IGU-Aug: Information-guided unsupervised augmentation and pixel-wise contrastive learning for medical image analysis.

Contrastive learning (CL) is a form of self-supervised learning and has been widely used for various tasks. Different from widely studied instance-level contrastive learning, pixel-wise contrastive learning mainly helps with pixel-wise dense prediction tasks. The counter-part to an instance in instance-level CL is a pixel, along with its neighboring context, in pixel-wise CL. Aiming to build better feature representation, there is a vast literature about designing instance augmentation strategies for instance-level CL; but there is little similar work on pixel augmentation for pixel-wise CL with a pixel granularity. In this paper, we attempt to bridge this gap. We first classify a pixel into three categories, namely low-, medium-, and high-informative, based on the information quantity the pixel contains. We then adaptively design separate augmentation strategies for each category in terms of augmentation intensity and sampling ratio. Extensive experiments validate that our information-guided pixel augmentation strategy succeeds in encoding more discriminative representations and surpassing other competitive approaches in unsupervised local feature matching. Furthermore, our pretrained model improves the performance of both one-shot and fully supervised models. To the best of our knowledge, we are the first to propose a pixel augmentation method with a pixel granularity for enhancing unsupervised pixel-wise contrastive learning. Code is available at https: //github.com/Curli-quan/IGU-Aug.

Which images to label for few-shot medical image analysis?

The success of deep learning methodologies hinges upon the availability of meticulously labeled extensive datasets. However, when dealing with medical images, the annotation process for such abundant training data often necessitates the involvement of experienced radiologists, thereby consuming their limited time resources. In order to alleviate this burden, few-shot learning approaches have been developed, which manage to achieve competitive performance levels with only several labeled images. Nevertheless, a crucial yet previously overlooked problem in few-shot learning is about the selection of template images for annotation before learning, which affects the final performance. In this study, we propose a novel TEmplate Choosing Policy (TECP) that aims to identify and select "the most worthy" images for annotation, particularly within the context of multiple few-shot medical tasks, including landmark detection, anatomy detection, and anatomy segmentation. TECP is composed of four integral components: (1) Self-supervised training, which entails training a pre-existing deep model to extract salient features from radiological images; (2) Alternative proposals for localizing informative regions within the images; and (3) Representative Score Estimation, which involves the evaluation and identification of the most representative samples or templates. (4) Ranking, which rank all candidates and select one with highest representative score. The efficacy of the TECP approach is demonstrated through a series of comprehensive experiments conducted on multiple public datasets. Across all three medical tasks, the utilization of TECP yields noticeable improvements in model performance.

High-efficiency integer multiplier for the orbital angular momentum of light.

The spiral transformation has attracted an increasing interest in switching orbital angular momentum (OAM) modes. However, the efficiency is deteriorated by the inevitable gap between the turns of the spiral strips. In order to overcome the problem, a multiple-ring conformal mapping scheme is proposed for efficient multiplication of the OAM of light. The OAM mode at the input plane is divided into concentric rings, which are mapped to multiple sectors and connected into a ring at the output plane. This point-to-point mapping mechanism can avoid the generation of high-order diffraction, leading to high conversion efficiency. The scheme may underpin the development of optical communication and quantum key distribution in OAM-based systems.

Adversarial Medical Image with Hierarchical Feature Hiding.

Deep learning based methods for medical images can be easily compromised by adversarial examples (AEs), posing a great security flaw in clinical decision-making. It has been discovered that conventional adversarial attacks like PGD which optimize the classification logits, are easy to distinguish in the feature space, resulting in accurate reactive defenses. To better understand this phenomenon and reassess the reliability of the reactive defenses for medical AEs, we thoroughly investigate the characteristic of conventional medical AEs. Specifically, we first theoretically prove that conventional adversarial attacks change the outputs by continuously optimizing vulnerable features in a fixed direction, thereby leading to outlier representations in the feature space. Then, a stress test is conducted to reveal the vulnerability of medical images, by comparing with natural images. Interestingly, this vulnerability is a double-edged sword, which can be exploited to hide AEs. We then propose a simple-yet-effective hierarchical feature constraint (HFC), a novel add-on to conventional white-box attacks, which assists to hide the adversarial feature in the target feature distribution. The proposed method is evaluated on three medical datasets, both 2D and 3D, with different modalities. The experimental results demonstrate the superiority of HFC, i.e., it bypasses an array of state-of-the-art adversarial medical AE detectors more efficiently than competing adaptive attacks1, which reveals the deficiencies of medical reactive defense and allows to develop more robust defenses in future.

Learning to Localize Cross-Anatomy Landmarks in X-Ray Images with a Universal Model.

Objective and Impact Statement. In this work, we develop a universal anatomical landmark detection model which learns once from multiple datasets corresponding to different anatomical regions. Compared with the conventional model trained on a single dataset, this universal model not only is more light weighted and easier to train but also improves the accuracy of the anatomical landmark location. Introduction. The accurate and automatic localization of anatomical landmarks plays an essential role in medical image analysis. However, recent deep learning-based methods only utilize limited data from a single dataset. It is promising and desirable to build a model learned from different regions which harnesses the power of big data. Methods. Our model consists of a local network and a global network, which capture local features and global features, respectively. The local network is a fully convolutional network built up with depth-wise separable convolutions, and the global network uses dilated convolution to enlarge the receptive field to model global dependencies. Results. We evaluate our model on four 2D X-ray image datasets totaling 1710 images and 72 landmarks in four anatomical regions. Extensive experimental results show that our model improves the detection accuracy compared to the state-of-the-art methods. Conclusion. Our model makes the first attempt to train a single network on multiple datasets for landmark detection. Experimental results qualitatively and quantitatively show that our proposed model performs better than other models trained on multiple datasets and even better than models trained on a single dataset separately.

Label-Free Segmentation of COVID-19 Lesions in Lung CT.

Scarcity of annotated images hampers the building of automated solution for reliable COVID-19 diagnosis and evaluation from CT. To alleviate the burden of data annotation, we herein present a label-free approach for segmenting COVID-19 lesions in CT via voxel-level anomaly modeling that mines out the relevant knowledge from normal CT lung scans. Our modeling is inspired by the observation that the parts of tracheae and vessels, which lay in the high-intensity range where lesions belong to, exhibit strong patterns. To facilitate the learning of such patterns at a voxel level, we synthesize 'lesions' using a set of simple operations and insert the synthesized 'lesions' into normal CT lung scans to form training pairs, from which we learn a normalcy-recognizing network (NormNet) that recognizes normal tissues and separate them from possible COVID-19 lesions. Our experiments on three different public datasets validate the effectiveness of NormNet, which conspicuously outperforms a variety of unsupervised anomaly detection (UAD) methods.

The autopsy study of 553 cases of sudden cardiac death in Chinese adults.

Despite a recent epidemiological study reporting a lower incidence of sudden cardiac death (SCD) in China as compared with that in Western countries, the exact causes of SCD are still unknown. Using a uniform review protocol and diagnostic criteria, a retrospective autopsy study identified 553 cases of SCD in 14,487 consecutive autopsies from eight regions in China representing different geographic and population features. Their ages ranged from 18 to 80 years (median 43.0 years) with a ratio of 4.3/1.0 for male/female. Out-of-hospital deaths and unwitnessed cases accounted for 74.3 and 22.6 %, respectively. The main causes of death were coronary atherosclerotic disease (CAD 50.3 %), myocarditis (14.8 %), and hypertrophic cardiomyopathy (4.5 %), with unexplained sudden death accounting for 12.1 % of the cases. CAD had a proportion of 10.4 % in victims <35 years, lower as compared with 59.0 and 83.0 % in victims aged 35-54 and in victims ≥55 years. On the other hand, myocarditis and unexplained sudden death were major causes and accounted for 34.7 and 22.5 % in victims <35 years. In order to differentiate the degree of the cause-effect relationship between autopsy findings and sudden death, a grading method was used in this series and characterized 24.3 % of findings as certain, 52.9 % as highly probable, and 22.8 % as uncertain. Our data indicated that there most likely are less CAD but more myocarditis and unexplained sudden death in Chinese youth with SCD than in populations from Western countries. Molecular genetic testing should be conducted in those cases with uncertain findings and unexplained sudden death in routine autopsy.

[Pathologic changes of the cardiac conduction system in 12 patients with abnormal ECG].

OBJECTIVE: To investigate the relationship between abnormal ECG and pathologic changes in the cardiac conduction system (CCS). METHOD: Pathological changes of the CCS in 12 cases with abnormal ECG out of 16 pre-death ECG were observed. RESULTS: (1) Among 7 cases of sudden cardiac death, ECG monitoring recorded bradyarrhythmia in 6 cases, tachyarrhythmia 6 cases, bradycardia-tachycardia syndrome 2 cases, conduction block 6 cases, atrial premature beats 6 cases, ventricular premature beats 6 cases, and ST-T changes 4 cases. (2) The histopathological findings in the CCS were noted in all cases. Of these 12 cases, three had signs of fatty infiltration, and/or fibrous 4 cases, three of amyloidosis, one of chronic inflammatory changes, two of acute inflammatory changes, two of developmental anomalies, two of hemorrhages and one of LAD stenosis. (3) Acute inflammation changes in the CCS corresponded to tachyarrhythmia and multiple ventricular premature beats, whereas chronic inflammation and degenerative changes in the CCS were often related to bradyarrhythmia, bradycardia-tachycardia syndrome and conduction block. (4) The CCS changes alone could lead to ST-T changes in ECG. CONCLUSION: The pathological changes in the CCS are related to ECG changes, and attributed to the pathological bases of arrhythmia.

[Traumatic cerebral infarction: a histopathological study of 17 cases].

OBJECTIVE: To assess the morphologic changes in traumatic cerebral infarction and to discuss its mechanism. METHODS: Specimens from seventeen cases of cerebral infarction were selected from 81 patients with severe brain injury, and subject to routine gross and histological examinations. RESULTS: (1) The cerebral infarction in all cases was hemorrhagic in nature with a wedged or irregular shape upon gross inspection. The lesions were found in occipital gyrus (8 cases), occipital lobes (3 cases), basal nuclei (3 cases), cingulate gyrus (2 cases), and lateral occipitotemporal gyrus (1 case). Histologically, the lesions were located at the junction between the cortex and medulla, showing congestion, edema, hemorrhage, necrotic nerve tissue and blood vessels. In severe cases, the lesion extended into the entire cortex and subarachnoid spaces. (2) Swelling of the brain and cerebral hernia were found in all cases, 8 of which demonstrated that the posterior cerebral artery was compressed and stenotic within the space between the crus cerebri and uncus. CONCLUSION: Brain tissue necrosis in traumatic cerebral infarction is the result of brain swelling and cerebral hernia formation, following congestion, bleeding and ischemia due to vasculature compression.

[Morphological changes of tissues after extensive soft tissue injury].

OBJECTIVE: To study the pathological morphological changes for diagnosing the cause of death of extensive soft tissue injury or crush syndrome. METHODS: The tissues were stained by HE and IHC. RESULTS: (1) The Mb positive rate was 60%, 75%, 95% respectively. (2) Both the HSP70 positive rate of hearts and brains were 90%. CONCLUSION: (1) The animal model of broad soft tissue injury was established. (2) Accumulated the pathological morphological data for diagnosing the cause of death of extensive soft tissue injury or crush syndrome.