Boundary uncertainty aware network for automated polyp segmentation.

Recently, leveraging deep neural networks for automated colorectal polyp segmentation has emerged as a hot topic due to the favored advantages in evading the limitations of visual inspection, e.g., overwork and subjectivity. However, most existing methods do not pay enough attention to the uncertain areas of colonoscopy images and often provide unsatisfactory segmentation performance. In this paper, we propose a novel boundary uncertainty aware network (BUNet) for precise and robust colorectal polyp segmentation. Specifically, considering that polyps vary greatly in size and shape, we first adopt a pyramid vision transformer encoder to learn multi-scale feature representations. Then, a simple yet effective boundary exploration module (BEM) is proposed to explore boundary cues from the low-level features. To make the network focus on the ambiguous area where the prediction score is biased to neither the foreground nor the background, we further introduce a boundary uncertainty aware module (BUM) that explores error-prone regions from the high-level features with the assistance of boundary cues provided by the BEM. Through the top-down hybrid deep supervision, our BUNet implements coarse-to-fine polyp segmentation and finally localizes polyp regions precisely. Extensive experiments on five public datasets show that BUNet is superior to thirteen competing methods in terms of both effectiveness and generalization ability.

Histological changes in retinal detachment: A systematic review for the clinician.

Although there have been numerous innovations in the management of retinal detachment (RD) over the past decades, there is still limited understanding of the pathophysiological processes that take place before and after repair. Summarizing key concepts using animal studies may allow for a better assessment of common pre- and postoperative microstructural abnormalities in RD. We performed a systematic literature review on Ovid MEDLINE, EMBASE, and Cochrane Controlled Register of Trials from January 1968 to January 2022, searching animal or human studies reporting retinal histologic changes following primary or induced RD. Thirty-two studies were included. Main cellular events were summarized: photoceptor apoptosis occurs as early as 12 hours after RD and, although most cells survive, there is extensive remodeling. Outer segments progressively degenerate, while inner segments are reorganized. Rod and cone opsins are redistributed, and rod axons retract while cones undergo changes in shape. Second- and third-order neurons rearrange their dendritic processes, and Müller cells become hypertrophic, growing into the subretinal space. Finally, retinal pigment epithelium cells undergo a change in their morphology. Acknowledging critical morphologic changes following RD is crucial in understanding why anatomical and functional outcomes can vary. Insights from histological studies, together with high-resolution imaging, may be key in identifying novel biomarkers in RD.

Adaptive Cross-Feature Fusion Network With Inconsistency Guidance for Multi-Modal Brain Tumor Segmentation.

In the context of contemporary artificial intelligence, increasing deep learning (DL) based segmentation methods have been recently proposed for brain tumor segmentation (BraTS) via analysis of multi-modal MRI. However, known DL-based works usually directly fuse the information of different modalities at multiple stages without considering the gap between modalities, leaving much room for performance improvement. In this paper, we introduce a novel deep neural network, termed ACFNet, for accurately segmenting brain tumor in multi-modal MRI. Specifically, ACFNet has a parallel structure with three encoder-decoder streams. The upper and lower streams generate coarse predictions from individual modality, while the middle stream integrates the complementary knowledge of different modalities and bridges the gap between them to yield fine prediction. To effectively integrate the complementary information, we propose an adaptive cross-feature fusion (ACF) module at the encoder that first explores the correlation information between the feature representations from upper and lower streams and then refines the fused correlation information. To bridge the gap between the information from multi-modal data, we propose a prediction inconsistency guidance (PIG) module at the decoder that helps the network focus more on error-prone regions through a guidance strategy when incorporating the features from the encoder. The guidance is obtained by calculating the prediction inconsistency between upper and lower streams and highlights the gap between multi-modal data. Extensive experiments on the BraTS 2020 dataset show that ACFNet is competent for the BraTS task with promising results and outperforms six mainstream competing methods.

Specificity-aware Federated Learning with Dynamic Feature Fusion Network for Imbalanced Medical Image Classification.

Recently, federated learning has become a powerful technique for medical image classification due to its ability to utilize datasets from multiple clinical clients while satisfying privacy constraints. However, there are still some obstacles in federated learning. Firstly, most existing methods directly average the model parameters collected by medical clients on the server, ignoring the specificities of the local models. Secondly, class imbalance is a common issue in medical datasets. In this paper, to handle these two challenges, we propose a novel specificity-aware federated learning framework that benefits from an Adaptive Aggregation Mechanism (AdapAM) and a Dynamic Feature Fusion Strategy (DFFS). Considering the specificity of each local model, we set the AdapAM on the server. The AdapAM utilizes reinforcement learning to adaptively weight and aggregate the parameters of local models based on their data distribution and performance feedback for obtaining the global model parameters. For the class imbalance in local datasets, we propose the DFFS to dynamically fuse the features of majority classes based on the imbalance ratio in the min-batch and collaborate the rest of features. We conduct extensive experiments on a dermoscopic dataset and a fundus image dataset. Experimental results show that our method can achieve state-of-the-art results in these two real-world medical applications.

A Meta-Analysis of Neurodevelopmental Outcomes following Intravitreal Bevacizumab for the Treatment of Retinopathy of Prematurity.

BACKGROUND: Retinopathy of prematurity (ROP) is the most common cause of preventable blindness in preterm infants. First-line treatments include intravitreal bevacizumab (IVB) or laser photocoagulation (LPC). OBJECTIVES: The aim of the study was to evaluate neurodevelopmental safety of IVB compared to LPC for ROP. METHODS: MEDLINE, Embase, and Cochrane library were searched up to September 2022. Studies were included with at least 12-month follow-up of primary outcomes such as severe neurodevelopmental impairment (sNDI), cerebral palsy (CP), and hearing impairment (HI). Secondary outcomes were moderate-to-severe neurodevelopmental impairment (msNDI), Bayley Scores of Infant Development (BSID-III), and visual impairment. RESULTS: 1,231 patients from 11 comparative studies were included. Quality of evidence was rated low for all outcomes. IVB was associated with a higher risk for sNDI (risk ratio [RR] = 1.25, 95% confidence interval [CI]: [1.01, 1.53], p = 0.04); and CP (RR = 1.40, CI: [1.08, 1.81], p = 0.01) compared to LPC. There was no significant difference between IVB and LPC for msNDI (RR = 1.15, CI: [0.98, 1.35], p = 0.08) and HI (RR = 1.43, CI: [0.86, 2.39], p = 0.17). BSID-III percentile scores were similar between IVB and LPC, with weighted mean differences of 1.51 [CI = -1.25, 4.27], 2.43 [CI = -1.36, 6.22], and 1.97 [CI = -1.06, 5.01] for cognitive, language, and motor domains, respectively (p > 0.05). CONCLUSION: To our knowledge, this is the largest meta-analysis on neurodevelopmental outcomes and the first to rigorously examine IVB monotherapy in ROP treatment. Compared to LPC, there was a marginally increased risk for sNDI and CP with IVB but little or no difference in the risk of msNDI and HI. Further randomized studies are needed to strengthen these findings.

Benchmarking Polyp Segmentation Methods in Narrow-Band Imaging Colonoscopy Images.

In recent years, there has been significant progress in polyp segmentation in white-light imaging (WLI) colonoscopy images, particularly with methods based on deep learning (DL). However, little attention has been paid to the reliability of these methods in narrow-band imaging (NBI) data. NBI improves visibility of blood vessels and helps physicians observe complex polyps more easily than WLI, but NBI images often include polyps with small/flat appearances, background interference, and camouflage properties, making polyp segmentation a challenging task. This paper proposes a new polyp segmentation dataset (PS-NBI2K) consisting of 2,000 NBI colonoscopy images with pixel-wise annotations, and presents benchmarking results and analyses for 24 recently reported DL-based polyp segmentation methods on PS-NBI2K. The results show that existing methods struggle to locate polyps with smaller sizes and stronger interference, and that extracting both local and global features improves performance. There is also a trade-off between effectiveness and efficiency, and most methods cannot achieve the best results in both areas simultaneously. This work highlights potential directions for designing DL-based polyp segmentation methods in NBI colonoscopy images, and the release of PS-NBI2K aims to drive further development in this field.

Toward Multicenter Skin Lesion Classification Using Deep Neural Network With Adaptively Weighted Balance Loss.

Recently, deep neural network-based methods have shown promising advantages in accurately recognizing skin lesions from dermoscopic images. However, most existing works focus more on improving the network framework for better feature representation but ignore the data imbalance issue, limiting their flexibility and accuracy across multiple scenarios in multi-center clinics. Generally, different clinical centers have different data distributions, which presents challenging requirements for the network's flexibility and accuracy. In this paper, we divert the attention from framework improvement to the data imbalance issue and propose a new solution for multi-center skin lesion classification by introducing a novel adaptively weighted balance (AWB) loss to the conventional classification network. Benefiting from AWB, the proposed solution has the following advantages: 1) it is easy to satisfy different practical requirements by only changing the backbone; 2) it is user-friendly with no tuning on hyperparameters; and 3) it adaptively enables small intraclass compactness and pays more attention to the minority class. Extensive experiments demonstrate that, compared with solutions equipped with state-of-the-art loss functions, the proposed solution is more flexible and more competent for tackling the multi-center imbalanced skin lesion classification task with considerable performance on two benchmark datasets. In addition, the proposed solution is proved to be effective in handling the imbalanced gastrointestinal disease classification task and the imbalanced DR grading task. Code is available at https://github.com/Weipeishan2021.

Prediction of Urban Taxi Travel Demand by Using Hybrid Dynamic Graph Convolutional Network Model.

The efficient and accurate prediction of urban travel demand, which is a hot topic in intelligent transportation research, is challenging due to its complicated spatial-temporal dependencies, dynamic nature, and uneven distribution. Most existing forecasting methods merely considered the static spatial dependencies while ignoring the influence of the diversity of dynamic demand patterns and/or uneven distribution. In this paper, we propose a traffic demand forecasting framework of a hybrid dynamic graph convolutional network (HDGCN) model to deeply capture the characteristics of urban travel demand and improve prediction accuracy. In HDGCN, traffic flow similarity graphs are designed according to the dynamic nature of travel demand, and a dynamic graph sequence is generated according to time sequence. Then, the dynamic graph convolution module and the standard graph convolution module are introduced to extract the spatial features from dynamic graphs and static graphs, respectively. Finally, the spatial features of the two components are fused and combined with the gated recurrent unit (GRU) to learn the temporal features. The efficiency and accuracy of the HDGCN model in predicting urban taxi travel demand are verified by using the taxi data from Manhattan, New York City. The modeling and comparison results demonstrate that the HDGCN model can achieve stable and effective prediction for taxi travel demand compared with the state-of-the-art baseline models. The proposed model could be used for the real-time, accurate, and efficient travel demand prediction of urban taxi and other urban transportation systems.

Boundary Constraint Network With Cross Layer Feature Integration for Polyp Segmentation.

Clinically, proper polyp localization in endoscopy images plays a vital role in the follow-up treatment (e.g., surgical planning). Deep convolutional neural networks (CNNs) provide a favoured prospect for automatic polyp segmentation and evade the limitations of visual inspection, e.g., subjectivity and overwork. However, most existing CNNs-based methods often provide unsatisfactory segmentation performance. In this paper, we propose a novel boundary constraint network, namely BCNet, for accurate polyp segmentation. The success of BCNet benefits from integrating cross-level context information and leveraging edge information. Specifically, to avoid the drawbacks caused by simple feature addition or concentration, BCNet applies a cross-layer feature integration strategy (CFIS) in fusing the features of the top-three highest layers, yielding a better performance. CFIS consists of three attention-driven cross-layer feature interaction modules (ACFIMs) and two global feature integration modules (GFIMs). ACFIM adaptively fuses the context information of the top-three highest layers via the self-attention mechanism instead of direct addition or concentration. GFIM integrates the fused information across layers with the guidance from global attention. To obtain accurate boundaries, BCNet introduces a bilateral boundary extraction module that explores the polyp and non-polyp information of the shallow layer collaboratively based on the high-level location information and boundary supervision. Through joint supervision of the polyp area and boundary, BCNet is able to get more accurate polyp masks. Experimental results on three public datasets show that the proposed BCNet outperforms seven state-of-the-art competing methods in terms of both effectiveness and generalization.

Comparing ProKera With Amniotic Membrane Transplantation: Indications, Outcomes, and Costs.

PURPOSE: The aim of this study was to compare the outcomes of ProKera versus amniotic membrane transplantation (AMT) in managing ocular surface disease. METHODS: This study is a retrospective case series of patients who received either ProKera or sutured AMT for ocular surface disease. Patient demographics, treatment indications, retention time, percentage healed area, changes in visual acuity, and costs to the health care system were analyzed. RESULTS: Fourteen patients were identified and analyzed for each group. The main indications for using ProKera and AMT were similar, including corneal ulcer or epithelial defect due to chemical burns, neurotropic state, or herpes zoster keratitis. The average time to dissolution or removal was 24.8 days in the ProKera group, compared with 50.1 days in the AMT group. The average percentage of healed corneal area was 59% for ProKera and 73% for AMT. There was no significant difference between the initial and the final visual acuity within groups and when comparing both groups. In our expense analysis, ProKera had a total cost of 699.00 Canadian dollars (CAD), whereas the cost of suture AMT was 1561.52 CAD. ProKera priced at 11.85 CAD for each percentage healed surface area and at 21.39 CAD for AMT. CONCLUSIONS: ProKera allowed for a faster corneal healing than sutured AMT, although its total healed area was less than the latter. Moreover, ProKera is more cost-effective than AMT, thus reducing financial burden to our health care system.

Extension of the Generalized Hydrodynamics to the Dimensional Crossover Regime.

In an effort to address integrability breaking in cold gas experiments, we extend the integrable hydrodynamics of the Lieb-Liniger model with two additional components representing the population of atoms in the first and second transverse excited states, thus enabling a description of quasi-1D condensates. Collisions between different components are accounted for through the inclusion of a Boltzmann-type collision integral in the hydrodynamic equation. Contrary to standard generalized hydrodynamics, our extended model captures thermalization of the condensate at a rate consistent with experimental observations from a quantum Newton's cradle setup.

Active Event-Triggered Control for Nonlinear Networked Control Systems With Communication Constraints.

In this paper, a novel reference input and hysteresis quantizer-based active event-triggered control (RIHQAETC) scheme is proposed for nonlinear networked control systems with quantizer, networked induced delay, and packet dropout. Different from the traditional methods, such a design method is constructed involving the structure of the hysteresis quantizer. In view of the network induced delay and the potential packet dropout, our RIHQAETC method is designed to actively compensate the negative effects caused by these two issues. The corresponding coder and decoder are also excogitated on account of the potential packet dropout based on the proposed triggering mechanism. Furthermore, the transmission of the important triggering information can be ensured as well as the finite-gain L2 stability performance. It is demonstrated by an example that our RIHQAETC method presents a more balanced updating frequency between the plant and the controller output sides and reduces the number of total triggering.

Overview of the Canadian Clinician Investigator Trainees' research presented at the 2019 CSCI-CITAC Joint Meeting.

The 2019 Annual General Meeting and Young Investigators' Forum of the Canadian Society for Clinical Investigation / Société Canadienne de Recherche Clinique (CSCI/SCRC) and Clinician Investigator Trainee Association of Canada / Association des Cliniciens-Chercheurs en Formation du Canada (CITAC/ACCFC) was held in Banff, Alberta on November 8-10th, 2019. The theme was "Positioning Early Career Investigators for Success: Strategy and Resilience". Lectures and workshops provided knowledge and tools to facilitate the attendees' development as clinician investigators. Dr. Jason Berman (President of CSCI/SCRC), Elina Cook (President of CITAC/ACCFC) and Drs. Doreen Rabi and Zelma Kiss (University of Calgary Organizing Co-Chairs) gave opening presentations. The keynote speakers were Dr. William Foulkes (McGill University) (Distinguished Scientist Award winner) and Dr. Andrés Finzi (Université de Montréal) (Joe Doupe Young Investigator Award winner). Dr. Robert Bortolussi (Dalhousie University) received the Distinguished Service Award for his work as the Editor-in-Chief of Clinical and Investigative Medicine and for being instrumental in the development of the Canadian Child Health Clinician Scientist Program. This meeting was the first to host a panel discussion with Drs. Stephen Robbins and Marcello Tonelli from the Canadian Institutes of Health Research. Workshops on communication, career planning and work-life balance were hosted by André Picard and Drs. Todd Anderson, Karen Tang, William Ghali, May Lynn Quan, Alicia Polachek and Shannon Ruzycki. The AGM showcased 90 presentations from clinician investigator trainees from across Canada. Most of the abstracts are summarized in this review. Eight outstanding abstracts were selected for oral presentation at the President's Forum.

Genome Data Uncover Conservation Status, Historical Relatedness and Candidate Genes Under Selection in Chinese Indigenous Pigs in the Taihu Lake Region.

Chinese indigenous pig breeds in the Taihu Lake (TL) region of Eastern China are well documented by their exceptional prolificacy. There are seven breeds in this region including Meishan (MS), Erhualian (EHL), Jiaxing Black (JXB), Fengjing (FJ), Shawutou (SWT), Mi (MI), and Hongdenglong (HDL). At present, these breeds are facing a great threat of population decline, inbreeding depression and lineage admixture since Western commercial pigs have dominated in Chinese pig industry. To provide better conservation strategies and identify candidate genes under selection for these breeds, we explored genome-wide single nucleotide polymorphism (SNP) markers to uncover genetic variability and relatedness, population structure, historical admixture and genomic signature of selection of 440 pigs representing the most comprehensive lineages of these breeds in TL region in a context of 1228 pigs from 45 Eurasian breeds. We showed that these breeds were more closely related to each other as compared to other Eurasian breeds, defining one of the main ancestral lineages of Chinese indigenous pigs. These breeds can be divided into two subgroups, one including JXB and FJ, and the other comprising of EHL, MI, HDL, MS, and SWT. In addition, HDL was highly inbred whereas EHL and MS had more abundant genetic diversity owing to their multiple conservation populations. Moreover, we identified a list of candidate genes under selection for body size and prolificacy. Our results would benefit the conservation of these valuable breeds and improve our understanding of the genetic mechanisms of body size and fecundity in pigs.

Novel Anti-Interleukin-1ß Therapy Preserves Retinal Integrity: A Longitudinal Investigation Using OCT Imaging and Automated Retinal Segmentation in Small Rodents.

Retinopathy of prematurity (ROP) is the leading cause of blindness in neonates. Inflammation, in particular interleukin-1ß (IL-1ß), is increased in early stages of the disorder, and contributes to inner and outer retinal vasoobliteration in the oxygen-induced retinopathy (OIR) model of ROP. A small peptide antagonist of IL-1 receptor, composed of the amino acid sequence, rytvela, has been shown to exert beneficial anti-inflammatory effects without compromising immunovigilance-related NF-κB in reproductive tissues. We conducted a longitudinal study to determine the efficacy of "rytvela" in preserving the integrity of the retina in OIR model, using optical coherence tomography (OCT) which provides high-resolution cross-sectional imaging of ocular structures in vivo. Sprague-Dawley rats subjected to OIR and treated or not with "rytvela" were compared to IL-1 receptor antagonist (Kineret). OCT imaging and custom automated segmentation algorithm used to measure retinal thickness (RT) were obtained at P14 and P30; gold-standard immunohistochemistry (IHC) was used to confirm retinal anatomical changes. OCT revealed significant retinal thinning in untreated animals by P30, confirmed by IHC; these changes were coherently associated with increased apoptosis. Both rytvela and Kineret subsided apoptosis and preserved RT. As anticipated, Kineret diminished both SAPK/JNK and NF-κB axes, whereas rytvela selectively abated the former which resulted in preserved monocyte phagocytic function. Altogether, OCT imaging with automated segmentation is a reliable non-invasive approach to study longitudinally retinal pathology in small animal models of retinopathy.

The Inability of the Choroid to Revascularize in Oxygen-Induced Retinopathy Results from Increased p53/miR-Let-7b Activity.

Retinopathy of prematurity (ROP) is characterized by an initial retinal avascularization, followed by pathologic neovascularization. Recently, choroidal thinning has also been detected in children formerly diagnosed with ROP; a similar sustained choroidal thinning is observed in ROP models. But the mechanism underlying the lack of choroidal revascularization remains unclear and was investigated in an oxygen-induced retinopathy (OIR) model. In OIR, evidence of senescence was detected, preceded by oxidative stress in the choroid and the retinal pigment epithelium. This was associated with a global reduction of proangiogenic factors, including insulin-like growth factor 1 receptor (Igf1R). Coincidentally, tumor suppressor p53 was highly expressed in the OIR retinae. Curtailing p53 activity resulted in reversal of senescence, normalization of Igf1r expression, and preservation of choroidal integrity. OIR-induced down-regulation of Igf1r was mediated at least partly by miR-let-7b as i) let-7b expression was augmented throughout and beyond the period of oxygen exposure, ii) let-7b directly targeted Igf1r mRNA, and iii) p53 knock-down blunted let-7b expression, restored Igf1r expression, and elicited choroidal revascularization. Finally, restoration of Igf1r expression rescued choroid thickness. Altogether, this study uncovers a significant mechanism for defective choroidal revascularization in OIR, revealing a new role for p53/let-7b/IGF-1R axis in the retina. Future investigations on this (and connected) pathway could further our understanding of other degenerative choroidopathies, such as geographic atrophy.