Distinguishing the Uterine Artery, the Ureter, and Nerves in Laparoscopic Surgical Images Using Ensembles of Binary Semantic Segmentation Networks.

Performing a minimally invasive surgery comes with a significant advantage regarding rehabilitating the patient after the operation. But it also causes difficulties, mainly for the surgeon or expert who performs the surgical intervention, since only visual information is available and they cannot use their tactile senses during keyhole surgeries. This is the case with laparoscopic hysterectomy since some organs are also difficult to distinguish based on visual information, making laparoscope-based hysterectomy challenging. In this paper, we propose a solution based on semantic segmentation, which can create pixel-accurate predictions of surgical images and differentiate the uterine arteries, ureters, and nerves. We trained three binary semantic segmentation models based on the U-Net architecture with the EfficientNet-b3 encoder; then, we developed two ensemble techniques that enhanced the segmentation performance. Our pixel-wise ensemble examines the segmentation map of the binary networks on the lowest level of pixels. The other algorithm developed is a region-based ensemble technique that takes this examination to a higher level and makes the ensemble based on every connected component detected by the binary segmentation networks. We also introduced and trained a classic multi-class semantic segmentation model as a reference and compared it to the ensemble-based approaches. We used 586 manually annotated images from 38 surgical videos for this research and published this dataset.

A multimodal deep learning architecture for smoking detection with a small data approach.

Covert tobacco advertisements often raise regulatory measures. This paper presents that artificial intelligence, particularly deep learning, has great potential for detecting hidden advertising and allows unbiased, reproducible, and fair quantification of tobacco-related media content. We propose an integrated text and image processing model based on deep learning, generative methods, and human reinforcement, which can detect smoking cases in both textual and visual formats, even with little available training data. Our model can achieve 74% accuracy for images and 98% for text. Furthermore, our system integrates the possibility of expert intervention in the form of human reinforcement. Using the pre-trained multimodal, image, and text processing models available through deep learning makes it possible to detect smoking in different media even with few training data.

Machine learning approach combined with causal relationship inferring unlocks the shared pathomechanism between COVID-19 and acute myocardial infarction.

Background: Increasing evidence suggests that people with Coronavirus Disease 2019 (COVID-19) have a much higher prevalence of Acute Myocardial Infarction (AMI) than the general population. However, the underlying mechanism is not yet comprehended. Therefore, our study aims to explore the potential secret behind this complication. Materials and methods: The gene expression profiles of COVID-19 and AMI were acquired from the Gene Expression Omnibus (GEO) database. After identifying the differentially expressed genes (DEGs) shared by COVID-19 and AMI, we conducted a series of bioinformatics analytics to enhance our understanding of this issue. Results: Overall, 61 common DEGs were filtered out, based on which we established a powerful diagnostic predictor through 20 mainstream machine-learning algorithms, by utilizing which we could estimate if there is any risk in a specific COVID-19 patient to develop AMI. Moreover, we explored their shared implications of immunology. Most remarkably, through the Bayesian network, we inferred the causal relationships of the essential biological processes through which the underlying mechanism of co-pathogenesis between COVID-19 and AMI was identified. Conclusion: For the first time, the approach of causal relationship inferring was applied to analyzing shared pathomechanism between two relevant diseases, COVID-19 and AMI. Our findings showcase a novel mechanistic insight into COVID-19 and AMI, which may benefit future preventive, personalized, and precision medicine.Graphical abstract.

A Novel Immune-Related Gene Prognostic Index (IRGPI) in Pancreatic Adenocarcinoma (PAAD) and Its Implications in the Tumor Microenvironment.

PURPOSE: Pancreatic adenocarcinoma (PAAD) is one of the most lethal malignancies, with less than 10% of patients surviving more than 5 years. Existing biomarkers for reliable survival rate prediction need to be enhanced. As a result, the objective of this study was to create a novel immune-related gene prognostic index (IRGPI) for estimating overall survival (OS) and to analyze the molecular subtypes based on this index. Materials and procedures: RNA sequencing and clinical data were retrieved from publicly available sources and analyzed using several R software packages. A unique IRGPI and optimum risk model were developed using a machine learning algorithm. The prediction capability of our model was then compared to that of previously proposed models. A correlation study was also conducted between the immunological tumor microenvironment, risk groups, and IRGPI genes. Furthermore, we classified PAAD into different molecular subtypes based on the expression of IRGPI genes and investigated their features in tumor immunology using the K-means clustering technique. RESULTS: A 12-gene IRGPI (FYN, MET, LRSAM1, PSPN, ERAP2, S100A1, IL20RB, MAP3K14, SEMA6C, PRKCG, CXCL11, and GH1) was established, and verified along with a risk model. OS prediction by our model outperformed previous gene signatures. According to the findings of our correlation studies, different risk groups and IRGPI genes were found to be tightly related to tumor microenvironments, and PAAD could be further subdivided into immunologically distinct molecular subtypes based on the expression of IRGPI genes. CONCLUSION: The current study constructed and verified a unique IRGPI. Furthermore, our findings revealed a connection between the IRGPI and the immunological microenvironment of tumors. PAAD was differentiated into several molecular subtypes that might react differently to immunotherapy. These findings could provide new insights for precision and translational medicine for more innovative immunotherapy strategies.

Cuproptosis-related gene index: A predictor for pancreatic cancer prognosis, immunotherapy efficacy, and chemosensitivity.

Aim: The term "Cuproptosis" was coined to describe a novel type of cell death triggered by intracellular copper buildup that is fundamentally distinct from other recognized types such as autophagy, ferroptosis, and pyroptosis in recent days. As the underlying mechanism was newly identified, its potential connection to pancreatic adenocarcinoma (PAAD) is still an open issue. Methods: A set of machine learning algorithms was used to develop a Cuproptosis-related gene index (CRGI). Its immunological characteristics were studied by exploring its implications on the expression of the immunological checkpoints, prospective immunotherapy responses, etc. Moreover, the sensitivity to chemotherapeutic drugs was predicted. Unsupervised consensus clustering was performed to more precisely identify different CRGI-based molecular subtypes and investigate the immunotherapy and chemotherapy efficacy. The expression of DLAT, LIPT1 and LIAS were also investigated, through real-time quantitative polymerase chain reaction (RT-qPCR), western blot, and immunofluorescence staining (IFS). Results: A novel CRGI was identified and validated. Additionally, correlation analysis revealed major changes in tumor immunology across the high- and low-CRGI groups. Through an in-depth study of each medication, it was determined that the predictive chemotherapeutic efficacy of 32 regularly used anticancer drugs differed between high- and low-CRGI groups. The results of the molecular subtyping provided more support for such theories. Expressional assays performed at transcriptomic and proteomic levels suggested that the aforementioned Cuproptosis-related genes might serve as reliable diagnostic biomarkers in PAAD. Significance: This is, to the best of our knowledge, the first study to examine prognostic prediction in PAAD from the standpoint of Cuproptosis. These findings may benefit future immunotherapy and chemotherapeutic therapies.

Isocyanide Substitution in Acridine Orange Shifts DNA Damage-Mediated Phototoxicity to Permeabilization of the Lysosomal Membrane in Cancer Cells.

In cancer therapy, immunogenic cell death eliminates tumor cells more efficiently than conventional apoptosis. During photodynamic therapy (PDT), some photosensitizer (PS) targeting lysosomes divert apoptosis to the immunologically more relevant necrosis-like cell death. Acridine orange (AO) is a PS targeting lysosome. We synthesized a new compound, 3-N,N-dimethylamino-6-isocyanoacridine (DM), a modified AO, aiming to target lysosomes better. To compare DM and AO, we studied optical properties, toxicity, cell internalization, and phototoxicity. In addition, light-mediated effects were monitored by the recently developed QUINESIn method on nuclei, and membrane stability, morphology, and function of lysosomes utilizing fluorescent probes by imaging cytometry in single cells. DM proved to be a better lysosomal marker at 405 nm excitation and lysed lysosomes more efficiently. AO injured DNA and histones more extensively than DM. Remarkably, DM's optical properties helped visualize shockwaves of nuclear DNA released from cells during the PDT. The asymmetric polar modification of the AO leads to a new compound, DM, which has increased efficacy in targeting and disrupting lysosomes. Suitable AO modification may boost adaptive immune response making PDT more efficient.

Predicting the epidemic curve of the coronavirus (SARS-CoV-2) disease (COVID-19) using artificial intelligence: An application on the first and second waves.

OBJECTIVES: The COVID-19 pandemic is considered a major threat to global public health. The aim of our study was to use the official epidemiological data to forecast the epidemic curves (daily new cases) of the COVID-19 using Artificial Intelligence (AI)-based Recurrent Neural Networks (RNNs), then to compare and validate the predicted models with the observed data. METHODS: We used publicly available datasets from the World Health Organization and Johns Hopkins University to create a training dataset, then we employed RNNs with gated recurring units (Long Short-Term Memory - LSTM units) to create two prediction models. Our proposed approach considers an ensemble-based system, which is realized by interconnecting several neural networks. To achieve the appropriate diversity, we froze some network layers that control the way how the model parameters are updated. In addition, we could provide country-specific predictions by transfer learning, and with extra feature injections from governmental constraints, better predictions in the longer term are achieved. We have calculated the Root Mean Squared Logarithmic Error (RMSLE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE) to thoroughly compare our model predictions with the observed data. RESULTS: We reported the predicted curves for France, Germany, Hungary, Italy, Spain, the United Kingdom, and the United States of America. The result of our study underscores that the COVID-19 pandemic is a propagated source epidemic, therefore repeated peaks on the epidemic curve are to be anticipated. Besides, the errors between the predicted and validated data and trends seem to be low. CONCLUSION: Our proposed model has shown satisfactory accuracy in predicting the new cases of COVID-19 in certain contexts. The influence of this pandemic is significant worldwide and has already impacted most life domains. Decision-makers must be aware, that even if strict public health measures are executed and sustained, future peaks of infections are possible. The AI-based models are useful tools for forecasting epidemics as these models can be recalculated according to the newly observed data to get a more precise forecasting.

IDRiD: Diabetic Retinopathy - Segmentation and Grading Challenge.

Diabetic Retinopathy (DR) is the most common cause of avoidable vision loss, predominantly affecting the working-age population across the globe. Screening for DR, coupled with timely consultation and treatment, is a globally trusted policy to avoid vision loss. However, implementation of DR screening programs is challenging due to the scarcity of medical professionals able to screen a growing global diabetic population at risk for DR. Computer-aided disease diagnosis in retinal image analysis could provide a sustainable approach for such large-scale screening effort. The recent scientific advances in computing capacity and machine learning approaches provide an avenue for biomedical scientists to reach this goal. Aiming to advance the state-of-the-art in automatic DR diagnosis, a grand challenge on "Diabetic Retinopathy - Segmentation and Grading" was organized in conjunction with the IEEE International Symposium on Biomedical Imaging (ISBI - 2018). In this paper, we report the set-up and results of this challenge that is primarily based on Indian Diabetic Retinopathy Image Dataset (IDRiD). There were three principal sub-challenges: lesion segmentation, disease severity grading, and localization of retinal landmarks and segmentation. These multiple tasks in this challenge allow to test the generalizability of algorithms, and this is what makes it different from existing ones. It received a positive response from the scientific community with 148 submissions from 495 registrations effectively entered in this challenge. This paper outlines the challenge, its organization, the dataset used, evaluation methods and results of top-performing participating solutions. The top-performing approaches utilized a blend of clinical information, data augmentation, and an ensemble of models. These findings have the potential to enable new developments in retinal image analysis and image-based DR screening in particular.

Automatic screening of fundus images using a combination of convolutional neural network and hand-crafted features.

Diabetic retinopathy (DR) and especially diabetic macular edema (DME) are common causes of vision loss as complications of diabetes. In this work, we consider an ensemble that organizes a convolutional neural network (CNN) and traditional hand-crafted features into a single architecture for retinal image classification. This approach allows the joint training of a CNN and the fine-tuning of the weights of handcrafted features to provide a final prediction. Our solution is dedicated to the automatic classification of fundus images according to the severity level of DR and DME. For an objective evaluation of our approach, we have tested its performance on the official test datasets of the IEEE International Symposium on Biomedical Imaging (ISBI) 2018 Challenge 2: Diabetic Retinopathy Segmentation and Grading Challenge, section B. Disease Grading: Classification of fundus images according to the severity level of diabetic retinopathy and diabetic macular edema. As for our experimental results based on testing on the Indian Diabetic Retinopathy Image Dataset (IDRiD), the classification accuracies have been found to be 90.07% for the 5-class DR challenge, and 96.85% for the 3-class DME one.

Segmentation of the uterine wall by an ensemble of fully convolutional neural networks.

In the past decades, the number of in vitro fertilization (IVF) procedures for the conception of a child has been rising continuously, however, the success rate of artificial insemination remained low. According to current statistics, large portion of unsuccessful IVF relates to some women' factors. As the directly related female organ, the proper investigation of the uterus has primary importance. Namely, visible markers may indicate inflammations or other negative effects that jeopardize successful implantation. The purpose of this study is to support the observability of the uterus from this aspect by providing computer-aided tools for the extraction of its wall from video hysteroscopy. As for methodology, fully convolutional neural networks (FCNNs) are used for the automatic segmentation of the video frames to determine the region of interest. We provide the necessary steps for the applicability of the general deep learning framework for this specific task. Moreover, we increase segmentation accuracy with applying ensemble-based approaches at two levels. First, the predictions of a given FCNN are aggregated for the overlapping regions of subimages, which are derived from the splitting of the original images. Next, the segmentation results of different FCNNs are fused via a weighted combination model; optimization for adjusting the weights are also provided. Based on our experimental results, we have achieved 91.56% segmentation accuracy regarding the recognition of the uterus wall.

Classification Of Skin Lesions Using An Ensemble Of Deep Neural Networks.

Skin cancer is among the deadliest variants of cancer if not recognized and treated in time. This work focuses on the identification of this disease using an ensemble of state-of-the-art deep learning approaches. More specifically, we propose the aggregation of robust convolutional neural networks (CNNs) into one neural net architecture, where the final classification is achieved based on the weighted output of the member CNNs. Since our framework is realized within a single neural net architecture, all the parameters of the member CNNs and the weights applied in the fusion can be determined by backpropagation routinely applied for such tasks. The presented ensemble consists of the CNNs AlexNet, VGGNet, GoogLeNet, all of which have been won in subsequent years the most prominent worldwide image classification challenge ImageNet. For an objective evaluation of our approach, we have tested its performance on the official test database of the IEEE International Symposium on Biomedical Imaging (ISBI) 2017 challenge on Skin Lesion Analysis Towards Melanoma Detection dedicated to skin cancer recognition. Our experimental studies show that the proposed approach is competitive in this field. Moreover, the ensemble-based approach outperformed all of its member CNNs.

Fusion of Deep Convolutional Neural Networks for Microaneurysm Detection in Color Fundus Images.

Microaneurysms (MAs) are common signsof several diseases, appearing as small circular darkish spots in color fundus images. The presence of even a single MA may suggest diseases (e.g. diabetic retinopathy), thus, their reliable recognition is a critical issue in both human clinical practice and computer-aided systems. As for their automatic recognition, deep learning techniques became very popular in the recent years. In this paper, we also apply such deep convolutional neural network (DCNN) based techniques; however, we organize them into a supernetwork with a fusionbased approach. The combination of the member DCNNs is achieved with interconnecting them in a joint fully-connected layer. The advantage of the method is that this large architecture can be trained as a single neural network, and thus, the member DCNNs are also trained with taking the predictions of the other members into consideration. The competitiveness of our approach is also validated with experimental studies, where the ensemble-based system outperformed each member DCNN. As a primary application domain with strong clinical motivation, the methodology was tested for image-level classification. More specifically, a retinal image is divided into subimages to provide the required inputs for the DCNN-based architecture, and the whole image is labeled as a positive case, if the presence of MA is predicted in any of the subimages. Additionally, we also demonstrate how our architecture can be trained to accurately localize MAs with training only the local neighborhoods of the lesions; empirical tests showing solid performance are also enclosed.

A review on automatic analysis techniques for color fundus photographs.

In this paper, we give a review on automatic image processing tools to recognize diseases causing specific distortions in the human retina. After a brief summary of the biology of the retina, we give an overview of the types of lesions that may appear as biomarkers of both eye and non-eye diseases. We present several state-of-the-art procedures to extract the anatomic components and lesions in color fundus photographs and decision support methods to help clinical diagnosis. We list publicly available databases and appropriate measurement techniques to compare quantitatively the performance of these approaches. Furthermore, we discuss on how the performance of image processing-based systems can be improved by fusing the output of individual detector algorithms. Retinal image analysis using mobile phones is also addressed as an expected future trend in this field.

HTSFinder: Powerful Pipeline of DNA Signature Discovery by Parallel and Distributed Computing.

Comprehensive effort for low-cost sequencing in the past few years has led to the growth of complete genome databases. In parallel with this effort, a strong need, fast and cost-effective methods and applications have been developed to accelerate sequence analysis. Identification is the very first step of this task. Due to the difficulties, high costs, and computational challenges of alignment-based approaches, an alternative universal identification method is highly required. Like an alignment-free approach, DNA signatures have provided new opportunities for the rapid identification of species. In this paper, we present an effective pipeline HTSFinder (high-throughput signature finder) with a corresponding k-mer generator GkmerG (genome k-mers generator). Using this pipeline, we determine the frequency of k-mers from the available complete genome databases for the detection of extensive DNA signatures in a reasonably short time. Our application can detect both unique and common signatures in the arbitrarily selected target and nontarget databases. Hadoop and MapReduce as parallel and distributed computing tools with commodity hardware are used in this pipeline. This approach brings the power of high-performance computing into the ordinary desktop personal computers for discovering DNA signatures in large databases such as bacterial genome. A considerable number of detected unique and common DNA signatures of the target database bring the opportunities to improve the identification process not only for polymerase chain reaction and microarray assays but also for more complex scenarios such as metagenomics and next-generation sequencing analysis.

A self-calibrating approach for the segmentation of retinal vessels by template matching and contour reconstruction.

The automated processing of retinal images is a widely researched area in medical image analysis. Screening systems based on the automated and accurate recognition of retinopathies enable the earlier diagnosis of diseases like diabetic retinopathy, hypertension and their complications. The segmentation of the vascular system is a crucial task in the field: on the one hand, the accurate extraction of the vessel pixels aids the detection of other anatomical parts (like the optic disc Hoover and Goldbaum, 2003) and lesions (like microaneurysms Sopharak et al., 2013); on the other hand, the geometrical features of the vascular system and their temporal changes are shown to be related to diseases, like the vessel tortuosity to Fabry disease Sodi et al., 2013 and the arteriolar-to-venus (A/V) ratio to hypertension (Pakter et al., 2005). In this study, a novel technique based on template matching and contour reconstruction is proposed for the segmentation of the vasculature. In the template matching step generalized Gabor function based templates are used to extract the center lines of vessels. Then, the intensity characteristics of vessel contours measured in training databases are reconstructed. The method was trained and tested on two publicly available databases, DRIVE and STARE; and reached an average accuracy of 0.9494 and 0.9610, respectively. We have also carried out cross-database tests and found that the accuracy scores are higher than that of any previous technique trained and tested on the same database.

Segmentation of retinal vessels by means of directional response vector similarity and region growing.

This paper presents a novel retinal vessel segmentation method. Opposed to the general approach in similar directional methods, where only the maximal or summed responses of a pixel are used, here, the directional responses of a pixel are considered as a vector. The segmentation method is a unique region growing procedure which combines a hysteresis thresholding scheme with the response vector similarity of adjacent pixels. A vessel score map is constructed as the combination of the statistical measures of the response vectors and its local maxima to provide the seeds for the region growing procedure. A nearest neighbor classifier based on a rotation invariant response vector similarity measure is used to filter the seed points. Many techniques in the literature that capture the Gaussian-like cross-section of vessels suffer from the drawback of giving false high responses to the steep intensity transitions at the boundary of the optic disc and bright lesions. To overcome this issue, we also propose a symmetry constrained multiscale matched filtering technique. The proposed vessel segmentation method has been tested on three publicly available image sets, where its performance proved to be competitive with the state-of-the-art and comparable to the accuracy of a human observer, as well.

Detection of the optic disc in fundus images by combining probability models.

In this paper, we propose a combination method for the automatic detection of the optic disc (OD) in fundus images based on ensembles of individual algorithms. We have studied and adapted some of the state-of-the-art OD detectors and finally organized them into a complex framework in order to maximize the accuracy of the localization of the OD. The detection of the OD can be considered as a single-object detection problem. This object can be localized with high accuracy by several algorithms extracting single candidates for the center of the OD and the final location can be defined using a single majority voting rule. To include more information to support the final decision, we can use member algorithms providing more candidates which can be ranked based on the confidence ordered by the algorithms. In this case, a spatial weighted graph is defined where the candidates are considered as its nodes, and the final OD position is determined in terms of finding a maximum-weighted clique. Now, we examine how to apply in our ensemble-based framework all the accessible information supplied by the member algorithms by making them return confidence values for each image pixel. These confidence values inform us about the probability that a given pixel is the center point of the object. We apply axiomatic and Bayesian approaches, as in the case of aggregation of judgments of experts in decision and risk analysis, to combine these confidence values. According to our experimental study, the accuracy of the localization of OD increases further. Besides single localization, this approach can be adapted for the precise detection of the boundary of the OD. Comparative experimental results are also given for several publicly available datasets.

Combined Methods for Diabetic Retinopathy Screening, Using Retina Photographs and Tear Fluid Proteomics Biomarkers.

Background. It is estimated that 347 million people suffer from diabetes mellitus (DM), and almost 5 million are blind due to diabetic retinopathy (DR). The progression of DR can be slowed down with early diagnosis and treatment. Therefore our aim was to develop a novel automated method for DR screening. Methods. 52 patients with diabetes mellitus were enrolled into the project. Of all patients, 39 had signs of DR. Digital retina images and tear fluid samples were taken from each eye. The results from the tear fluid proteomics analysis and from digital microaneurysm (MA) detection on fundus images were used as the input of a machine learning system. Results. MA detection method alone resulted in 0.84 sensitivity and 0.81 specificity. Using the proteomics data for analysis 0.87 sensitivity and 0.68 specificity values were achieved. The combined data analysis integrated the features of the proteomics data along with the number of detected MAs in the associated image and achieved sensitivity/specificity values of 0.93/0.78. Conclusions. As the two different types of data represent independent and complementary information on the outcome, the combined model resulted in a reliable screening method that is comparable to the requirements of DR screening programs applied in clinical routine.

Improving the performance of an ensemble-based exudate detection system using stochastic parameter optimization.

In this paper, we propose an efficient way to determine the optimal parameter setting of an ensemble based system dedicated to the detection of exudates in retinal images. We show that the optimal parameter settings of an individual detector may be different, when it becomes a member of an ensemble. We consider a stochastic search algorithm to solve this optimization problem. However, since the computational demand is extremely high, we introduce a specific speed-up by sampling the test dataset in every search step. We show that this approach is equivalent to the noisy evaluation of the energy function and fits the corresponding theoretical results to our case. Experimental results showing improvement with respect to the member exudate detectors and individually optimal parameter settings are presented for publicly available datasets.

SRIdent: A novel pipeline for real-time identification of species from high-throughput sequencing reads in Metagenomics and clinical diagnostic assays.

New advances in rapid sequencing of large amounts of DNA have brought a great potential for the study of complex communities of microorganisms. One of the challenging problems is rapid identification of species from sequenced reads. Delays in the identification of pathogens are a barrier to the early diagnosis and proper treatment of infectious diseases. In this paper we proposed SRIdent (Short Read Identifier), an effective pipeline for real-time identification of species from high-throughput sequencing reads in Metagenomics and clinical diagnostic assays. This pipeline is based on generating k-mers from the short reads and searching the existence of DNA signatures in the Reads k-mers, by using Apache Hive data-warehousing. RkmerG (Read k-mers Generator) is a software program presented in this paper, for producing k-mers of the short reads, in order to use in the pipeline. The purpose of this study is to identify the species in a sample, directly from the reads without assembling and alignment.