OutSingle: a novel method of detecting and injecting outliers in RNA-Seq count data using the optimal hard threshold for singular values.

MOTIVATION: Finding outliers in RNA-sequencing (RNA-Seq) gene expression (GE) can help in identifying genes that are aberrant and cause Mendelian disorders. Recently developed models for this task rely on modeling RNA-Seq GE data using the negative binomial distribution (NBD). However, some of those models either rely on procedures for inferring NBD's parameters in a nonbiased way that are computationally demanding and thus make confounder control challenging, while others rely on less computationally demanding but biased procedures and convoluted confounder control approaches that hinder interpretability. RESULTS: In this article, we present OutSingle (Outlier detection using Singular Value Decomposition), an almost instantaneous way of detecting outliers in RNA-Seq GE data. It uses a simple log-normal approach for count modeling. For confounder control, it uses the recently discovered optimal hard threshold (OHT) method for noise detection, which itself is based on singular value decomposition (SVD). Due to its SVD/OHT utilization, OutSingle's model is straightforward to understand and interpret. We then show that our novel method, when used on RNA-Seq GE data with real biological outliers masked by confounders, outcompetes the previous state-of-the-art model based on an ad hoc denoising autoencoder. Additionally, OutSingle can be used to inject artificial outliers masked by confounders, which is difficult to achieve with previous approaches. We describe a way of using OutSingle for outlier injection and proceed to show how OutSingle outperforms its competition on 16 out of 18 datasets that were generated from three real datasets using OutSingle's injection procedure with different outlier types and magnitudes. Our methods are applicable to other types of similar problems involving finding outliers in matrices under the presence of confounders. AVAILABILITY AND IMPLEMENTATION: The code for OutSingle is available at https://github.com/esalkovic/outsingle.

Venetoclax in biomarker-selected multiple myeloma patients: Impact of exposure on clinical efficacy and safety.

Venetoclax, a potent BCL-2 inhibitor, is currently under development for treatment of t(11;14) Multiple myeloma (MM). The objective of this research was to investigate the exposure-response relationships of venetoclax for a phase 1/2 study evaluating venetoclax monotherapy or in combination with dexamethasone in relapsed or refractory MM. A total of 117 patients receiving venetoclax at 300, 600, 800, 900, or 1200 mg were included in the analysis. The impact of venetoclax exposures on efficacy (objective response rate [ORR], progression-free survival [PFS] and overall survival [OS]) as well as safety (treatment-emergent adverse effects (grade ≥3) of neutropenia, infection, and any grade of serious treatment-emergent adverse effects) was evaluated. In the t(11;14)-positive subpopulation, venetoclax exposure relationships to PFS and OS indicated a trend of longer PFS and OS with higher exposures. Moreover, logistic regression analyses for clinical response (ORR and ≥VGPR rate) demonstrated a statistically significant (p < 0.05) relationship with exposure. Evaluation of the exposure-safety relationships demonstrated a lack of a relationship between venetoclax exposures (AUCavg ) and grade ≥3 infections, grade ≥3 neutropenia, grade ≥3 treatment-emergent adverse events or any grade serious treatment-emergent adverse events. These findings support further study of venetoclax at 800 mg QD dose in combination with dexamethasone in the t(11;14)-positive patient population where increased efficacy was observed without an increase in safety events.Clinical Trial: NCT01794520 registered 20 February 2013.

Advances in endoscopic management of colonic diverticular bleeding.

PURPOSE OF REVIEW: This review is focused on diagnostic and management strategies for colonic diverticular bleeding (CDB). It aims to present the current state of the field, highlighting the available techniques, and emphasizing findings that influence the choice of therapy. RECENT FINDINGS: Recent guidelines recommend nonurgent colonoscopy (>24 h) for CDB. However, factors such as a shock index ≥1, which may warrant an urgent colonoscopy, remain under investigation.The standard approach to detecting the source of CDB requires a water-jet scope equipped with a cap. Innovative diagnostic techniques, such as the long-cap and tapered-cap, have proven effective in identifying stigmata of recent hemorrhage (SRH). Furthermore, the water or gel immersion methods may aid in managing massive hemorrhage by improving the visualization and stabilization of the bleeding site for subsequent intervention. Innovations in endoscopic hemostasis have significantly improved the management of CDB. New therapeutic methods such as endoscopic band ligation and direct clipping have substantially diminished the incidence of recurrent bleeding. Recent reports also have demonstrated the efficacy of cutting-edge techniques such as over-the-scope clips, which have significantly improved outcomes in complex cases that have historically necessitated surgical intervention. SUMMARY: Currently available endoscopic diagnostic and hemostatic methods for CDB have evolved with improved outcomes. Further research is necessary to refine the criteria for urgent colonoscopy and to confirm the effectiveness of new endoscopic hemostasis techniques.

HbA1c or fructosamine on evaluating glucose intolerance in children with beta- thalassemia.

BACKGROUND: Beta-thalassemia major (ß-TM) patients are more likely to experience blood glucose intolerance and to date; the blood markers that could evaluate this are debatable. So, this study aimed to assess the roles of glycated hemoglobin A1c (HbA1c) and fructosamine in evaluating glucose intolerance in children with ß-TM and figuring out role of insulin resistance in these patients. METHODS: One hundred children diagnosed with ß-TM and 100 age and sex-matched controls were enrolled. Fasting plasma glucose (FPG), 2-h post-prandial blood glucose (2-h PG), HbA1c, fructosamine, fasting insulin level (FINS), insulin resistance index (HOMA-IR), and insulin sensitivity index (HOMA-IS) were evaluated. RESULTS: FPG and 2-h PG revealed glucose intolerance in 51 patients (51%), 19 of them had diabetes mellitus. HbA1c, fructosamine, FINS, and HOMA-IR showed a high statistically significant increase in patients compared to controls, (P < 0.001). Results revealed fructosamine was more specific in detecting prediabetes state and more sensitive in identifying diabetes mellitus in our patients when compared to HbA1c. CONCLUSIONS: Despite controversies on HbA1c in children with ß-TM, it is still valuable in glucose intolerance detection. Fructosamine showed more sensitivity and specificity. Furthermore, insulin resistance was prevalent in children with ß-TM highlighting the necessity of regular glycemic state evaluation. IMPACT: Glucose intolerance is a common complication in beta thalassemia patients. Conflicting data was reported about the role of HbA1c and fructosamine in evaluating glucose intolerance in thalassemic patients. Fructosamine does not yet have a threshold that may be used to distinguish between patients who have diabetes mellitus and those who do not. Fructosamine was more specific in detecting blood glucose intolerance compared to HbA1c and was more sensitive for diagnosing diabetes mellitus. Insulin resistance was common in patients with beta-thalassemia and often present before the onset of overt diabetes.

Venetoclax pharmacokinetics in participants with end-stage renal disease undergoing haemodialysis.

AIMS: Renal insufficiency is a common comorbidity in patients with haematological malignancies. This study aimed to assess how end-stage renal disease (ESRD) might affect the pharmacokinetics of venetoclax, a Bcl-2 inhibitor, in participants with ESRD undergoing haemodialysis. METHODS: Venetoclax was administered as a single 100-mg dose to 6 female participants with ESRD (estimated glomerular filtration rate <15 mL/min) both prior to haemodialysis and between haemodialysis days and 7 healthy female participants with normal renal function (estimated glomerular filtration rate >90 mL/min). Intensive pharmacokinetic and protein binding samples were collected from all participants. Arterial and venous samples were collected from ESRD participants during haemodialysis to assess the effect of haemodialysis on venetoclax pharmacokinetics. Pharmacokinetic parameters were estimated using noncompartmental methods. RESULTS: There was no difference in plasma venetoclax concentrations between arterial and venous samples, suggesting that haemodialysis did not affect the pharmacokinetics of venetoclax. The fraction unbound (fu ) of venetoclax was ~2-fold higher for participants with ESRD compared to participants with normal renal function. The unbound maximum plasma concentration and area under the plasma concentration-time curve from time 0 to 48 h were comparable between ESRD and normal function groups. The mean half-life ranged from 10.4 to 12.2 h across groups, demonstrating that ESRD did not affect the half-life of venetoclax. No new safety signals were observed during this study. CONCLUSION: ESRD and dialysis do not alter unbound venetoclax plasma concentrations. No pharmacokinetics driven dose adjustment is needed for patients with renal insufficiency.

Efficacy of venetoclax plus rituximab for relapsed CLL: 5-year follow-up of continuous or limited- duration therapy.

We report long-term follow-up of the phase 1b study of venetoclax and rituximab (VenR) in patients with relapsed chronic lymphocytic leukemia (CLL), including outcomes with continuous or limited-duration therapy. Patients received venetoclax daily (200-600 mg) and rituximab over 6 months and then received venetoclax monotherapy. Patients achieving complete response (CR), CR with incomplete marrow recovery (CRi), or undetectable minimal residual disease (uMRD) assessed by flow cytometry (<10-4 cutoff) were allowed, but not required, to discontinue therapy, while remaining in the study and could be retreated with VenR upon progression. Median follow-up for all patients (N = 49) was 5.3 years. Five-year rates (95% CI) for overall survival, progression-free survival, and duration of response were 86% (72-94), 56% (40-70), and 58% (40-73), respectively. Of the 33 deep responders (CR/CRi or uMRD), 14 remained on venetoclax monotherapy (continuous therapy), and 19 stopped venetoclax therapy (limited-duration therapy) after a median of 1.4 years. Five-year estimates of ongoing response were similar between continuous (71%; 95% CI, 39-88) or limited-duration therapy (79% [49-93]). Six of 19 patients in the latter group had subsequent disease progression, all >2 years off venetoclax (range, 2.1-6.4). Four patients were retreated with VenR, with partial responses observed in the 3 evaluable to date. VenR induced deep responses that were highly durable with either continuous or limited-duration therapy. Retreatment with VenR induced responses in patients with CLL progression after discontinuing therapy. Continuous exposure to venetoclax in deep responders does not appear to provide incremental benefit.

Insights into the Role of Erythrina corallodendron L. in Alzheimer's Disease: in Vitro and in Silico Approach.

Alzheimer's disease (AD) is a major health problem. Cholinergic transmission is greatly affected in AD. Phytochemical investigation of the alkaloid rich fraction (AF) of Erythrina corallodendron L leaves resulted in isolation of five known alkaloids: erysodine, erythrinine, 8-oxoerythrinine, erysovine N-oxide and erythrinine N-oxide. In this study, eysovine N-oxide was reported for the second time in nature. AF was assayed for cholinesterase inhibition at the concentration of 100 µg mL-1 . AF showed a higher percent inhibition for butyrylcholinesterase enzyme (BuChE) (83.28 %) compared to acetylcholinesterase enzyme (AChE) (64.64 %). The isolated alkaloids were also assayed for their anti-BuChE effect. In-silico docking study was done for the isolated compounds at the binding sites of AChE and BuChE to determine their binding pattern and interactions, also molecular dynamics were estimated for the compound displaying the best fit for AChE and BuChE. In addition, ADME parameters and toxicity were predicted for the isolated alkaloids compared to donepezil.

Light Field Image Super-Resolution Using Deep Residual Networks on Lenslet Images.

Due to its widespread usage in many applications, numerous deep learning algorithms have been proposed to overcome Light Field's trade-off (LF). The sensor's low resolution limits angular and spatial resolution, which causes this trade-off. The proposed method should be able to model the non-local properties of the 4D LF data fully to mitigate this problem. Therefore, this paper proposes a different approach to increase spatial and angular information interaction for LF image super-resolution (SR). We achieved this by processing the LF Sub-Aperture Images (SAI) independently to extract the spatial information and the LF Macro-Pixel Image (MPI) to extract the angular information. The MPI or Lenslet LF image is characterized by its ability to integrate more complementary information between different viewpoints (SAIs). In particular, we extract initial features and then process MAI and SAIs alternately to incorporate angular and spatial information. Finally, the interacted features are added to the initial extracted features to reconstruct the final output. We trained the proposed network to minimize the sum of absolute errors between low-resolution (LR) input and high-resolution (HR) output images. Experimental results prove the high performance of our proposed method over the state-of-the-art methods on LFSR for small baseline LF images.

Effect of Dietary Incorporation of Yellow Mealworm as a Partial Fishmeal Replacer on Growth, Metabolism, and Intestinal Histomorphology in Juvenile Meagre (Argyrosomus regius).

Efforts have been made to find alternatives to fish meal (FM), as the sustainability of aquaculture depends on it. Insect meal (IM) is a potential candidate to partially replace FM, being more sustainable and economically viable. In this experimental trial, three diets were tested with different yellow mealworm incorporation: a control diet with no IM, a diet with an inclusion of 10% IM (Ins10), and a diet with an incorporation of 20% IM (Ins20). The diets were tested on 10.5 g meagre for 47 days. The results showed that an IM inclusion higher than 10% affected both growth (2.6 vs. 2.2) and FCR (1.5 vs. 1.9) of meagre juveniles. However, this reduction in growth did not result from lower protein retention or changes in muscle fibre area or density. Little differences were observed in the activity of pancreatic and intestinal enzymes except for aminopeptidase total activity which was higher in the control and Ins10 compared to Ins20 (3847 vs. 3540 mU/mg protein), suggesting no limitations in protein synthesis. Also, the alkaline phosphatase intestinal maturation index was higher in the control group compared to the IM groups (437 vs. 296). On the contrary, several differences were also found in the proteolytic activity in the hepatic and muscle tissues of meagre juveniles fed the Ins10 diet. The inclusion of IM had no impact on intestine histomorphology but changes were detected in the enterocytes of fish from control and Ins10 which showed hypervacuolization and nucleus misplacement compared to the Ins20 treatment. Nevertheless, a higher percentage of Vibrionaceae was recorded for meagre fed on the Ins20 diet. Since no signs of inflammation were observed in the distal intestine, this suggests IM incorporation could have had an important impact on intestinal health due to its antimicrobial properties. This is supported by an increase in the haematocrit in the treatments where IM was added (20 to 25%). In conclusion, incorporations of IM at percentages up to 10% do not seem to have a negative impact on meagre performance at this age but can enhance the fish immune system and protection against intestinal inflammation.

Venetoclax exposure-efficacy and exposure-safety relationships in patients with treatment-naïve acute myeloid leukemia who are ineligible for intensive chemotherapy.

This study evaluated venetoclax population pharmacokinetics (popPK) in patients with treatment-naïve acute myeloid leukemia and assessed the relationship between venetoclax exposure and clinical response for venetoclax in combination with either a hypomethylating agent (HMA) or low-dose cytarabine (LDAC). A total of 771 patients who received venetoclax from 5 Phase 1-3 studies were included in the popPK model. Exposure-response analyses included data from 575 patients for venetoclax/placebo plus HMA and 279 patients for venetoclax/placebo plus LDAC. The popPK model successfully characterized venetoclax plasma concentrations over time and confirmed venetoclax exposure did not vary significantly with age, weight, sex, mild to moderate hepatic impairment, or mild to severe renal impairment. Asian patients had 67% higher mean relative bioavailability than non-Asian patients, however the range of exposures in Asian patients was similar to non-Asian patients. For all efficacy endpoints with both treatment combinations, efficacy was higher in the venetoclax treatment groups compared with the respective control arm of placebo plus azacitidine or LDAC. Within patients who received venetoclax, no significant exposure-efficacy relationships were identified for either treatment combination, indicating that the beneficial effects of venetoclax were already maximized in the dose ranges studied. There was no apparent effect of venetoclax exposure on treatment-emergent Grade ≥3 thrombocytopenia or infections for either combination. Rates of treatment-emergent Grade ≥3 neutropenia were higher in the venetoclax treatment arms compared with the respective control arms; however, within patients who received venetoclax, there was only a shallow relationship or no apparent relationship with venetoclax exposure for venetoclax plus HMA or LDAC, respectively. Along with the efficacy and safety data previously published, the exposure-response analyses support the venetoclax dose regimens of 400 mg once daily (QD) plus HMA and 600 mg QD plus LDAC in treatment-naïve AML patients who are ineligible for intensive chemotherapy.

Anticoagulation and bleeding in the cancer patient.

Cancer patients have an increased risk of bleeding compared to non-cancer patients with anticoagulant therapy. A bleeding risk assessment before initiation of anticoagulation is recommended. Currently low molecular weight heparin (LMWH) and direct oral anticoagulants (DOACs) are the mainstays of treatment for cancer-associated venous thromboembolism (VTE). Since DOACs are administered orally, they offer some convenience and ease of administration; however, LMWH may be preferred in certain cancers. Given the prevalence of anticoagulant therapies in cancer patients, clinical providers must be able to recognize potentially critical bleeding sites and modalities to reverse major hemorrhage. Reversal agents or antidotes to bleeding may be required when bleeding is persistent or life-threatening. These include vitamin K, fresh frozen plasma (FFP), protamine, prothrombin complex concentrate (PCC) or andexanet alfa, and idarucizumab. Inferior vena cava (IVC) filter insertion can be also considered in those with major bleeding. Evidence for timing and need for re-initiation of anticoagulant therapy after a major bleeding remains sparse, but a multi-disciplinary approach and shared decision-making can be implemented in the interim.

Long-term effects of smoking on serum concentrations of oxidative stress biomarkers: Results of a large, population-based cohort study.

INTRODUCTION: Cross-sectional studies have consistently shown an association between current smoking and oxidative stress biomarkers. However, no longitudinal studies have been performed so far. METHODS: The oxidative stress biomarkers "total thiol groups of serum proteins" (TTP), and "derivatives of reactive oxygen metabolites" (D-ROM) were measured in serum samples of 3835 participants of a population-based, German cohort study of older adults (age: 60-84 years) with repeated measurements for 2834 participants three years later. Multivariable linear regression models were applied and ß-coefficients with 95% confidence intervals were obtained. RESULTS: In both cross-sectional and longitudinal analysis, current smoking was statistically significantly associated with increased D-ROM levels, and a dose-response relationship between the amount of daily tobacco consumption and the D-ROM concentrations was observed that plateaued at ≥15 g of tobacco consumption per day. Former smoking was also associated with D-ROM levels. Only former smokers who quitted smoking more than 10 years ago had no increased D-ROM levels compared to never smokers. There was neither a cross-sectional nor longitudinal association between any of the smoking variables and TTP levels. CONCLUSION: This large population-based cohort of older German adults suggests that smoking has long-term effects on the oxidative stress burden. The further increasing D-ROM levels of current smokers at an older age and the observation that it may take more than 10 years until the redox balance is restored are solid arguments for quitting smoking as soon as possible at any age.

Mapping the Distribution of Special Structured Pulmonary Vasculatures in the Dromedary Camel (Camelus dromedarius): Histological and Histochemical Study.

Dromedary camel (Camelus dromedarius) is adapted to survive the harsh environments. It has some key adaptation peculiarities in various organs. In this study, we aimed to map the distribution pattern of unique regulatory devices along the course of the pulmonary vessels using histological and histochemical analyses. Arteries with variable wall thickness and spirally oriented course were recorded within the adventitia of the main pulmonary artery. Throttle arteries and glomus bolsters were found within the wall of the lobar pulmonary artery. The bronchial artery was located within the wall of all bronchi reaching the subsegmental branches and it had elastic longitudinal muscular intima bolsters. Arteries with double muscular media were demonstrated in the pulmonary pleura. These bolsters are suggested to play a complicated role that allows for hemodynamic, humeral, and thermoregulatory activities. The lumen of some subsegmental pulmonary veins revealed occasional constrictions arising from the corresponding muscular pad-like protrusions of the tunica media. These veins may possess occlusive or constrictive mechanisms and their obstruction induces engorgement of the associated capillary bed in addition to restricting venous outflow. Collectively, these data strongly recommend a crucial role for the special regulatory devices in preserving the camel pulmonary function in the harsh desert environment.

DTS-Net: Depth-to-Space Networks for Fast and Accurate Semantic Object Segmentation.

We propose Depth-to-Space Net (DTS-Net), an effective technique for semantic segmentation using the efficient sub-pixel convolutional neural network. This technique is inspired by depth-to-space (DTS) image reconstruction, which was originally used for image and video super-resolution tasks, combined with a mask enhancement filtration technique based on multi-label classification, namely, Nearest Label Filtration. In the proposed technique, we employ depth-wise separable convolution-based architectures. We propose both a deep network, that is, DTS-Net, and a lightweight network, DTS-Net-Lite, for real-time semantic segmentation; these networks employ Xception and MobileNetV2 architectures as the feature extractors, respectively. In addition, we explore the joint semantic segmentation and depth estimation task and demonstrate that the proposed technique can efficiently perform both tasks simultaneously, outperforming state-of-art (SOTA) methods. We train and evaluate the performance of the proposed method on the PASCAL VOC2012, NYUV2, and CITYSCAPES benchmarks. Hence, we obtain high mean intersection over union (mIOU) and mean pixel accuracy (Pix.acc.) values using simple and lightweight convolutional neural network architectures of the developed networks. Notably, the proposed method outperforms SOTA methods that depend on encoder-decoder architectures, although our implementation and computations are far simpler.

Exploration of Semantic Label Decomposition and Dataset Size in Semantic Indoor Scenes Synthesis via Optimized Residual Generative Adversarial Networks.

In this paper, we revisit the paired image-to-image translation using the conditional generative adversarial network, the so-called "Pix2Pix", and propose efficient optimization techniques for the architecture and the training method to maximize the architecture's performance to boost the realism of the generated images. We propose a generative adversarial network-based technique to create new artificial indoor scenes using a user-defined semantic segmentation map as an input to define the location, shape, and category of each object in the scene, exactly similar to Pix2Pix. We train different residual connections-based architectures of the generator and discriminator on the NYU depth-v2 dataset and a selected indoor subset from the ADE20K dataset, showing that the proposed models have fewer parameters, less computational complexity, and can generate better quality images than the state of the art methods following the same technique to generate realistic indoor images. We also prove that using extra specific labels and more training samples increases the quality of the generated images; however, the proposed residual connections-based models can learn better from small datasets (i.e., NYU depth-v2) and can improve the realism of the generated images in training on bigger datasets (i.e., ADE20K indoor subset) in comparison to Pix2Pix. The proposed method achieves an LPIPS value of 0.505 and an FID value of 81.067, generating better quality images than that produced by Pix2Pix and other recent paired Image-to-image translation methods and outperforming them in terms of LPIPS and FID.

LEOD-Net: Learning Line-Encoded Bounding Boxes for Real-Time Object Detection.

This paper proposes a learnable line encoding technique for bounding boxes commonly used in the object detection task. A bounding box is simply encoded using two main points: the top-left corner and the bottom-right corner of the bounding box; then, a lightweight convolutional neural network (CNN) is employed to learn the lines and propose high-resolution line masks for each category of classes using a pixel-shuffle operation. Post-processing is applied to the predicted line masks to filtrate them and estimate clear lines based on a progressive probabilistic Hough transform. The proposed method was trained and evaluated on two common object detection benchmarks: Pascal VOC2007 and MS-COCO2017. The proposed model attains high mean average precision (mAP) values (78.8% for VOC2007 and 48.1% for COCO2017) while processing each frame in a few milliseconds (37 ms for PASCAL VOC and 47 ms for COCO). The strength of the proposed method lies in its simplicity and ease of implementation unlike the recent state-of-the-art methods in object detection, which include complex processing pipelines.

RT-ViT: Real-Time Monocular Depth Estimation Using Lightweight Vision Transformers.

The latest research in computer vision highlighted the effectiveness of the vision transformers (ViT) in performing several computer vision tasks; they can efficiently understand and process the image globally unlike the convolution which processes the image locally. ViTs outperform the convolutional neural networks in terms of accuracy in many computer vision tasks but the speed of ViTs is still an issue, due to the excessive use of the transformer layers that include many fully connected layers. Therefore, we propose a real-time ViT-based monocular depth estimation (depth estimation from single RGB image) method with encoder-decoder architectures for indoor and outdoor scenes. This main architecture of the proposed method consists of a vision transformer encoder and a convolutional neural network decoder. We started by training the base vision transformer (ViT-b16) with 12 transformer layers then we reduced the transformer layers to six layers, namely ViT-s16 (the Small ViT) and four layers, namely ViT-t16 (the Tiny ViT) to obtain real-time processing. We also try four different configurations of the CNN decoder network. The proposed architectures can learn the task of depth estimation efficiently and can produce more accurate depth predictions than the fully convolutional-based methods taking advantage of the multi-head self-attention module. We train the proposed encoder-decoder architecture end-to-end on the challenging NYU-depthV2 and CITYSCAPES benchmarks then we evaluate the trained models on the validation and test sets of the same benchmarks showing that it outperforms many state-of-the-art methods on depth estimation while performing the task in real-time (∼20 fps). We also present a fast 3D reconstruction (∼17 fps) experiment based on the depth estimated from our method which is considered a real-world application of our method.

RCA-LF: Dense Light Field Reconstruction Using Residual Channel Attention Networks.

Dense multi-view image reconstruction has played an active role in research for a long time and interest has recently increased. Multi-view images can solve many problems and enhance the efficiency of many applications. This paper presents a more specific solution for reconstructing high-density light field (LF) images. We present this solution for images captured by Lytro Illum cameras to solve the implicit problem related to the discrepancy between angular and spatial resolution resulting from poor sensor resolution. We introduce the residual channel attention light field (RCA-LF) structure to solve different LF reconstruction tasks. In our approach, view images are grouped in one stack where epipolar information is available. We use 2D convolution layers to process and extract features from the stacked view images. Our method adopts the channel attention mechanism to learn the relation between different views and give higher weight to the most important features, restoring more texture details. Finally, experimental results indicate that the proposed model outperforms earlier state-of-the-art methods for visual and numerical evaluation.

DTS-Depth: Real-Time Single-Image Depth Estimation Using Depth-to-Space Image Construction.

As most of the recent high-resolution depth-estimation algorithms are computationally so expensive that they cannot work in real time, the common solution is using a low-resolution input image to reduce the computational complexity. We propose a different approach, an efficient and real-time convolutional neural network-based depth-estimation algorithm using a single high-resolution image as the input. The proposed method efficiently constructs a high-resolution depth map using a small encoding architecture and eliminates the need for a decoder, which is typically used in the encoder-decoder architectures employed for depth estimation. The proposed algorithm adopts a modified MobileNetV2 architecture, which is a lightweight architecture, to estimate the depth information through the depth-to-space image construction, which is generally employed in image super-resolution. As a result, it realizes fast frame processing and can predict a high-accuracy depth in real time. We train and test our method on the challenging KITTI, Cityscapes, and NYUV2 depth datasets. The proposed method achieves low relative absolute error (0.028 for KITTI, 0.167 for CITYSCAPES, and 0.069 for NYUV2) while working at speed reaching 48 frames per second on a GPU and 20 frames per second on a CPU for high-resolution test images. We compare our method with the state-of-the-art methods on depth estimation, showing that our method outperforms those methods. However, the architecture is less complex and works in real time.

Light Field Reconstruction Using Residual Networks on Raw Images.

Although Light-Field (LF) technology attracts attention due to its large number of applications, especially with the introduction of consumer LF cameras and its frequent use, reconstructing densely sampled LF images represents a great challenge to the use and development of LF technology. Our paper proposes a learning-based method to reconstruct densely sampled LF images from a sparse set of input images. We trained our model with raw LF images rather than using multiple images of the same scene. Raw LF can represent the two-dimensional array of images captured in a single image. Therefore, it enables the network to understand and model the relationship between different images of the same scene well and thus restore more texture details and provide better quality. Using raw images has transformed the task from image reconstruction into image-to-image translation. The feature of small-baseline LF was used to define the images to be reconstructed using the nearest input view to initialize input images. Our network was trained end-to-end to minimize the sum of absolute errors between the reconstructed and ground-truth images. Experimental results on three challenging real-world datasets demonstrate the high performance of our proposed method and its outperformance over the state-of-the-art methods.