available at https://github.com/CapsuleEndoscope/EndoSLAM. A video demonstrating the experimental setup and procedure is accessible as Supplementary Video 1.

Unsupervised Monocular Depth Estimation for Colonoscope System Using Feedback Network

A colonoscopy is a medical examination used to check disease or abnormalities in the large intestine. If necessary, polyps or adenomas would be removed through the scope during a colonoscopy. Colorectal cancer can be prevented through this. However, the polyp detection rate differs depending on the condition and skill level of the endoscopist. Even some endoscopists have a 90% chance of missing an adenoma. Artificial intelligence and robot technologies for colonoscopy are being studied to compensate for these problems. In this study, we propose a self-supervised monocular depth estimation using spatiotemporal consistency in the colon environment. It is our contribution to propose a loss function for reconstruction errors between adjacent predicted depths and a depth feedback network that uses predicted depth information of the previous frame to predict the depth of the next frame. We performed quantitative and qualitative evaluation of our approach, and the proposed FBNet (depth FeedBack Network) outperformed state-of-the-art results for unsupervised depth estimation on the UCL datasets.

Application and Construction of Deep Learning Networks in Medical Imaging

Deep learning (DL) approaches are part of the machine learning (ML) subfield concerned with the development of computational models to train artificial intelligence systems. DL models are characterized by automatically extracting high-level features from the input data to learn the relationship between matching datasets. Thus, its implementation offers an advantage over common ML methods that often require the practitioner to have some domain knowledge of the input data to select the best latent representation. As a result of this advantage, DL has been successfully applied within the medical imaging field to address problems, such as disease classification and tumor segmentation for which it is difficult or impossible to determine which image features are relevant. Therefore, taking into consideration the positive impact of DL on the medical imaging field, this article reviews the key concepts associated with its evolution and implementation. The sections of this review summarize the milestones related to the development of the DL field, followed by a description of the elements of deep neural network and an overview of its application within the medical imaging field. Subsequently, the key steps necessary to implement a supervised DL application are defined, and associated limitations are discussed.

Stomach 3D Reconstruction Using Virtual Chromoendoscopic Images

Gastric endoscopy is a golden standard in the clinical process that enables medical practitioners to diagnose various lesions inside a patient’s stomach. If a lesion is found, a success in identifying the location of the found lesion relative to the global view of the stomach will lead to better decision making for the next clinical treatment. Our previous research showed that the lesion localization could be achieved by reconstructing the whole stomach shape from chromoendoscopic indigo carmine (IC) dye-sprayed images using a structure-from-motion (SfM) pipeline. However, spraying the IC dye to the whole stomach requires additional time, which is not desirable for both patients and practitioners. Our objective is to propose an alternative way to achieve whole stomach 3D reconstruction without the need of the IC dye. We generate virtual IC-sprayed (VIC) images based on image-to-image style translation trained on unpaired real no-IC and IC-sprayed images, where we have investigated the effect of input and output color channel selection for generating the VIC images. We validate our reconstruction results by comparing them with the results using real IC-sprayed images and confirm that the obtained stomach 3D structures are comparable to each other. We also propose a local reconstruction technique to obtain a more detailed surface and texture around an interesting region. The proposed method achieves the whole stomach reconstruction without the need of real IC dye using SfM. We have found that translating no-IC green-channel images to IC-sprayed red-channel images gives the best SfM reconstruction result. Clinical impact We offer a method of the frame localization and local 3D reconstruction of a found gastric lesion using standard endoscopy images, leading to better clinical decision.

Deep virtual adversarial self-training with consistency regularization for semi-supervised medical image classification

Convolutional neural networks have achieved prominent success on a variety of medical imaging tasks when a large amount of labeled training data is available. However, the acquisition of expert annotations for medical data is usually expensive and time-consuming, which poses a great challenge for supervised learning approaches. In this work, we proposed a novel semi-supervised deep learning method, i.e., deep virtual adversarial self-training with consistency regularization, for large-scale medical image classification. To effectively exploit useful information from unlabeled data, we leverage self-training and consistency regularization to harness the underlying knowledge, which helps improve the discrimination capability of training models. More concretely, the model first uses its prediction for pseudo-labeling on the weakly-augmented input image. A pseudo-label is kept only if the corresponding class probability is of high confidence. Then the model prediction is encouraged to be consistent with the strongly-augmented version of the same input image. To improve the robustness of the network against virtual adversarial perturbed input, we incorporate virtual adversarial training (VAT) on both labeled and unlabeled data into the course of training. Hence, the network is trained by minimizing a combination of three types of losses, including a standard supervised loss on labeled data, a consistency regularization loss on unlabeled data, and a VAT loss on both labeled and labeled data. We extensively evaluate the proposed semi-supervised deep learning methods on two challenging medical image classification tasks: breast cancer screening from ultrasound images and multi-class ophthalmic disease classification from optical coherence tomography B-scan images. Experimental results demonstrate that the proposed method outperforms both supervised baseline and other state-of-the-art methods by a large margin on all tasks.

DeepCDA: deep cross-domain compound-protein affinity prediction through LSTM and convolutional neural networks

MOTIVATION

An essential part of drug discovery is the accurate prediction of the binding affinity of new compound-protein pairs. Most of the standard computational methods assume that compounds or proteins of the test data are observed during the training phase. However, in real-world situations, the test and training data are sampled from different domains with different distributions. To cope with this challenge, we propose a deep learning-based approach that consists of three steps. In the first step, the training encoder network learns a novel representation of compounds and proteins. To this end, we combine convolutional layers and long-short-term memory layers so that the occurrence patterns of local substructures through a protein and a compound sequence are learned. Also, to encode the interaction strength of the protein and compound substructures, we propose a two-sided attention mechanism. In the second phase, to deal with the different distributions of the training and test domains, a feature encoder network is learned for the test domain by utilizing an adversarial domain adaptation approach. In the third phase, the learned test encoder network is applied to new compound-protein pairs to predict their binding affinity.

RESULTS

To evaluate the proposed approach, we applied it to KIBA, Davis and BindingDB datasets. The results show that the proposed method learns a more reliable model for the test domain in more challenging situations.

AVAILABILITY AND IMPLEMENTATION

https://github.com/LBBSoft/DeepCDA.

VR-Caps: A Virtual Environment for Capsule Endoscopy

Current capsule endoscopes and next-generation robotic capsules for diagnosis and treatment of gastrointestinal diseases are complex cyber-physical platforms that must orchestrate complex software and hardware functions. The desired tasks for these systems include visual localization, depth estimation, 3D mapping, disease detection and segmentation, automated navigation, active control, path realization and optional therapeutic modules such as targeted drug delivery and biopsy sampling. Data-driven algorithms promise to enable many advanced functionalities for capsule endoscopes, but real-world data is challenging to obtain. Physically-realistic simulations providing synthetic data have emerged as a solution to the development of data-driven algorithms. In this work, we present a comprehensive simulation platform for capsule endoscopy operations and introduce VR-Caps, a virtual active capsule environment that simulates a range of normal and abnormal tissue conditions (e.g., inflated, dry, wet etc.) and varied organ types, capsule endoscope designs (e.g., mono, stereo, dual and 360^∘ camera), and the type, number, strength, and placement of internal and external magnetic sources that enable active locomotion. VR-Caps makes it possible to both independently or jointly develop, optimize, and test medical imaging and analysis software for the current and next-generation endoscopic capsule systems. To validate this approach, we train state-of-the-art deep neural networks to accomplish various medical image analysis tasks using simulated data from VR-Caps and evaluate the performance of these models on real medical data. Results demonstrate the usefulness and effectiveness of the proposed virtual platform in developing algorithms that quantify fractional coverage, camera trajectory, 3D map reconstruction, and disease classification. All of the code, pre-trained weights and created 3D organ models of the virtual environment with detailed instructions how to setup and use the environment are made publicly available at https://github.com/CapsuleEndoscope/VirtualCapsuleEndoscopy and a video demonstration can be seen in the supplementary videos (Video-I).

Transformation-Consistent Self-Ensembling Model for Semisupervised Medical Image Segmentation

A common shortfall of supervised deep learning for medical imaging is the lack of labeled data, which is often expensive and time consuming to collect. This article presents a new semisupervised method for medical image segmentation, where the network is optimized by a weighted combination of a common supervised loss only for the labeled inputs and a regularization loss for both the labeled and unlabeled data. To utilize the unlabeled data, our method encourages consistent predictions of the network-in-training for the same input under different perturbations. With the semisupervised segmentation tasks, we introduce a transformation-consistent strategy in the self-ensembling model to enhance the regularization effect for pixel-level predictions. To further improve the regularization effects, we extend the transformation in a more generalized form including scaling and optimize the consistency loss with a teacher model, which is an averaging of the student model weights. We extensively validated the proposed semisupervised method on three typical yet challenging medical image segmentation tasks: 1) skin lesion segmentation from dermoscopy images in the International Skin Imaging Collaboration (ISIC) 2017 data set; 2) optic disk (OD) segmentation from fundus images in the Retinal Fundus Glaucoma Challenge (REFUGE) data set; and 3) liver segmentation from volumetric CT scans in the Liver Tumor Segmentation Challenge (LiTS) data set. Compared with state-of-the-art, our method shows superior performance on the challenging 2-D/3-D medical images, demonstrating the effectiveness of our semisupervised method for medical image segmentation.

Secure and Robust Machine Learning for Healthcare: A Survey

Recent years have witnessed widespread adoption of machine learning (ML)/deep learning (DL) techniques due to their superior performance for a variety of healthcare applications ranging from the prediction of cardiac arrest from one-dimensional heart signals to computer-aided diagnosis (CADx) using multi-dimensional medical images. Notwithstanding the impressive performance of ML/DL, there are still lingering doubts regarding the robustness of ML/DL in healthcare settings (which is traditionally considered quite challenging due to the myriad security and privacy issues involved), especially in light of recent results that have shown that ML/DL are vulnerable to adversarial attacks. In this paper, we present an overview of various application areas in healthcare that leverage such techniques from security and privacy point of view and present associated challenges. In addition, we present potential methods to ensure secure and privacy-preserving ML for healthcare applications. Finally, we provide insight into the current research challenges and promising directions for future research.

Augmented Reality Guided Laparoscopic Surgery of the Uterus

A major research area in Computer Assisted Intervention (CAI) is to aid laparoscopic surgery teams with Augmented Reality (AR) guidance. This involves registering data from other modalities such as MR and fusing it with the laparoscopic video in real-time, to reveal the location of hidden critical structures. We present the first system for AR guided laparoscopic surgery of the uterus. This works with pre-operative MR or CT data and monocular laparoscopes, without requiring any additional interventional hardware such as optical trackers. We present novel and robust solutions to two main sub-problems: the initial registration, which is solved using a short exploratory video, and update registration, which is solved with real-time tracking-by-detection. These problems are challenging for the uterus because it is a weakly-textured, highly mobile organ that moves independently of surrounding structures. In the broader context, our system is the first that has successfully performed markerless real-time registration and AR of a mobile human organ with monocular laparoscopes in the OR.

Automatic Segmentation and Visualization of Choroid in OCT with Knowledge Infused Deep Learning

The choroid provides oxygen and nourishment to the outer retina thus is related to the pathology of various ocular diseases. Optical coherence tomography (OCT) is advantageous in visualizing and quantifying the choroid in vivo. However, its application in the study of the choroid is still limited for two reasons. (1) The lower boundary of the choroid (choroid-sclera interface) in OCT is fuzzy, which makes the automatic segmentation difficult and inaccurate. (2) The visualization of the choroid is hindered by the vessel shadows from the superficial layers of the inner retina. In this paper, we propose to incorporate medical and imaging prior knowledge with deep learning to address these two problems. We propose a biomarker-infused global-to-local network (Bio-Net) for the choroid segmentation, which not only regularizes the segmentation via predicted choroid thickness, but also leverages a global-to-local segmentation strategy to provide global structure information and suppress overfitting. For eliminating the retinal vessel shadows, we propose a deep-learning pipeline, which firstly locate the shadows using their projection on the retinal pigment epithelium layer, then the contents of the choroidal vasculature at the shadow locations are predicted with an edge-to-texture generative adversarial inpainting network. The results show our method outperforms the existing methods on both tasks. We further apply the proposed method in a clinical prospective study for understanding the pathology of glaucoma, which demonstrates its capacity in detecting the structure and vascular changes of the choroid related to the elevation of intra-ocular pressure.

Target-Independent Domain Adaptation for WBC Classification Using Generative Latent Search

Automating the classification of camera-obtained microscopic images of White Blood Cells (WBCs) and related cell subtypes has assumed importance since it aids the laborious manual process of review and diagnosis. Several State-Of-The-Art (SOTA) methods developed using Deep Convolutional Neural Networks suffer from the problem of domain shift - severe performance degradation when they are tested on data (target) obtained in a setting different from that of the training (source). The change in the target data might be caused by factors such as differences in camera/microscope types, lenses, lighting-conditions etc. This problem can potentially be solved using Unsupervised Domain Adaptation (UDA) techniques albeit standard algorithms presuppose the existence of a sufficient amount of unlabelled target data which is not always the case with medical images. In this paper, we propose a method for UDA that is devoid of the need for target data. Given a test image from the target data, we obtain its 'closest-clone' from the source data that is used as a proxy in the classifier. We prove the existence of such a clone given that infinite number of data points can be sampled from the source distribution. We propose a method in which a latent-variable generative model based on variational inference is used to simultaneously sample and find the 'closest-clone' from the source distribution through an optimization procedure in the latent space. We demonstrate the efficacy of the proposed method over several SOTA UDA methods for WBC classification on datasets captured using different imaging modalities under multiple settings.

EndoL2H: Deep Super-Resolution for Capsule Endoscopy

Although wireless capsule endoscopy is the preferred modality for diagnosis and assessment of small bowel diseases, the poor camera resolution is a substantial limitation for both subjective and automated diagnostics. Enhanced-resolution endoscopy has shown to improve adenoma detection rate for conventional endoscopy and is likely to do the same for capsule endoscopy. In this work, we propose and quantitatively validate a novel framework to learn a mapping from low-to-high-resolution endoscopic images. We combine conditional adversarial networks with a spatial attention block to improve the resolution by up to factors of 8× , 10× , 12× , respectively. Quantitative and qualitative studies demonstrate the superiority of EndoL2H over state-of-the-art deep super-resolution methods Deep Back-Projection Networks (DBPN), Deep Residual Channel Attention Networks (RCAN) and Super Resolution Generative Adversarial Network (SRGAN). Mean Opinion Score (MOS) tests were performed by 30 gastroenterologists qualitatively assess and confirm the clinical relevance of the approach. EndoL2H is generally applicable to any endoscopic capsule system and has the potential to improve diagnosis and better harness computational approaches for polyp detection and characterization. Our code and trained models are available at https://github.com/CapsuleEndoscope/EndoL2H.

Application of Artificial Intelligence Technology in Oncology: Towards the Establishment of Precision Medicine

In recent years, advances in artificial intelligence (AI) technology have led to the rapid clinical implementation of devices with AI technology in the medical field. More than 60 AI-equipped medical devices have already been approved by the Food and Drug Administration (FDA) in the United States, and the active introduction of AI technology is considered to be an inevitable trend in the future of medicine. In the field of oncology, clinical applications of medical devices using AI technology are already underway, mainly in radiology, and AI technology is expected to be positioned as an important core technology. In particular, "precision medicine," a medical treatment that selects the most appropriate treatment for each patient based on a vast amount of medical data such as genome information, has become a worldwide trend; AI technology is expected to be utilized in the process of extracting truly useful information from a large amount of medical data and applying it to diagnosis and treatment. In this review, we would like to introduce the history of AI technology and the current state of medical AI, especially in the oncology field, as well as discuss the possibilities and challenges of AI technology in the medical field.

Detecting Deficient Coverage in Colonoscopies

Colonoscopy is tool of choice for preventing Colorectal Cancer, by detecting and removing polyps before they become cancerous. However, colonoscopy is hampered by the fact that endoscopists routinely miss 22-28% of polyps. While some of these missed polyps appear in the endoscopist's field of view, others are missed simply because of substandard coverage of the procedure, i.e. not all of the colon is seen. This paper attempts to rectify the problem of substandard coverage in colonoscopy through the introduction of the C2D2 (Colonoscopy Coverage Deficiency via Depth) algorithm which detects deficient coverage, and can thereby alert the endoscopist to revisit a given area. More specifically, C2D2 consists of two separate algorithms: the first performs depth estimation of the colon given an ordinary RGB video stream; while the second computes coverage given these depth estimates. Rather than compute coverage for the entire colon, our algorithm computes coverage locally, on a segment-by-segment basis; C2D2 can then indicate in real-time whether a particular area of the colon has suffered from deficient coverage, and if so the endoscopist can return to that area. Our coverage algorithm is the first such algorithm to be evaluated in a large-scale way; while our depth estimation technique is the first calibration-free unsupervised method applied to colonoscopies. The C2D2 algorithm achieves state of the art results in the detection of deficient coverage. On synthetic sequences with ground truth, it is 2.4 times more accurate than human experts; while on real sequences, C2D2 achieves a 93.0% agreement with experts.

Deep Adversarial Training for Multi-Organ Nuclei Segmentation in Histopathology Images

Nuclei mymargin segmentation is a fundamental task for various computational pathology applications including nuclei morphology analysis, cell type classification, and cancer grading. Deep learning has emerged as a powerful approach to segmenting nuclei but the accuracy of convolutional neural networks (CNNs) depends on the volume and the quality of labeled histopathology data for training. In particular, conventional CNN-based approaches lack structured prediction capabilities, which are required to distinguish overlapping and clumped nuclei. Here, we present an approach to nuclei segmentation that overcomes these challenges by utilizing a conditional generative adversarial network (cGAN) trained with synthetic and real data. We generate a large dataset of H&E training images with perfect nuclei segmentation labels using an unpaired GAN framework. This synthetic data along with real histopathology data from six different organs are used to train a conditional GAN with spectral normalization and gradient penalty for nuclei segmentation. This adversarial regression framework enforces higher-order spacial-consistency when compared to conventional CNN models. We demonstrate that this nuclei segmentation approach generalizes across different organs, sites, patients and disease states, and outperforms conventional approaches, especially in isolating individual and overlapping nuclei.

Effective Immunohistochemistry Pathology Microscopy Image Generation Using CycleGAN

Immunohistochemistry detection technology is able to detect more difficult tumors than regular pathology detection technology only with hematoxylin-eosin stained pathology microscopy images, - for example, neuroendocrine tumor detection. However, making immunohistochemistry pathology microscopy images costs much time and money. In this paper, we propose an effective immunohistochemistry pathology microscopic image-generation method that can generate synthetic immunohistochemistry pathology microscopic images from hematoxylin-eosin stained pathology microscopy images without any annotation. CycleGAN is adopted as the basic architecture for the unpaired and unannotated dataset. Moreover, multiple instances learning algorithms and the idea behind conditional GAN are considered to improve performance. To our knowledge, this is the first attempt to generate immunohistochemistry pathology microscopic images, and our method can achieve good performance, which will be very useful for pathologists and patients when applied in clinical practice.

Domain gap in adapting self-supervised depth estimation methods for stereo-endoscopy

In endoscopy, depth estimation is a task that potentially helps in quantifying visual information for better scene understanding. A plethora of depth estimation algorithms have been proposed in the computer vision community. The endoscopic domain however, differs from the typical depth estimation scenario due to differences in the setup and nature of the scene. Furthermore, it is unfeasible to obtain ground truth depth information owing to an unsuitable detection range of off-the-shelf depth sensors and difficulties in setting up a depth-sensor in a surgical environment. In this paper, an existing self-supervised approach, called Monodepth [1], from the field of autonomous driving is applied to a novel dataset of stereo-endoscopic images from reconstructive mitral valve surgery. While it is already known that endoscopic scenes are more challenging than outdoor driving scenes, the paper performs experiments to quantify the comparison, and describe the domain gap and challenges involved in the transfer of these methods.

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.

A Survey of Unsupervised Deep Domain Adaptation

Deep learning has produced state-of-the-art results for a variety of tasks. While such approaches for supervised learning have performed well, they assume that training and testing data are drawn from the same distribution, which may not always be the case. As a complement to this challenge, single-source unsupervised domain adaptation can handle situations where a network is trained on labeled data from a source domain and unlabeled data from a related but different target domain with the goal of performing well at test-time on the target domain. Many single-source and typically homogeneous unsupervised deep domain adaptation approaches have thus been developed, combining the powerful, hierarchical representations from deep learning with domain adaptation to reduce reliance on potentially-costly target data labels. This survey will compare these approaches by examining alternative methods, the unique and common elements, results, and theoretical insights. We follow this with a look at application areas and open research directions.

Advancing Medical Imaging Informatics by Deep Learning-Based Domain Adaptation

INTRODUCTION

There has been a rapid development of deep learning (DL) models for medical imaging. However, DL requires a large labeled dataset for training the models. Getting large-scale labeled data remains a challenge, and multi-center datasets suffer from heterogeneity due to patient diversity and varying imaging protocols. Domain adaptation (DA) has been developed to transfer the knowledge from a labeled data domain to a related but unlabeled domain in either image space or feature space. DA is a type of transfer learning (TL) that can improve the performance of models when applied to multiple different datasets.

OBJECTIVE

In this survey, we review the state-of-the-art DL-based DA methods for medical imaging. We aim to summarize recent advances, highlighting the motivation, challenges, and opportunities, and to discuss promising directions for future work in DA for medical imaging.

METHODS

We surveyed peer-reviewed publications from leading biomedical journals and conferences between 2017-2020, that reported the use of DA in medical imaging applications, grouping them by methodology, image modality, and learning scenarios.

RESULTS

We mainly focused on pathology and radiology as application areas. Among various DA approaches, we discussed domain transformation (DT) and latent feature-space transformation (LFST). We highlighted the role of unsupervised DA in image segmentation and described opportunities for future development.

CONCLUSION

DA has emerged as a promising solution to deal with the lack of annotated training data. Using adversarial techniques, unsupervised DA has achieved good performance, especially for segmentation tasks. Opportunities include domain transferability, multi-modal DA, and applications that benefit from synthetic data.

Uncertainty-aware domain alignment for anatomical structure segmentation

Automatic and accurate segmentation of anatomical structures on medical images is crucial for detecting various potential diseases. However, the segmentation performance of established deep neural networks may degenerate on different modalities or devices owing to the significant difference across the domains, a problem known as domain shift. In this work, we propose an uncertainty-aware domain alignment framework to address the domain shift problem in the cross-domain Unsupervised Domain Adaptation (UDA) task. Specifically, we design an Uncertainty Estimation and Segmentation Module (UESM) to obtain the uncertainty map estimation. Then, a novel Uncertainty-aware Cross Entropy (UCE) loss is proposed to leverage the uncertainty information to boost the segmentation performance on highly uncertain regions. To further improve the performance in the UDA task, an Uncertainty-aware Self-Training (UST) strategy is developed to choose the optimal target samples by uncertainty guidance. In addition, the Uncertainty Feature Recalibration Module (UFRM) is applied to enforce the framework to minimize the cross-domain discrepancy. The proposed framework is evaluated on a private cross-device Optical Coherence Tomography (OCT) dataset and a public cross-modality cardiac dataset released by MMWHS 2017. Extensive experiments indicate that the proposed UESM is both efficient and effective for the uncertainty estimation in the UDA task, achieving state-of-the-art performance on both cross-modality and cross-device datasets.

Augmenting Colonoscopy using Extended and Directional CycleGAN for Lossy Image Translation

Colorectal cancer screening modalities, such as optical colonoscopy (OC) and virtual colonoscopy (VC), are critical for diagnosing and ultimately removing polyps (precursors of colon cancer). The non-invasive VC is normally used to inspect a 3D reconstructed colon (from CT scans) for polyps and if found, the OC procedure is performed to physically traverse the colon via endoscope and remove these polyps. In this paper, we present a deep learning framework, Extended and Directional CycleGAN, for lossy unpaired image-to-image translation between OC and VC to augment OC video sequences with scale-consistent depth information from VC, and augment VC with patient-specific textures, color and specular highlights from OC (e.g, for realistic polyp synthesis). Both OC and VC contain structural information, but it is obscured in OC by additional patient-specific texture and specular highlights, hence making the translation from OC to VC lossy. The existing CycleGAN approaches do not handle lossy transformations. To address this shortcoming, we introduce an extended cycle consistency loss, which compares the geometric structures from OC in the VC domain. This loss removes the need for the CycleGAN to embed OC information in the VC domain. To handle a stronger removal of the textures and lighting, a Directional Discriminator is introduced to differentiate the direction of translation (by creating paired information for the discriminator), as opposed to the standard CycleGAN which is direction-agnostic. Combining the extended cycle consistency loss and the Directional Discriminator, we show state-of-the-art results on scale-consistent depth inference for phantom, textured VC and for real polyp and normal colon video sequences. We also present results for realistic pendunculated and flat polyp synthesis from bumps introduced in 3D VC models.

Adversarial convolutional network for esophageal tissue segmentation on OCT images

Automatic segmentation is important for esophageal OCT image processing, which is able to provide tissue characteristics such as shape and thickness for disease diagnosis. Existing automatical segmentation methods based on deep convolutional networks may not generate accurate segmentation results due to limited training set and various layer shapes. This study proposed a novel adversarial convolutional network (ACN) to segment esophageal OCT images using a convolutional network trained by adversarial learning. The proposed framework includes a generator and a discriminator, both with U-Net alike fully convolutional architecture. The discriminator is a hybrid network that discriminates whether the generated results are real and implements pixel classification at the same time. Leveraging on the adversarial training, the discriminator becomes more powerful. In addition, the adversarial loss is able to encode high order relationships of pixels, thus eliminating the requirements of post-processing. Experiments on segmenting esophageal OCT images from guinea pigs confirmed that the ACN outperforms several deep learning frameworks in pixel classification accuracy and improves the segmentation result. The potential clinical application of ACN for detecting eosinophilic esophagitis (EoE), an esophageal disease, is also presented in the experiment.

Adversarial convolutional network for esophageal tissue segmentation on OCT images

Automatic segmentation is important for esophageal OCT image processing, which is able to provide tissue characteristics such as shape and thickness for disease diagnosis. Existing automatical segmentation methods based on deep convolutional networks may not generate accurate segmentation results due to limited training set and various layer shapes. This study proposed a novel adversarial convolutional network (ACN) to segment esophageal OCT images using a convolutional network trained by adversarial learning. The proposed framework includes a generator and a discriminator, both with U-Net alike fully convolutional architecture. The discriminator is a hybrid network that discriminates whether the generated results are real and implements pixel classification at the same time. Leveraging on the adversarial training, the discriminator becomes more powerful. In addition, the adversarial loss is able to encode high order relationships of pixels, thus eliminating the requirements of post-processing. Experiments on segmenting esophageal OCT images from guinea pigs confirmed that the ACN outperforms several deep learning frameworks in pixel classification accuracy and improves the segmentation result. The potential clinical application of ACN for detecting eosinophilic esophagitis (EoE), an esophageal disease, is also presented in the experiment.

GANPOP: Generative Adversarial Network Prediction of Optical Properties From Single Snapshot Wide-Field Images

We present a deep learning framework for wide-field, content-aware estimation of absorption and scattering coefficients of tissues, called Generative Adversarial Network Prediction of Optical Properties (GANPOP). Spatial frequency domain imaging is used to obtain ground-truth optical properties at 660 nm from in vivo human hands and feet, freshly resected human esophagectomy samples, and homogeneous tissue phantoms. Images of objects with either flat-field or structured illumination are paired with registered optical property maps and are used to train conditional generative adversarial networks that estimate optical properties from a single input image. We benchmark this approach by comparing GANPOP to a single-snapshot optical property (SSOP) technique, using a normalized mean absolute error (NMAE) metric. In human gastrointestinal specimens, GANPOP with a single structured-light input image estimates the reduced scattering and absorption coefficients with 60% higher accuracy than SSOP while GANPOP with a single flat-field illumination image achieves similar accuracy to SSOP. When applied to both in vivo and ex vivo swine tissues, a GANPOP model trained solely on structured-illumination images of human specimens and phantoms estimates optical properties with approximately 46% improvement over SSOP, indicating adaptability to new, unseen tissue types. Given a training set that appropriately spans the target domain, GANPOP has the potential to enable rapid and accurate wide-field measurements of optical properties.

Generalization of intensity distribution of medical images using GANs

The performance of a CNN based medical-image classification network depends on the intensities of the trained images. Therefore, it is necessary to generalize medical images of various intensities against degradation of performance. For lesion classification, features of generalized images should be carefully maintained. To maintain the performance of the medical image classification network and minimize the loss of features, we propose a method using a generative adversarial network (GAN) as a generator to adapt the arbitrary intensity distribution to the specific intensity distribution of the training set. We also select CycleGAN and UNIT to train unpaired medical image data sets. The following was done to evaluate each method’s performance: the similarities between the generalized image and the original were measured via the structural similarity index (SSIM) and histogram, and the original domain data set was passed to a classifier that trained only the original domain images for accuracy comparisons. The results show that the performance evaluation of the generalized images is better than that of the originals, confirming that our proposed method is a simple but powerful solution to the performance degradation of a classification network.

Coupled Real-Synthetic Domain Adaptation for Real-World Deep Depth Enhancement

Advances in depth sensing technologies have allowed simultaneous acquisition of both color and depth data under different environments. However, most depth sensors have lower resolution than that of the associated color channels and such a mismatch can affect applications that require accurate depth recovery. Existing depth enhancement methods use simplistic noise models and cannot generalize well under real-world conditions. In this paper, a coupled real-synthetic domain adaptation method is proposed, which enables domain transfer between high-quality depth simulators and real depth camera information for super-resolution depth recovery. The method first enables the realistic degradation from synthetic images, and then enhances degraded depth data to high quality with a color-guided sub-network. The key advantage of the work is that it generalizes well to real-world datasets without further training or fine-tuning. Detailed quantitative and qualitative results are presented, and it is demonstrated that the proposed method achieves improved performance compared to previous methods fine-tuned on the specific datasets.

Boundary-Weighted Domain Adaptive Neural Network for Prostate MR Image Segmentation

Accurate segmentation of the prostate from magnetic resonance (MR) images provides useful information for prostate cancer diagnosis and treatment. However, automated prostate segmentation from 3D MR images faces several challenges. The lack of clear edge between the prostate and other anatomical structures makes it challenging to accurately extract the boundaries. The complex background texture and large variation in size, shape and intensity distribution of the prostate itself make segmentation even further complicated. Recently, as deep learning, especially convolutional neural networks (CNNs), emerging as the best performed methods for medical image segmentation, the difficulty in obtaining large number of annotated medical images for training CNNs has become much more pronounced than ever. Since large-scale dataset is one of the critical components for the success of deep learning, lack of sufficient training data makes it difficult to fully train complex CNNs. To tackle the above challenges, in this paper, we propose a boundary-weighted domain adaptive neural network (BOWDA-Net). To make the network more sensitive to the boundaries during segmentation, a boundary-weighted segmentation loss is proposed. Furthermore, an advanced boundary-weighted transfer leaning approach is introduced to address the problem of small medical imaging datasets. We evaluate our proposed model on three different MR prostate datasets. The experimental results demonstrate that the proposed model is more sensitive to object boundaries and outperformed other state-of-the-art methods.

Convolutional Neural Network Technology in Endoscopic Imaging: Artificial Intelligence for Endoscopy

Recently, significant improvements have been made in artificial intelligence. The artificial neural network was introduced in the 1950s. However, because of the low computing power and insufficient datasets available at that time, artificial neural networks suffered from overfitting and vanishing gradient problems for training deep networks. This concept has become more promising owing to the enhanced big data processing capability, improvement in computing power with parallel processing units, and new algorithms for deep neural networks, which are becoming increasingly successful and attracting interest in many domains, including computer vision, speech recognition, and natural language processing. Recent studies in this technology augur well for medical and healthcare applications, especially in endoscopic imaging. This paper provides perspectives on the history, development, applications, and challenges of deep-learning technology.

Radiomics: from qualitative to quantitative imaging

Historically, medical imaging has been a qualitative or semi-quantitative modality. It is difficult to quantify what can be seen in an image, and to turn it into valuable predictive outcomes. As a result of advances in both computational hardware and machine learning algorithms, computers are making great strides in obtaining quantitative information from imaging and correlating it with outcomes. Radiomics, in its two forms "handcrafted and deep," is an emerging field that translates medical images into quantitative data to yield biological information and enable radiologic phenotypic profiling for diagnosis, theragnosis, decision support, and monitoring. Handcrafted radiomics is a multistage process in which features based on shape, pixel intensities, and texture are extracted from radiographs. Within this review, we describe the steps: starting with quantitative imaging data, how it can be extracted, how to correlate it with clinical and biological outcomes, resulting in models that can be used to make predictions, such as survival, or for detection and classification used in diagnostics. The application of deep learning, the second arm of radiomics, and its place in the radiomics workflow is discussed, along with its advantages and disadvantages. To better illustrate the technologies being used, we provide real-world clinical applications of radiomics in oncology, showcasing research on the applications of radiomics, as well as covering its limitations and its future direction.

CAI4CAI: The Rise of Contextual Artificial Intelligence in Computer Assisted Interventions

Data-driven computational approaches have evolved to enable extraction of information from medical images with a reliability, accuracy and speed which is already transforming their interpretation and exploitation in clinical practice. While similar benefits are longed for in the field of interventional imaging, this ambition is challenged by a much higher heterogeneity. Clinical workflows within interventional suites and operating theatres are extremely complex and typically rely on poorly integrated intra-operative devices, sensors, and support infrastructures. Taking stock of some of the most exciting developments in machine learning and artificial intelligence for computer assisted interventions, we highlight the crucial need to take context and human factors into account in order to address these challenges. Contextual artificial intelligence for computer assisted intervention, or CAI4CAI, arises as an emerging opportunity feeding into the broader field of surgical data science. Central challenges being addressed in CAI4CAI include how to integrate the ensemble of prior knowledge and instantaneous sensory information from experts, sensors and actuators; how to create and communicate a faithful and actionable shared representation of the surgery among a mixed human-AI actor team; how to design interventional systems and associated cognitive shared control schemes for online uncertainty-aware collaborative decision making ultimately producing more precise and reliable interventions.

Simulated perfusion MRI data to boost training of convolutional neural networks for lesion fate prediction in acute stroke

The problem of final tissue outcome prediction of acute ischemic stroke is assessed from physically realistic simulated perfusion magnetic resonance images. Different types of simulations with a focus on the arterial input function are discussed. These simulated perfusion magnetic resonance images are fed to convolutional neural network to predict real patients. Performances close to the state-of-the-art performances are obtained with a patient specific approach. This approach consists in training a model only from simulated images tuned to the arterial input function of a tested real patient. This demonstrates the added value of physically realistic simulated images to predict the final infarct from perfusion.

Deep Transferable Compound Representation across Domains and Tasks for Low Data Drug Discovery

The main problem of small molecule-based drug discovery is to find a candidate molecule with increased pharmacological activity, proper ADME, and low toxicity. Recently, machine learning has driven a significant contribution to drug discovery. However, many machine learning methods, such as deep learning-based approaches, require a large amount of training data to form accurate predictions for unseen data. In lead optimization step, the amount of available biological data on small molecule compounds is low, which makes it a challenging problem to apply machine learning methods. The main goal of this study is to design a new approach to handle these situations. To this end, source assay (auxiliary assay) knowledge is utilized to learn a better model to predict the property of new compounds in the target assay. Up to now, the current approaches did not consider that source and target assays are adapted to different target groups with different compounds distribution. In this paper, we propose a new architecture by utilizing graph convolutional network and adversarial domain adaptation network to tackle this issue. To evaluate the proposed approach, we applied it to Tox21, ToxCast, SIDER, HIV, and BACE collections. The results showed the effectiveness of the proposed approach in transferring the related knowledge from source to target data set.

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.

Generative adversarial network in medical imaging: A review

Generative adversarial networks have gained a lot of attention in the computer vision community due to their capability of data generation without explicitly modelling the probability density function. The adversarial loss brought by the discriminator provides a clever way of incorporating unlabeled samples into training and imposing higher order consistency. This has proven to be useful in many cases, such as domain adaptation, data augmentation, and image-to-image translation. These properties have attracted researchers in the medical imaging community, and we have seen rapid adoption in many traditional and novel applications, such as image reconstruction, segmentation, detection, classification, and cross-modality synthesis. Based on our observations, this trend will continue and we therefore conducted a review of recent advances in medical imaging using the adversarial training scheme with the hope of benefiting researchers interested in this technique.

Homologous Component Analysis for Domain Adaptation

Covariate shift assumption based domain adaptation approaches usually utilize only one common transformation to align marginal distributions and make conditional distributions preserved. However, one common transformation may cause loss of useful information, such as variances and neighborhood relationship in both source and target domain. To address this problem, we propose a novel method called homologous component analysis (HCA) where we try to find two totally different but homologous transformations to align distributions with side information and make conditional distributions preserved. As it is hard to find a closed form solution to the corresponding optimization problem, we solve them by means of the alternating direction minimizing method (ADMM) in the context of Stiefel manifolds. We also provide a generalization error bound for domain adaptation in semi-supervised case and two transformations can help to decrease this upper bound more than only one common transformation does. Extensive experiments on synthetic and real data show the effectiveness of the proposed method by comparing its classification accuracy with the state-of-the-art methods and numerical evidence on chordal distance and Frobenius distance shows that resulting optimal transformations are different.

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.

DeepLSR: a deep learning approach for laser speckle reduction

Speckle artifacts degrade image quality in virtually all modalities that utilize coherent energy, including optical coherence tomography, reflectance confocal microscopy, ultrasound, and widefield imaging with laser illumination. We present an adversarial deep learning framework for laser speckle reduction, called DeepLSR (https://durr.jhu.edu/DeepLSR), that transforms images from a source domain of coherent illumination to a target domain of speckle-free, incoherent illumination. We apply this method to widefield images of objects and tissues illuminated with a multi-wavelength laser, using light emitting diode-illuminated images as ground truth. In images of gastrointestinal tissues, DeepLSR reduces laser speckle noise by 6.4 dB, compared to a 2.9 dB reduction from optimized non-local means processing, a 3.0 dB reduction from BM3D, and a 3.7 dB reduction from an optical speckle reducer utilizing an oscillating diffuser. Further, DeepLSR can be combined with optical speckle reduction to reduce speckle noise by 9.4 dB. This dramatic reduction in speckle noise may enable the use of coherent light sources in applications that require small illumination sources and high-quality imaging, including medical endoscopy.

Not-so-supervised: A survey of semi-supervised, multi-instance, and transfer learning in medical image analysis

Machine learning (ML) algorithms have made a tremendous impact in the field of medical imaging. While medical imaging datasets have been growing in size, a challenge for supervised ML algorithms that is frequently mentioned is the lack of annotated data. As a result, various methods that can learn with less/other types of supervision, have been proposed. We give an overview of semi-supervised, multiple instance, and transfer learning in medical imaging, both in diagnosis or segmentation tasks. We also discuss connections between these learning scenarios, and opportunities for future research. A dataset with the details of the surveyed papers is available via https://figshare.com/articles/Database_of_surveyed_literature_in_Not-so-supervised_a_survey_of_semi-supervised_multi-instance_and_transfer_learning_in_medical_image_analysis_/7479416.

Development of Algorithms for Automated Detection of Cervical Pre-Cancers With a Low-Cost, Point-of-Care, Pocket Colposcope

GOAL

In this paper, we propose methods for (1) automatic feature extraction and classification for acetic acid and Lugol's iodine cervigrams and (2) methods for combining features/diagnosis of different contrasts in cervigrams for improved performance.

METHODS

We developed algorithms to pre-process pathology-labeled cervigrams and extract simple but powerful color and textural-based features. The features were used to train a support vector machine model to classify cervigrams based on corresponding pathology for visual inspection with acetic acid, visual inspection with Lugol's iodine, and a combination of the two contrasts.

RESULTS

The proposed framework achieved a sensitivity, specificity, and accuracy of 81.3%, 78.6%, and 80.0%, respectively, when used to distinguish cervical intraepithelial neoplasia (CIN+) relative to normal and benign tissues. This is superior to the average values achieved by three expert physicians on the same data set for discriminating normal/benign cases from CIN+ (77% sensitivity, 51% specificity, and 63% accuracy).

CONCLUSION

The results suggest that utilizing simple color- and textural-based features from visual inspection with acetic acid and visual inspection with Lugol's iodine images may provide unbiased automation of cervigrams.

SIGNIFICANCE

This would enable automated, expert-level diagnosis of cervical pre-cancer at the point of care.