No abstract

In this paper, we present a study on learning visual recognition models from large scale noisy web data. We build a new database called WebVision, which contains more than 2.4 million web images crawled from the Internet by using queries generated from the 1, 000 semantic concepts of the ILSVRC 2012 benchmark. Meta information along with those web images (e.g., title, description, tags, etc.) are also crawled. A validation set and test set containing human annotated images are also provided to facilitate algorithmic development. Based on our new database, we obtain a few interesting observations: 1) the noisy web images are sufficient for training a good deep CNN model for visual recognition; 2) the model learnt from our WebVision database exhibits comparable or even better generalization ability than the one trained from the ILSVRC 2012 dataset when being transferred to new datasets and tasks; 3) a domain adaptation issue (a.k.a., dataset bias) is observed, which means the dataset can be used as the largest benchmark dataset for visual domain adaptation. Our new WebVision database and relevant studies in this work would benefit the advance of learning state-of-the-art visual models with minimum supervision based on web data.

In this paper, we propose to use golden angle modulation (GAM) points to construct codebooks for uplink and downlink sparse code multiple access (SCMA) systems. We provide two categories of codebooks with one and two optimization parameters respectively. The advantages of the proposed design method are twofold: 1) the number of optimization variables is independent of codebook and system parameters; 2) it is simple to implement. In the downlink, we use GAM points to build a multidimensional mother constellation for SCMA codebooks, while in the uplink GAM points are directly mapped to user codebooks. The proposed codebooks exhibit good performance with low peak to average power ratio (PAPR) compared to the codebooks proposed in the literature based on constellation rotation and interleaving. Index Terms-Sparse code multiple access, codebook design, golden angle modulation, mapping, PAPR reduction. NOMA techniques can be classed under three categories: power-domain NOMA (PD-NOMA) [1], code-domain NOMA (CD-NOMA) [2], [3], [4] and a combination of PD-NOMA and CD-NOMA called power domain sparse code multiple access [5]. Low-density signature (LDS) [2], [3] is an efficient CD-NOMA that allows operating at overloaded conditions with performance close to single user case with affordable complexity. Due to the low-density property of signatures, only few users will use the same resource in order to reduce the interference and the data symbol for each user will be spread on a few number of resources. Like low-density parity-check coding, the low-density characteristic of LDS signatures allows to use the message passing algorithm (MPA) for multiuser

Background: Inflammation-related parameters have been revealed to have prognostic value in multiple caners. However, the significance of some inflammation-related parameters, including the peripheral blood neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), lymphocyte-to-monocyte ratio (LMR) and prognostic nutritional index (PNI), remains controversial in T1-2 rectal cancer (RC). Methods: Clinical data of 154 T1-2 RC patients were retrospectively reviewed. The cutoff values for NLR, PLR, LMR, and PNI were determined by receiver operating characteristic curves. The relationships of these parameters with postoperative morbidities and prognosis were statistically analysed. Results: The optimal cutoff values for preoperative NLR, PLR, LMR and PNI were 2.8, 140.0, 3.9, and 47.1, respectively. Significant but heterogeneous associations were found between NLR, PLR, LMR and PNI and clinicopathological factors. In addition, high NLR, high PLR, and low PNI were correlated with an increased postoperative morbidity rate. Patients with high NLR/PLR or low LMR/PNI had lower OS and DFS rates. On multivariate analysis, only high NLR was identified as an independent risk factor for poor DFS. Conclusions: NLR, PLR, and PNI are valuable factors for predicting postoperative complications in T1-2 RC patients. A preoperative NLR of more than 2.8 is an independent prognostic factor for poor DFS in T1-2 RC patients.

Two guidelines are reproducible and reliable in AC diagnosis but different in severity grading. TG13 are more practical for immediate severity grading, enabling planning treatment upon admission. Intrahepatic obstruction is a new candidate predictor of 30-day mortality for further assessment.

Purpose To assess whether a generative adversarial network (GAN) could synthesize realistic optical coherence tomography (OCT) images that satisfactorily serve as the educational images for retinal specialists, and the training datasets for the classification of various retinal disorders using deep learning (DL). Methods The GANs architecture was adopted to synthesize high-resolution OCT images trained on a publicly available OCT dataset, including urgent referrals (37,206 OCT images from eyes with choroidal neovascularization, and 11,349 OCT images from eyes with diabetic macular edema) and nonurgent referrals (8617 OCT images from eyes with drusen, and 51,140 OCT images from normal eyes). Four hundred real and synthetic OCT images were evaluated by two retinal specialists (with over 10 years of clinical retinal experience) to assess image quality. We further trained two DL models on either real or synthetic datasets and compared the performance of urgent versus nonurgent referrals diagnosis tested on a local (1000 images from the public dataset) and clinical validation dataset (278 images from Shanghai Shibei Hospital). Results The image quality of real versus synthetic OCT images was similar as assessed by two retinal specialists. The accuracy of discrimination of real versus synthetic OCT images was 59.50% for retinal specialist 1 and 53.67% for retinal specialist 2. For the local dataset, the DL model trained on real (DL_Model_R) and synthetic OCT images (DL_Model_S) had an area under the curve (AUC) of 0.99, and 0.98, respectively. For the clinical dataset, the AUC was 0.94 for DL_Model_R and 0.90 for DL_Model_S. Conclusions The GAN synthetic OCT images can be used by clinicians for educational purposes and for developing DL algorithms. Translational Relevance The medical image synthesis based on GANs is promising in humans and machines to fulfill clinical tasks.

No abstract

No abstract