Your search

In this work, we propose a multi-task learning-based approach towards the localization of optic disc and fovea from human retinal fundus images using a deep learning-based approach. Formulating the task as an image-based regression problem, we propose a Densenet121-based architecture through an extensive set of experiments with a variety of CNN architectures. Our proposed approach achieved an average mean absolute error of only 13pixels (0.04%), mean squared error of 11 pixels (0.005%), and a root mean square error of only 0.02 (13%) on the IDRiD dataset.

The optical disc in the human retina can reveal important information about a person's health and well-being. We propose a deep learning-based approach to automatically identify the region in human retinal images that corresponds to the optical disc. We formulated the task as an image segmentation problem that leverages multiple public-domain datasets of human retinal fundus images. Using an attention-based residual U-Net, we showed that the optical disc in a human retina image can be detected with more than 99% pixel-level accuracy and around 95% in Matthew's Correlation Coefficient. A comparison with variants of UNet with different encoder CNN architectures ascertains the superiority of the proposed approach across multiple metrics.

Subcellular localization of messenger RNA (mRNAs) plays a pivotal role in the regulation of gene expression, cell migration as well as in cellular adaptation. Experiment techniques for pinpointing the subcellular localization of mRNAs are laborious, time-consuming and expensive. Therefore, in silico approaches for this purpose are attaining great attention in the RNA community.

Winning football matches is the major goal of all football clubs in the world. Football being the most popular game in the world, many studies have been conducted to analyze and predict match winners based on players’ physical and technical performance. In this study, we analyzed the matches from the professional football league of Qatar Stars League (QSL) covering the matches held in the last ten seasons. We incorporated the highest number of professional matches from the last ten seasons covering from 2011 up to 2022 and proposed SoccerNet, a Gated Recurrent Unit (GRU)-based deep learning-based model to predict match winners with over 80% accuracy. We considered match- and player-related information captured by STATS platform in a time slot of 15 minutes. Then we analyzed players’ performance at different positions on the field at different stages of the match. Our results indicated that in QSL, the defenders’ role in matches is more dominant than midfielders and forwarders. Moreover, our analysis suggests that the last 15–30 minutes of match segments of the matches from QSL have a more significant impact on the match result than other match segments. To the best of our knowledge, the proposed model is the first DL-based model in predicting match winners from any professional football leagues in the Middle East and North Africa (MENA) region. We believe the results will support the coaching staff and team management for QSL in designing game strategies and improve the overall quality of performance of the players.

Retinal Detachment (RD) is one of the major problems with retinal disorder patients. Till to date there existing no confirmatory sign or marker on retina for the early detection of RD. Therefore, patients may have sudden RD at any time of their life. Moreover, it is completely dependent upon the subjective judgement of ophthalmologist to make the final diagnostic decision on RD. To support the decision making process for the ophthalmologist, in this article we proposed RDNet, a SqueezeNet architecture based deep learning model for the early detection of RD. We used publicly available dataset of 1017 images covering rhegmatogenous RD and control group. The proposed model built on this image set achieved 97.55% sensitivity, 99.26% specificity and 98.23% accuracy in detecting RD. The proposed model outperformed the existing models for the same purpose with the highest area under the ROC curve (AUC) of 0.995. We believe our model will support the early detection of RD in clinical setup and assist the ophthalmologist in identifying RD at its early stage.

Diabetes is one of the leading fatal diseases globally, putting a huge burden on the global healthcare system. Early diagnosis of diabetes is hence, of utmost importance and could save many lives. However, current techniques to determine whether a person has diabetes or has the risk of developing diabetes are primarily reliant upon clinical biomarkers. In this article, we propose a novel deep learning architecture to predict if a person has diabetes or not from a photograph of his/her retina. Using a relatively small-sized dataset, we develop a multi-stage convolutional neural network (CNN)-based model DiaNet that can reach an accuracy level of over 84% on this task, and in doing so, successfully identifies the regions on the retina images that contribute to its decision-making process, as corroborated by the medical experts in the field. This is the first study that highlights the distinguishing capability of the retinal images for diabetes patients in the Qatari population to the best of our knowledge. Comparing the performance of DiaNet against the existing clinical data-based machine learning models, we conclude that the retinal images contain sufficient information to distinguish the Qatari diabetes cohort from the control group. In addition, our study reveals that retinal images may contain prognosis markers for diabetes and other comorbidities like hypertension and ischemic heart disease. The results led us to believe that the inclusion of retinal images into the clinical setup for the diagnosis of diabetes is warranted in the near future.

Human genes often, through alternative splicing of pre-messenger RNAs, produce multiple mRNAs and protein isoforms that may have similar or completely different functions. Identification of splice sites is, therefore, crucial to understand the gene structure and variants of mRNA and protein isoforms produced by the primary RNA transcripts. Although many computational methods have been developed to detect the splice sites in humans, this is still substantially a challenging problem and further improvement of the computational model is still foreseeable. Accordingly, we developed DeepDSSR (deep donor splice site recognizer), a novel deep learning based architecture, for predicting human donor splice sites. The proposed method, built upon publicly available and highly imbalanced benchmark dataset, is comparable with the leading deep learning based methods for detecting human donor splice sites. Performance evaluation metrics show that DeepDSSR outperformed the existing deep learning based methods. Future work will improve the predictive capabilities of our model, and we will build a model for the prediction of acceptor splice sites.

Diabetes mellitus (DM) and osteoporosis/osteopenia affect millions of people globally and are major health conditions in several countries including Qatar. Bone mineral density (BMD) is a widely accepted indicator for diagnosing osteoporosis (OP) and osteopenia (OPN). The best method for determining bone mineral density and OP/OPN risk is via dual energy X-ray absorptiometry (DXA) technology. The risk of osteoporosis-related fracture may increase for people with diabetes. Therefore, it is necessary to develop a system that can support the early detection of OP/OPN in diabetic patients. In this study, we analyzed Qatar diabetic cohorts including 500 subjects, among which 68 were OP/OPN (target) subjects and 432 were without osteoporosis/osteopenia (control) subjects. The objective of this study is to develop an ML model to distinguish diabetic OP/OPN patients from diabetic non-OP/non-OPN subjects based on their bone health indicators from full body DXA scan measurements. Based on our experiments, AdaBoost model performed the best for classifying the target group from the control group. 10-fold cross validation-based results indicate that the proposed ML model was able to distinguish the target group from the control group at 80% sensitivity, 96% specificity. To the best of our knowledge, our study is the first ML-based approach to detect the early onset of OP/OPN in diabetic cohort from Qatar. Our analyses revealed the higher level of lean mass, fat mass and bone mass for the control group compared to the target group. Higher levels of BMC, BMD from different body parts in the control group compared to the osteoporosis/osteopenia group indicate the protective effects of obesity on bone health in the Qatari diabetic cohort. Moreover, higher value of anthropometric measurements in troch, lumbar spine (L1, L2, L3, L4), pelvis and other body parts in the control group indicates that the WHO guideline can be applied to the Qatari diabetic cohort for the early detection of OP/OPN based on the proposed ML model. Further research on OP/OPN in diabetic patients is warranted in future to confirm the role of DM on bone health.

Epidermolysis bullosa acquisita (EBA) represents a big challenge as a rare skin disorder, with no established markers for early detection for patients. Moreover, as a rare disease, it is extremely difficult to acquire good number of patient sample to diagnose accurately with high confidence. EBA has many biomarkers very similar to other bullosa diseases and needs specific clinical expertise to detect it using immunofluorescence microscopy. In this study, we introduce a deep learningbased method, EBAnet, that leveraged Convolutional Neural Network (CNN) based model for the detection of EBA based on Direct immunofluorescence (DIF) microscopy image. The proposed EfficientNet-based model achieved 97.3% sensitivity, 96.1% precision, and 96.7% accuracy in distinguishing EBA from other class and outperformed the existing model for the same purpose. GradCAM based class activation map also highlighted the important region of the DIF images that was focused by the proposed model leveraging the explainability of the model. We believe, EBAnet will add value in the early and accurate detection of EBA, addressing a critical need in clinical practice.

Promoter region of protein-coding genes are gradually being well understood, yet no comparable studies exist for the promoter of long non-coding RNA (lncRNA) genes which has emerged as a global potential regulator in multiple cellular process and different diseases for human. To understand the difference in the transcriptional regulation pattern of these genes, previously, we proposed a machine learning based model to classify the promoter of protein-coding genes and lncRNA genes. In this study, we are presenting DeepCNPP (deep coding non-coding promoter predictor), an improved model based on deep learning (DL) framework to classify the promoter of lncRNA genes and protein-coding genes. We used convolution neural network (CNN) based deep network to classify the promoter of these two broad categories of human genes. Our computational model, built upon the sequence information only, was able to classify these two groups of promoters from human at a rate of 83.34% accuracy and outperformed the existing model. Further analysis and interpretation of the output from DeepCNPP architecture will enable us to understand the difference in transcription regulatory pattern for these two groups of genes.

Lung Adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC) are the two main histology subtypes of non-small cell lung cancer (NSCLC) with 70% of total Lung Cancer. In this article we proposed an ensemble-based model for the identification of subtypes of NSCLC using methylation data. Proposed Random Forest-based model along with out of bag (OOB) error based feature selection technique identified the top ten most important CpG sites that are highly differentiator between LUSC and LUAD subtypes of NSCLC with an accuracy, precision and F1 Score of \(97\%\) . The proposed model outperformed the other existing models for the same purpose with huge margin of 12%. Pathway analysis of the proposed 10 CpG sites revealed different pathways for LUAD and LUSC associated genes, LUAD-associated genes primarily participated in TP53, PTEN, GLP-1, Incretin regulation, and apoptosis. Conversely, LUSC-associated genes were predominantly involved in pathways for platelet degranulation, serine biosynthesis, and Nephrin family interaction.

Cardiovascular diseases (CVD) are the leading cause of death worldwide. People affected by CVDs may go undiagnosed until the occurrence of a serious heart failure event such as stroke, heart attack, and myocardial infraction. In Qatar, there is a lack of studies focusing on CVD diagnosis based on non-invasive methods such as retinal image or dual-energy X-ray absorptiometry (DXA). In this study, we aimed at diagnosing CVD using a novel approach integrating information from retinal images and DXA data. We considered an adult Qatari cohort of 500 participants from Qatar Biobank (QBB) with an equal number of participants from the CVD and the control groups. We designed a case-control study with a novel multi-modal (combining data from multiple modalities—DXA and retinal images)—to propose a deep learning (DL)-based technique to distinguish the CVD group from the control group. Uni-modal models based on retinal images and DXA data achieved 75.6% and 77.4% accuracy, respectively. The multi-modal model showed an improved accuracy of 78.3% in classifying CVD group and the control group. We used gradient class activation map (GradCAM) to highlight the areas of interest in the retinal images that influenced the decisions of the proposed DL model most. It was observed that the model focused mostly on the centre of the retinal images where signs of CVD such as hemorrhages were present. This indicates that our model can identify and make use of certain prognosis markers for hypertension and ischemic heart disease. From DXA data, we found higher values for bone mineral density, fat content, muscle mass and bone area across majority of the body parts in CVD group compared to the control group indicating better bone health in the Qatari CVD cohort. This seminal method based on DXA scans and retinal images demonstrate major potentials for the early detection of CVD in a fast and relatively non-invasive manner.

More than half a billion people worldwide are affected by diabetes, which is a prevalent non-communicable disease that can lead to critical health conditions, including vision loss. Diabetic Macular Edema (DME) is a primary cause of vision impairment and can eventually lead to blindness in diabetic patients. Early detection of DME and proper health management are crucial to controlling the disease. Retinal image-based AI-enabled diabetes diagnosis has gained significant attention as a non-invasive, fast, and reasonably accurate method for diagnosing DME. To make this technology accessible to underserved communities or areas lacking proper clinical facilities, a mobile application-based solution could have a significant impact. In this article, we describe how we transformed our previously published AI-enabled model into an Android-based mobile application, which is part of a two-phase research study. In the first phase, we developed a deep learning-based model that predicts DME grading using retinal images. In the second phase, we built a mobile application DMEgrader to make our model accessible via a mobile device. To the best of our knowledge, this is the first article to demonstrate necessary steps and code snippets to support developers in transforming deep learning models into Android based mobile applications for DME grading prediction. © 2023 IEEE.

Cardiovascular disease (CVD) is reported to be the leading cause of mortality in the middle eastern countries, including Qatar. But no comprehensive study has been conducted on the Qatar specific CVD risk factors identification. The objective of this case-control study was to develop machine learning (ML) model distinguishing healthy individuals from people having CVD, which could ultimately reveal the list of potential risk factors associated to CVD in Qatar. To the best of our knowledge, this study considered the largest collection of biomedical measurements representing the anthropometric measurements, clinical biomarkers, bioimpedance, spirometry, VICORDER readings, and behavioral factors of the CVD group from Qatar Biobank (QBB). CatBoost model achieved 93% accuracy, thereby outperforming the existing model for the same purpose. Interestingly, combining multimodal datasets into the proposed ML model outperformed the ML model built upon currently known risk factors for CVD, emphasizing the importance of incorporating other clinical biomarkers into consideration for CVD diagnosis plan. The ablation study on the multimodal dataset from QBB revealed that physio-clinical and bioimpedance measurements have the most distinguishing power to classify these two groups irrespective of gender and age of the participants. Multiple feature subset selection techniques confirmed known CVD risk factors (blood pressure, lipid profile, smoking, sedentary life, and diabetes), and identified potential novel risk factors linked to CVD-related comorbidities such as renal disorder (e.g., creatinine, uric acid, homocysteine, albumin), atherosclerosis (intima media thickness), hypercoagulable state (fibrinogen), and liver function (e.g., alkaline phosphate, gamma-glutamyl transferase). Moreover, the inclusion of the proposed novel factors into the ML model provides better performance than the model with traditional known risk factors for CVD. The association of the proposed risk factors and comorbidities are required to be investigated in clinical setup to understand their role in CVD better. © 2013 IEEE.

Diabetes, affecting more than 500 million individuals worldwide, is the most widespread non-communicable disease, globally. The early identification and effective management of diabetes are crucial for controlling its spread. Currently, the HbA1c test is the gold standard for the detection of diabetes with high confidence. But this is an invasive, expensive pathology test. Therefore, alternative non-invasive and inexpensive methods have been proposed in the literature for the early detection of diabetes.

Alzheimer's Disease (AD) is a neurodegenerative disease that causes complications with thinking capability, memory and behavior. AD is a major public health problem among the elderly in developed and developing countries. With the growth of AD around the world, there is a need to further expand our understanding of the roles different clinical measurements can have in the diagnosis of AD. In this work, we propose a machine learning-based technique to distinguish control subjects with no cognitive impairments, AD subjects, and subjects with mild cognitive impairment (MCI), often seen as precursors of AD. We utilized several machine learning (ML) techniques and found that Gradient Boosting Decision Trees achieved the highest performance above 84% classification accuracy. Also, we determined the importance of the features (clinical biomarkers) contributing to the proposed multi-class classification system. Further investigation on the biomarkers will pave the way to introduce better treatment plan for AD patients. © 2020 The authors and IOS Press.