Recent advances and clinical applications of deep learning in medical image analysis

Deep learning has received extensive research interest in developing new medical image processing algorithms, and deep learning based models have been remarkably successful in a variety of medical imaging tasks to support disease detection and diagnosis. Despite the success, the further improvement of deep learning models in medical image analysis is majorly bottlenecked by the lack of large-sized and well-annotated datasets. In the past five years, many studies have focused on addressing this challenge. In this paper, we reviewed and summarized these recent studies to provide a comprehensive overview of applying deep learning methods in various medical image analysis tasks. Especially, we emphasize the latest progress and contributions of state-of-the-art unsupervised and semi-supervised deep learning in medical image analysis, which are summarized based on different application scenarios, including classification, segmentation, detection, and image registration. We also discuss major technical challenges and suggest possible solutions in the future research efforts.     

Deep label fusion: A generalizable hybrid multi-atlas and deep convolutional neural network for medical image segmentation

Deep convolutional neural networks (DCNN) achieve very high accuracy in segmenting various anatomical structures in medical images but often suffer from relatively poor generalizability. Multi-atlas segmentation (MAS), while less accurate than DCNN in many applications, tends to generalize well to unseen datasets with different characteristics from the training dataset. Several groups have attempted to integrate the power of DCNN to learn complex data representations and the robustness of MAS to changes in image characteristics. However, these studies primarily focused on replacing individual components of MAS with DCNN models and reported marginal improvements in accuracy. In this study we describe and evaluate a 3D end-to-end hybrid MAS and DCNN segmentation pipeline, called Deep Label Fusion (DLF). The DLF pipeline consists of two main components with learnable weights, including a weighted voting subnet that mimics the MAS algorithm and a fine-tuning subnet that corrects residual segmentation errors to improve final segmentation accuracy. We evaluate DLF on five datasets that represent a diversity of anatomical structures (medial temporal lobe subregions and lumbar vertebrae) and imaging modalities (multi-modality, multi-field-strength MRI and Computational Tomography). These experiments show that DLF achieves comparable segmentation accuracy to nnU-Net (Isensee et al., 2020), the state-of-the-art DCNN pipeline, when evaluated on a dataset with similar characteristics to the training datasets, while outperforming nnU-Net on tasks that involve generalization to datasets with different characteristics (different MRI field strength or different patient population). DLF is also shown to consistently improve upon conventional MAS methods. In addition, a modality augmentation strategy tailored for multimodal imaging is proposed and demonstrated to be beneficial in improving the segmentation accuracy of learning-based methods, including DLF and DCNN, in missing data scenarios in test time as well as increasing the interpretability of the contribution of each individual modality.     

Interpretability-Guided Inductive Bias For Deep Learning Based Medical Image

Deep learning methods provide state of the art performance for supervised learning based medical image analysis. However it is essential that trained models extract clinically relevant features for downstream tasks as, otherwise, shortcut learning and generalization issues can occur. Furthermore in the medical field, trustability and transparency of current deep learning systems is a much desired property. In this paper we propose an interpretability-guided inductive bias approach enforcing that learned features yield more distinctive and spatially consistent saliency maps for different class labels of trained models, leading to improved model performance. We achieve our objectives by incorporating a class-distinctiveness loss and a spatial-consistency regularization loss term. Experimental results for medical image classification and segmentation tasks show our proposed approach outperforms conventional methods, while yielding saliency maps in higher agreement with clinical experts. Additionally, we show how information from unlabeled images can be used to further boost performance. In summary, the proposed approach is modular, applicable to existing network architectures used for medical imaging applications, and yields improved learning rates, model robustness, and model interpretability.     

A survey on incorporating domain knowledge into deep learning for medical image analysis

Although deep learning models like CNNs have achieved great success in medical image analysis, the small size of medical datasets remains a major bottleneck in this area. To address this problem, researchers have started looking for external information beyond current available medical datasets. Traditional approaches generally leverage the information from natural images via transfer learning. More recent works utilize the domain knowledge from medical doctors, to create networks that resemble how medical doctors are trained, mimic their diagnostic patterns, or focus on the features or areas they pay particular attention to. In this survey, we summarize the current progress on integrating medical domain knowledge into deep learning models for various tasks, such as disease diagnosis, lesion, organ and abnormality detection, lesion and organ segmentation. For each task, we systematically categorize different kinds of medical domain knowledge that have been utilized and their corresponding integrating methods. We also provide current challenges and directions for future research.     

Deep learning with noisy labels: Exploring techniques and remedies in medical image analysis

Supervised training of deep learning models requires large labeled datasets. There is a growing interest in obtaining such datasets for medical image analysis applications. However, the impact of label noise has not received sufficient attention. Recent studies have shown that label noise can significantly impact the performance of deep learning models in many machine learning and computer vision applications. This is especially concerning for medical applications, where datasets are typically small, labeling requires domain expertise and suffers from high inter- and intra-observer variability, and erroneous predictions may influence decisions that directly impact human health. In this paper, we first review the state-of-the-art in handling label noise in deep learning. Then, we review studies that have dealt with label noise in deep learning for medical image analysis. Our review shows that recent progress on handling label noise in deep learning has gone largely unnoticed by the medical image analysis community. To help achieve a better understanding of the extent of the problem and its potential remedies, we conducted experiments with three medical imaging datasets with different types of label noise, where we investigated several existing strategies and developed new methods to combat the negative effect of label noise. Based on the results of these experiments and our review of the literature, we have made recommendations on methods that can be used to alleviate the effects of different types of label noise on deep models trained for medical image analysis. We hope that this article helps the medical image analysis researchers and developers in choosing and devising new techniques that effectively handle label noise in deep learning.     

Suggestive annotation of brain MR images with gradient-guided sampling

Machine learning has been widely adopted for medical image analysis in recent years given its promising performance in image segmentation and classification tasks. The success of machine learning, in particular supervised learning, depends on the availability of manually annotated datasets. For medical imaging applications, such annotated datasets are not easy to acquire, it takes a substantial amount of time and resource to curate an annotated medical image set. In this paper, we propose an efficient annotation framework for brain MR images that can suggest informative sample images for human experts to annotate. We evaluate the framework on two different brain image analysis tasks, namely brain tumour segmentation and whole brain segmentation. Experiments show that for brain tumour segmentation task on the BraTS 2019 dataset, training a segmentation model with only 7% suggestively annotated image samples can achieve a performance comparable to that of training on the full dataset. For whole brain segmentation on the MALC dataset, training with 42% suggestively annotated image samples can achieve a comparable performance to training on the full dataset. The proposed framework demonstrates a promising way to save manual annotation cost and improve data efficiency in medical imaging applications.     

Enhancing MR image segmentation with realistic adversarial data augmentation

The success of neural networks on medical image segmentation tasks typically relies on large labeled datasets for model training. However, acquiring and manually labeling a large medical image set is resource-intensive, expensive, and sometimes impractical due to data sharing and privacy issues. To address this challenge, we propose AdvChain, a generic adversarial data augmentation framework, aiming at improving both the diversity and effectiveness of training data for medical image segmentation tasks. AdvChain augments data with dynamic data augmentation, generating randomly chained photo-metric and geometric transformations to resemble realistic yet challenging imaging variations to expand training data. By jointly optimizing the data augmentation model and a segmentation network during training, challenging examples are generated to enhance network generalizability for the downstream task. The proposed adversarial data augmentation does not rely on generative networks and can be used as a plug-in module in general segmentation networks. It is computationally efficient and applicable for both low-shot supervised and semi-supervised learning. We analyze and evaluate the method on two MR image segmentation tasks: cardiac segmentation and prostate segmentation with limited labeled data. Results show that the proposed approach can alleviate the need for labeled data while improving model generalization ability, indicating its practical value in medical imaging applications.     

Dynamic-weighting hierarchical segmentation network for medical images

Automatic medical image segmentation plays a crucial role in many medical image analysis applications, such as disease diagnosis and prognosis. Despite the extensive progress of existing deep learning based models for medical image segmentation, they focus on extracting accurate features by designing novel network structures and solely utilize fully connected (FC) layer for pixel-level classification. Considering the insufficient capability of FC layer to encode the extracted diverse feature representations, we propose a Hierarchical Segmentation (HieraSeg) Network for medical image segmentation and devise a Hierarchical Fully Connected (HFC) layer. Specifically, it consists of three classifiers and decouples each category into several subcategories by introducing multiple weight vectors to denote the diverse characteristics in each category. A subcategory-level and a category-level learning schemes are then designed to explicitly enforce the discrepant subcategories and automatically capture the most representative characteristics. Hence, the HFC layer can fit the variant characteristics so as to derive an accurate decision boundary. To enhance the robustness of HieraSeg Network with the variability of lesions, we further propose a Dynamic-Weighting HieraSeg (DW-HieraSeg) Network, which introduces an Image-level Weight Net (IWN) and a Pixel-level Weight Net (PWN) to learn data-driven curriculum. Through progressively incorporating informative images and pixels in an easy-to-hard manner, DW-HieraSeg Network is able to eliminate local optimums and accelerate the training process. Additionally, a class balanced loss is proposed to constrain the PWN for preventing the overfitting problem in minority category. Comprehensive experiments on three benchmark datasets, EndoScene, ISIC and Decathlon, show our newly proposed HieraSeg and DW-HieraSeg Networks achieve state-of-the-art performance, which clearly demonstrates the effectiveness of the proposed approaches for medical image segmentation.     

Learning joint segmentation of tissues and brain lesions from task-specific hetero-modal domain-shifted datasets

Brain tissue segmentation from multimodal MRI is a key building block of many neuroimaging analysis pipelines. Established tissue segmentation approaches have, however, not been developed to cope with large anatomical changes resulting from pathology, such as white matter lesions or tumours, and often fail in these cases. In the meantime, with the advent of deep neural networks (DNNs), segmentation of brain lesions has matured significantly. However, few existing approaches allow for the joint segmentation of normal tissue and brain lesions. Developing a DNN for such a joint task is currently hampered by the fact that annotated datasets typically address only one specific task and rely on task-specific imaging protocols including a task-specific set of imaging modalities. In this work, we propose a novel approach to build a joint tissue and lesion segmentation model from aggregated task-specific hetero-modal domain-shifted and partially-annotated datasets. Starting from a variational formulation of the joint problem, we show how the expected risk can be decomposed and optimised empirically. We exploit an upper bound of the risk to deal with heterogeneous imaging modalities across datasets. To deal with potential domain shift, we integrated and tested three conventional techniques based on data augmentation, adversarial learning and pseudo-healthy generation. For each individual task, our joint approach reaches comparable performance to task-specific and fully-supervised models. The proposed framework is assessed on two different types of brain lesions: White matter lesions and gliomas. In the latter case, lacking a joint ground-truth for quantitative assessment purposes, we propose and use a novel clinically-relevant qualitative assessment methodology.     

Deep learning for computational cytology: A survey

Computational cytology is a critical, rapid-developing, yet challenging topic in medical image computing concerned with analyzing digitized cytology images by computer-aided technologies for cancer screening. Recently, an increasing number of deep learning (DL) approaches have made significant achievements in medical image analysis, leading to boosting publications of cytological studies. In this article, we survey more than 120 publications of DL-based cytology image analysis to investigate the advanced methods and comprehensive applications. We first introduce various deep learning schemes, including fully supervised, weakly supervised, unsupervised, and transfer learning. Then, we systematically summarize public datasets, evaluation metrics, versatile cytology image analysis applications including cell classification, slide-level cancer screening, nuclei or cell detection and segmentation. Finally, we discuss current challenges and potential research directions of computational cytology.     

ACEnet: Anatomical context-encoding network for neuroanatomy segmentation

Segmentation of brain structures from magnetic resonance (MR) scans plays an important role in the quantification of brain morphology. Since 3D deep learning models suffer from high computational cost, 2D deep learning methods are favored for their computational efficiency. However, existing 2D deep learning methods are not equipped to effectively capture 3D spatial contextual information that is needed to achieve accurate brain structure segmentation. In order to overcome this limitation, we develop an Anatomical Context-Encoding Network (ACEnet) to incorporate 3D spatial and anatomical contexts in 2D convolutional neural networks (CNNs) for efficient and accurate segmentation of brain structures from MR scans, consisting of 1) an anatomical context encoding module to incorporate anatomical information in 2D CNNs and 2) a spatial context encoding module to integrate 3D image information in 2D CNNs. In addition, a skull stripping module is adopted to guide the 2D CNNs to attend to the brain. Extensive experiments on three benchmark datasets have demonstrated that our method achieves promising performance compared with state-of-the-art alternative methods for brain structure segmentation in terms of both computational efficiency and segmentation accuracy.     

深度学习在医学影像中的应用

医学影像是放射科医生做出医学诊断的重要依据。但随着医学影像技术的快速发展, 逐渐增多的影像图像和复杂的图像信息对医生的工作产生了巨大的挑战。而深度学习是人工智能研究中最热门的领域, 在处理大数据和提取有效信息方面具有优势, 因此逐渐成为分析医学影像方面的首选方法。本文阐述了深度学习的概念, 并简要总结深度学习在医学影像中的常见模型, 包括卷积神经网络、循环神经网络、深度置信网络和自动编码器。卷积神经网络的基本结构是卷积层、池化层和全连接层; 循环神经网络由输入层、隐藏层和输出层组成; 深度置信网络的基础是玻尔兹曼机; 自动编码器包含编码层、隐藏层和解码层。通过对CT肺结节和MRI脑部疾病的分类, 阐明目前深度学习在疾病自动分类上准确性较高; 通过分割左心室、椎旁肌肉和肝脏的结构, 可见深度学习方法在医学图像分割上与人为分割具有一致性; 深度学习在肺结节和乳腺癌疾病的检测上已相对成熟。但目前为止, 仍存在标注的样本量少和过拟合的问题, 希望通过共享图像数据库来解决此问题。总之, 深度学习在医学影像中具有广阔前景, 且对临床医生的工作具有重大意义。      

HAL-IA: A Hybrid Active Learning framework using Interactive Annotation for medical image segmentation

High performance of deep learning models on medical image segmentation greatly relies on large amount of pixel-wise annotated data, yet annotations are costly to collect. How to obtain high accuracy segmentation labels of medical images with limited cost (e.g. time) becomes an urgent problem. Active learning can reduce the annotation cost of image segmentation, but it faces three challenges: the cold start problem, an effective sample selection strategy for segmentation task and the burden of manual annotation. In this work, we propose a Hybrid Active Learning framework using Interactive Annotation (HAL-IA) for medical image segmentation, which reduces the annotation cost both in decreasing the amount of the annotated images and simplifying the annotation process. Specifically, we propose a novel hybrid sample selection strategy to select the most valuable samples for segmentation model performance improvement. This strategy combines pixel entropy, regional consistency and image diversity to ensure that the selected samples have high uncertainty and diversity. In addition, we propose a warm-start initialization strategy to build the initial annotated dataset to avoid the cold-start problem. To simplify the manual annotation process, we propose an interactive annotation module with suggested superpixels to obtain pixel-wise label with several clicks. We validate our proposed framework with extensive segmentation experiments on four medical image datasets. Experimental results showed that the proposed framework achieves high accuracy pixel-wise annotations and models with less labeled data and fewer interactions, outperforming other state-of-the-art methods. Our method can help physicians efficiently obtain accurate medical image segmentation results for clinical analysis and diagnosis.     

Towards annotation-efficient segmentation via image-to-image translation

An important challenge and limiting factor in deep learning methods for medical imaging segmentation is the lack of available of annotated data to properly train models. For the specific task of tumor segmentation, the process entails clinicians labeling every slice of volumetric scans for every patient, which becomes prohibitive at the scale of datasets required to train neural networks to optimal performance. To address this, we propose a novel semi-supervised framework that allows training any segmentation (encoder–decoder) model using only information readily available in radiological data, namely the presence of a tumor in the image, in addition to a few annotated images. Specifically, we conjecture that a generative model performing domain translation on this weak label — healthy vs diseased scans — helps achieve tumor segmentation. The proposed GenSeg method first disentangles tumoral tissue from healthy “background” tissue. The latent representation is separated into (1) the common background information across both domains, and (2) the unique tumoral information. GenSeg then achieves diseased-to-healthy image translation by decoding a healthy version of the image from just the common representation, as well as a residual image that allows adding back the tumors. The same decoder that produces this residual tumor image, also outputs a tumor segmentation. Implicit data augmentation is achieved by re-using the same framework for healthy-to-diseased image translation, where a residual tumor image is produced from a prior distribution. By performing both image translation and segmentation simultaneously, GenSeg allows training on only partially annotated datasets. To test the framework, we trained U-Net-like architectures using GenSeg and evaluated their performance on 3 variants of a synthetic task, as well as on 2 benchmark datasets: brain tumor segmentation in MRI (derived from BraTS) and liver metastasis segmentation in CT (derived from LiTS). Our method outperforms the baseline semi-supervised (autoencoder and mean teacher) and supervised segmentation methods, with improvements ranging between 8–14% Dice score on the brain task and 5–8% on the liver task, when only 1% of the training images were annotated. These results show the proposed framework is ideal at addressing the problem of training deep segmentation models when a large portion of the available data is unlabeled and unpaired, a common issue in tumor segmentation.     

Weakly supervised histopathology image segmentation with self-attention

Accurate segmentation in histopathology images at pixel-level plays a critical role in the digital pathology workflow. The development of weakly supervised methods for histopathology image segmentation liberates pathologists from time-consuming and labor-intensive works, opening up possibilities of further automated quantitative analysis of whole-slide histopathology images. As an effective subgroup of weakly supervised methods, multiple instance learning (MIL) has achieved great success in histopathology images. In this paper, we specially treat pixels as instances so that the histopathology image segmentation task is transformed into an instance prediction task in MIL. However, the lack of relations between instances in MIL limits the further improvement of segmentation performance. Therefore, we propose a novel weakly supervised method called SA-MIL for pixel-level segmentation in histopathology images. SA-MIL introduces a self-attention mechanism into the MIL framework, which captures global correlation among all instances. In addition, we use deep supervision to make the best use of information from limited annotations in the weakly supervised method. Our approach makes up for the shortcoming that instances are independent of each other in MIL by aggregating global contextual information. We demonstrate state-of-the-art results compared to other weakly supervised methods on two histopathology image datasets. It is evident that our approach has generalization ability for the high performance on both tissue and cell histopathology datasets. There is potential in our approach for various applications in medical images.     

Test-time adaptable neural networks for robust medical image segmentation

Convolutional Neural Networks (CNNs) work very well for supervised learning problems when the training dataset is representative of the variations expected to be encountered at test time. In medical image segmentation, this premise is violated when there is a mismatch between training and test images in terms of their acquisition details, such as the scanner model or the protocol. Remarkable performance degradation of CNNs in this scenario is well documented in the literature. To address this problem, we design the segmentation CNN as a concatenation of two sub-networks: a relatively shallow image normalization CNN, followed by a deep CNN that segments the normalized image. We train both these sub-networks using a training dataset, consisting of annotated images from a particular scanner and protocol setting. Now, at test time, we adapt the image normalization sub-network for each test image, guided by an implicit prior on the predicted segmentation labels. We employ an independently trained denoising autoencoder (DAE) in order to model such an implicit prior on plausible anatomical segmentation labels. We validate the proposed idea on multi-center Magnetic Resonance imaging datasets of three anatomies: brain, heart and prostate. The proposed test-time adaptation consistently provides performance improvement, demonstrating the promise and generality of the approach. Being agnostic to the architecture of the deep CNN, the second sub-network, the proposed design can be utilized with any segmentation network to increase robustness to variations in imaging scanners and protocols. Our code is available at: https://github.com/neerakara/test-time-adaptable-neural-networks-for-domain-generalization.     

Models Genesis

Transfer learning from natural images to medical images has been established as one of the most practical paradigms in deep learning for medical image analysis. To fit this paradigm, however, 3D imaging tasks in the most prominent imaging modalities (e.g., CT and MRI) have to be reformulated and solved in 2D, losing rich 3D anatomical information, thereby inevitably compromising its performance. To overcome this limitation, we have built a set of models, called Generic Autodidactic Models, nicknamed Models Genesis, because they are created ex nihilo (with no manual labeling), self-taught (learnt by self-supervision), and generic (served as source models for generating application-specific target models). Our extensive experiments demonstrate that our Models Genesis significantly outperform learning from scratch and existing pre-trained 3D models in all five target 3D applications covering both segmentation and classification. More importantly, learning a model from scratch simply in 3D may not necessarily yield performance better than transfer learning from ImageNet in 2D, but our Models Genesis consistently top any 2D/2.5D approaches including fine-tuning the models pre-trained from ImageNet as well as fine-tuning the 2D versions of our Models Genesis, confirming the importance of 3D anatomical information and significance of Models Genesis for 3D medical imaging. This performance is attributed to our unified self-supervised learning framework, built on a simple yet powerful observation: the sophisticated and recurrent anatomy in medical images can serve as strong yet free supervision signals for deep models to learn common anatomical representation automatically via self-supervision. As open science, all codes and pre-trained Models Genesis are available at https://github.com/MrGiovanni/ModelsGenesis.     

Embracing imperfect datasets: A review of deep learning solutions for medical image segmentation

The medical imaging literature has witnessed remarkable progress in high-performing segmentation models based on convolutional neural networks. Despite the new performance highs, the recent advanced segmentation models still require large, representative, and high quality annotated datasets. However, rarely do we have a perfect training dataset, particularly in the field of medical imaging, where data and annotations are both expensive to acquire. Recently, a large body of research has studied the problem of medical image segmentation with imperfect datasets, tackling two major dataset limitations: scarce annotations where only limited annotated data is available for training, and weak annotations where the training data has only sparse annotations, noisy annotations, or image-level annotations. In this article, we provide a detailed review of the solutions above, summarizing both the technical novelties and empirical results. We further compare the benefits and requirements of the surveyed methodologies and provide our recommended solutions. We hope this survey article increases the community awareness of the techniques that are available to handle imperfect medical image segmentation datasets.     

Distributed contrastive learning for medical image segmentation

Supervised deep learning needs a large amount of labeled data to achieve high performance. However, in medical imaging analysis, each site may only have a limited amount of data and labels, which makes learning ineffective. Federated learning (FL) can learn a shared model from decentralized data. But traditional FL requires fully-labeled data for training, which is very expensive to obtain. Self-supervised contrastive learning (CL) can learn from unlabeled data for pre-training, followed by fine-tuning with limited annotations. However, when adopting CL in FL, the limited data diversity on each site makes federated contrastive learning (FCL) ineffective. In this work, we propose two federated self-supervised learning frameworks for volumetric medical image segmentation with limited annotations. The first one features high accuracy and fits high-performance servers with high-speed connections. The second one features lower communication costs, suitable for mobile devices. In the first framework, features are exchanged during FCL to provide diverse contrastive data to each site for effective local CL while keeping raw data private. Global structural matching aligns local and remote features for a unified feature space among different sites. In the second framework, to reduce the communication cost for feature exchanging, we propose an optimized method FCLOpt that does not rely on negative samples. To reduce the communications of model download, we propose the predictive target network update (PTNU) that predicts the parameters of the target network. Based on PTNU, we propose the distance prediction (DP) to remove most of the uploads of the target network. Experiments on a cardiac MRI dataset show the proposed two frameworks substantially improve the segmentation and generalization performance compared with state-of-the-art techniques.