GraphSKT: Graph-Guided Structured Knowledge Transfer for Domain Adaptive Lesion Detection.

Adversarial-based adaptation has dominated the area of domain adaptive detection over the past few years. Despite their general efficacy for various tasks, the learned representations may not capture the intrinsic topological structures of the whole images and thus are vulnerable to distributional shifts especially in real-world applications, such as geometric distortions across imaging devices in medical images. In this case, forcefully matching data distributions across domains cannot ensure precise knowledge transfer and are prone to result in the negative transfer. In this paper, we explore the problem of domain adaptive lesion detection from the perspective of relational reasoning, and propose a Graph-Structured Knowledge Transfer (GraphSKT) framework to perform hierarchical reasoning by modeling both the intra- and inter-domain topological structures. To be specific, we utilize cross-domain correspondence to mine meaningful foreground regions for representing graph nodes and explicitly endow each node with contextual information. Then, the intra- and inter-domain graphs are built on the top of instance-level features to achieve a high-level understanding of the lesion and whole medical image, and transfer the structured knowledge from source to target domains. The contextual and semantic information is propagated through graph nodes methodically, enhancing the expressive power of learned features for the lesion detection tasks. Extensive experiments on two types of challenging datasets demonstrate that the proposed GraphSKT significantly outperforms the state-of-the-art approaches for detection of polyps in colonoscopy images and of mass in mammographic images.

Highly efficient and stable catalysts-covalent organic framework-supported palladium particles for 4-nitrophenol catalytic hydrogenation.

A covalent organic framework (COF) was used as the support of the catalyst in this work in order to obtain an environmentally friendly catalyst with high catalytic performance, selectivity and stability for 4-nitrophenol hydrogenation. Pd tiny particles are fixed in the cavity of COF to obtain Pd/COF catalysts, which has a quite narrow particle size distribution (5.09 ± 1.30 nm). As-prepared Pd/COF catalysts (Pd loading-2.11 wt%) shows excellent catalytic performance (conversion - 99.3%, selectivity >99.0% and turnover frequency (TOF)-989.4 h-1) for 4-nitrophenol hydrogenation under relatively mild reaction conditions of reaction temperature-40 °C and reaction pressure-3.0 MPa H2, and Pd/COF catalysts have high stability. Pd/COF catalysts were characterized by X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), scanning electron microscope energy-dispersive X-ray spectroscopy (SEM-EDS), transmission electron microscope (TEM), high resolution TEM (HRTEM), Brunauer-Emmett-Teller (BET), scanning TEM energy-dispersive X-ray spectroscopy (STEM-EDS) elemental analysis techniques to prove that the Pd nanoparticles are highly dispersed on the COF. Pd/COF catalysts have good stability and reusability hence with certain industrial application value.

Cardiac segmentation on late gadolinium enhancement MRI: A benchmark study from multi-sequence cardiac MR segmentation challenge.

Accurate computing, analysis and modeling of the ventricles and myocardium from medical images are important, especially in the diagnosis and treatment management for patients suffering from myocardial infarction (MI). Late gadolinium enhancement (LGE) cardiac magnetic resonance (CMR) provides an important protocol to visualize MI. However, compared with the other sequences LGE CMR images with gold standard labels are particularly limited. This paper presents the selective results from the Multi-Sequence Cardiac MR (MS-CMR) Segmentation challenge, in conjunction with MICCAI 2019. The challenge offered a data set of paired MS-CMR images, including auxiliary CMR sequences as well as LGE CMR, from 45 patients who underwent cardiomyopathy. It was aimed to develop new algorithms, as well as benchmark existing ones for LGE CMR segmentation focusing on myocardial wall of the left ventricle and blood cavity of the two ventricles. In addition, the paired MS-CMR images could enable algorithms to combine the complementary information from the other sequences for the ventricle segmentation of LGE CMR. Nine representative works were selected for evaluation and comparisons, among which three methods are unsupervised domain adaptation (UDA) methods and the other six are supervised. The results showed that the average performance of the nine methods was comparable to the inter-observer variations. Particularly, the top-ranking algorithms from both the supervised and UDA methods could generate reliable and robust segmentation results. The success of these methods was mainly attributed to the inclusion of the auxiliary sequences from the MS-CMR images, which provide important label information for the training of deep neural networks. The challenge continues as an ongoing resource, and the gold standard segmentation as well as the MS-CMR images of both the training and test data are available upon registration via its homepage (www.sdspeople.fudan.edu.cn/zhuangxiahai/0/mscmrseg/).

Enhanced Oil Recovery Mechanism and Technical Boundary of Gel Foam Profile Control System for Heterogeneous Reservoirs in Changqing.

The gel plugging and flooding system has a long history of being researched and applied, but the Changqing reservoir geological characteristics are complex, and the synergistic performance of the composite gel foam plugging system is not fully understood, resulting in poor field application. Additionally, the technique boundary chart of the heterogeneous reservoir plugging system has hardly appeared. In this work, reservoir models of porous, fracture, and pore-fracture were constructed, a composite gel foam plugging system was developed, and its static injection and dynamic profile control and oil displacement performance were evaluated. Finally, combined with the experimental studies, a technical boundary chart of plugging systems for heterogeneous reservoirs is proposed. The research results show that the adsorption effect of microspheres (WQ-100) on the surface of elastic gel particles-1 (PEG-1) is more potent than that of pre-crosslinked particle gel (PPG) and the deposition is mainly on the surface of PPG. The adsorption effect of PEG-1 on the surface of PPG is not apparent, primarily manifested as deposition stacking. The gel was synthesized with 0.2% hydrolyzed polyacrylamide (HPAM) + 0.2% organic chromium cross-linking agent, and the strength of enhanced gel with WQ-100 was higher than that of PEG-1 and PPG. The comprehensive value of WQ-100 reinforced foam is greater than that of PEG-1, and PPG reinforced foam, and the enhanced foam with gel has a thick liquid film and poor foaming effect. For the heterogeneous porous reservoir with the permeability of 5/100 mD, the enhanced foam with WQ-100 shows better performance in plugging control and flooding, and the recovery factor increases by 28.05%. The improved foam with gel enhances the fluid flow diversion ability and the recovery factor of fractured reservoirs with fracture widths of 50 µm and 180 µm increases by 29.41% and 24.39%, respectively. For pore-fractured reservoirs with a permeability of 52/167 mD, the PEG + WQ-100 microsphere and enhanced foam with WQ-100 systems show better plugging and recovering performance, and the recovery factor increases are 20.52% and 17.08%, 24.44%, and 21.43%, respectively. The smaller the particle size of the prefabricated gel, the more uniform the adsorption on the foam liquid film and the stronger the stability of the foam system. The plugging performance of the composite gel system is stronger than that of the enhanced gel with foam, but the oil displacement performance of the gel-enhanced foam is better than that of the composite gel system due to the "plug-flooding-integrated" feature of the foam. Combined with the plugging and flooding performance of each plugging system, a technique boundary chart for the plugging system was established for the coexisting porous, fracture, and pore-fracture heterogeneous reservoirs in Changqing Oilfield.

Closed or open platform? the nature of platform and a qualitative comparative analysis of the performance effect of platform openness.

Internet platform enterprises have become one of the dominant organizational forms for internet-based businesses. Despite the strategically crucial role that openness decision plays for Internet platform enterprises, the results of existing research on the relationship between platform openness and platform performance are not conclusive. As to the nature of platform, its transaction attribute has been overemphasized while its innovation attribute is mostly neglected. Through decomposing platform openness into supply-side openness and demand-side openness, as well as introducing demand diversity and knowledge complexity as contextual variables, this study attempts to understand the impact of both types of attributes on performance by considering their configuration. Using fuzzy sets qualitative comparative analysis (fsQCA) method, we find that high demand diversity of platform users and high supply-side openness will lead to better platform performance. Moreover, the high knowledge complexity required for platform innovation together with high supply-side and demand-side openness will contribute to a high level of platform performance.

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.

An Adversarial Learning Approach to Medical Image Synthesis for Lesion Detection.

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