Abeer Mahmoud Mahmoud
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Abeer Mahmoud Mahmoud has a total of 37 co-authors. Here are the top co-authors by number of shared publications:
- Mohamed Mabrouk Morsey
Department of Computer Science, Faculty of Computer and Information Sciences, Ain Shams University, Cairo 11566, Egypt
4 shared publications - Mostafa Aref
Department Computer Science, Faculty of Computer and Information Sciences,Ain Shams University, Cairo, Egypt
3 shared publications - Tokuji Okada
Information Science and Engineering, Graduate School of Science and Technology, Niigata University, Niigata 950-2181, Japan
3 shared publications Hanen Karamti
Computer Sciences Department, College of Computer and Information Sciences, Princess Nourah bint Abdulrahman University, Riyadh, Kingdom of Saudi Arabia
2 shared publications- Mohamed Gawish
Ain Shams University,Faculty of Computer and Information Sciences,Computer Science Department,Cairo,Egypt
2 shared publications - Nada S. El-Askary
Ain Shams University,Faculty of Computer and Information Sciences,Computer Science Department,Cairo,Egypt
2 shared publications
Latest Publications
- Open Access
- 6 October 2026
- Published by MDPI AG in Journal of Imaging
- Vol. 12 (10) , 491
Adaptive Decision-Level Fusion for Osteosarcoma Classification Using Unpaired X-Ray and Histopathology Data
Osteosarcoma is an aggressive bone malignancy in children and adolescents, where early and accurate classification is essential for improving clinical outcomes. While deep learning methods have shown promising results in medical image analysis, most existing approaches rely on single-modality data or assume patient-level correspondence in multimodal learning, which is rarely available in real-world clinical settings. In addition, many fusion strategies are designed for paired datasets and depend on fixed weighting schemes, limiting their applicability to heterogeneous and independently collected data. To address these challenges, this study proposes a modality-agnostic ensemble learning framework across independent cohorts for osteosarcoma classification using histopathology and X-ray images. Each modality is first modeled independently using EfficientNet-B0, and decision-level integration is performed without requiring cross-modal alignment. Two fusion strategies are investigated: fixed-weight probability fusion and a proposed adaptive sample-level gating network that dynamically learns fusion weights based on modality-specific predictive confidence. Experiments conducted on publicly available histopathology (TCIA) and X-ray (BTXRD) datasets demonstrate that the proposed adaptive fusion approach achieves superior performance compared to fixed fusion. The adaptive gating network reaches an accuracy of 97.08%, outperforming fixed-weight fusion (94.15%) and unimodal baselines. These results confirm that adaptive decision-level fusion can effectively leverage complementary information across unpaired datasets. Overall, the proposed framework provides a practical solution for multimodal medical image classification under data fragmentation conditions, where paired datasets are not available.
- 6 October 2026
- Published by MDPI AG in Journal of Imaging
- Vol. 12 (10) , 491
Open AccessAdaptive Decision-Level Fusion for Osteosarcoma Classification Using Unpaired X-Ray and Histopathology Data
Osteosarcoma is an aggressive bone malignancy in children and adolescents, where early and accurate classification is essential for improving clinical outcomes. While deep learning methods have shown promising results in medical image analysis, most existing approaches rely on single-modality data or assume patient-level correspondence in multimodal learning, which is rarely available in real-world clinical settings. In addition, many fusion strategies are designed for paired datasets and depend on fixed weighting schemes, limiting their applicability to heterogeneous and independently collected data. To address these challenges, this study proposes a modality-agnostic ensemble learning framework across independent cohorts for osteosarcoma classification using histopathology and X-ray images. Each modality is first modeled independently using EfficientNet-B0, and decision-level integration is performed without requiring cross-modal alignment. Two fusion strategies are investigated: fixed-weight probability fusion and a proposed adaptive sample-level gating network that dynamically learns fusion weights based on modality-specific predictive confidence. Experiments conducted on publicly available histopathology (TCIA) and X-ray (BTXRD) datasets demonstrate that the proposed adaptive fusion approach achieves superior performance compared to fixed fusion. The adaptive gating network reaches an accuracy of 97.08%, outperforming fixed-weight fusion (94.15%) and unimodal baselines. These results confirm that adaptive decision-level fusion can effectively leverage complementary information across unpaired datasets. Overall, the proposed framework provides a practical solution for multimodal medical image classification under data fragmentation conditions, where paired datasets are not available.
- 25 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 210-218
Adaptive Dynamic RGB Filtering in Immersive Metaverse Environments to Enhance Red-Green Color Differentiation for Color Blindness Children
The Color Vision Deficiency (CVD), especially Red-Green, heavily constrain equitable learning that Color-Dependent due to assured that it prevents learning of millions of children. The existing static improvement methods fail in an immersive environment, which lacks accuracy and naturalness. This article presents an adaptive dynamic RGB filtration system for the immersive Metaverse space. Our new AI-Adaptive model provides Real-Time, conceptually color improvement, which corresponds to individual CVD profiles, which are originally integrated through Python, OpenCV, PyTorch/Tensorflow and OpenXR. A strong KNIME Analytics Platform line facilitates continuous processing through user behavior analysis. Empirical results show important effect: Improve visual discrimination, increase learning outcomes and user engagement and increase in comfort. This research provides a transformative path for inclusive education, and use AI to coordinate the visual reality with individual perceptual needs in Metaverse.
- 25 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 210-218
Adaptive Dynamic RGB Filtering in Immersive Metaverse Environments to Enhance Red-Green Color Differentiation for Color Blindness Children
The Color Vision Deficiency (CVD), especially Red-Green, heavily constrain equitable learning that Color-Dependent due to assured that it prevents learning of millions of children. The existing static improvement methods fail in an immersive environment, which lacks accuracy and naturalness. This article presents an adaptive dynamic RGB filtration system for the immersive Metaverse space. Our new AI-Adaptive model provides Real-Time, conceptually color improvement, which corresponds to individual CVD profiles, which are originally integrated through Python, OpenCV, PyTorch/Tensorflow and OpenXR. A strong KNIME Analytics Platform line facilitates continuous processing through user behavior analysis. Empirical results show important effect: Improve visual discrimination, increase learning outcomes and user engagement and increase in comfort. This research provides a transformative path for inclusive education, and use AI to coordinate the visual reality with individual perceptual needs in Metaverse.
- 25 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 560-567
Towards Explainable Multi-Modal Fusion Strategies for ASD Detection: A Review
Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder marked by complexities that has attracted considerable attention recently, leading to extensive research on automated diagnosis through artificial intelligence. As largescale neuroimaging data becomes more accessible datasets, researchers have begun to build systems incorporating multi-model fusion strategies and Explainable Artificial Intelligence (XAI) techniques to enhance classification performance and model interpretability. This review presents a comprehensive overview with extensive analysis of recent advancements in multi-model fusion approaches for ASD detection including early (featurelevel), intermediate (model-level), and late (decision-level) fusion techniques applied to structural MRI, functional MRI, and phenotypic data, focusing on studies utilizing the ABIDE dataset. In parallel, the growing importance of XAI is addressed, highlighting methods such as “GMDH”, “SHAP”, and “LIME” that provide transparency and insight into model decisions, enabling clinicians to understand these decisions. The paper discusses common challenges in multi-site datasets, model generalizability, and interpretability, and offers guidance for selecting appropriate fusion strategies and explanation tools. Finally, future directions are proposed to improve the clinical relevance, robustness, and ethical deployment of AI systems for ASD diagnosis.
- 25 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 560-567
Towards Explainable Multi-Modal Fusion Strategies for ASD Detection: A Review
Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder marked by complexities that has attracted considerable attention recently, leading to extensive research on automated diagnosis through artificial intelligence. As largescale neuroimaging data becomes more accessible datasets, researchers have begun to build systems incorporating multi-model fusion strategies and Explainable Artificial Intelligence (XAI) techniques to enhance classification performance and model interpretability. This review presents a comprehensive overview with extensive analysis of recent advancements in multi-model fusion approaches for ASD detection including early (featurelevel), intermediate (model-level), and late (decision-level) fusion techniques applied to structural MRI, functional MRI, and phenotypic data, focusing on studies utilizing the ABIDE dataset. In parallel, the growing importance of XAI is addressed, highlighting methods such as “GMDH”, “SHAP”, and “LIME” that provide transparency and insight into model decisions, enabling clinicians to understand these decisions. The paper discusses common challenges in multi-site datasets, model generalizability, and interpretability, and offers guidance for selecting appropriate fusion strategies and explanation tools. Finally, future directions are proposed to improve the clinical relevance, robustness, and ethical deployment of AI systems for ASD diagnosis.
- Open Access
- 30 June 2025
- Published by Egyptian Knowledge Bank in International Journal of Intelligent Computing and Information Sciences
- Vol. 25 (2) , 54-62
A MACHINE LEARNING CASE STUDY ON EARLY DETECTION OF AUTISM SPECTRUM DISORDER USING PHENOTYPIC DATA
Early Autism Spectrum Disorder (ASD) detection is crucial for promoting cognitive, motor skills, and social development. Artificial intelligence-powered systems present an exciting chance to transform ASD detection. The Autism Brain Imaging Data Exchange (ABIDE) represents a significant repository of brain imaging and phenotypic data collected from nineteen sites, encompassing a total of 1,114 cases of both ASD and typical control individuals. Each case includes 347 descriptive variables. This article demonstrates a case study on detecting ASD based on a machine learning (ML) pipeline utilizing phenotypic data from ABIDE. The ML pipeline involves four primary steps: (1) collecting and integrating data, (2) preprocessing the data, (3) training an ML model, and (4) evaluating the ML model. This article employs seven distinct ML algorithms for training the model and documenting the classification accuracy of each algorithm. During the case study, the accuracy ranged from 80.50% to 95.10%. The model trained using the random forest algorithm achieved the preeminent accuracy for ASD detection using phenotypic data.
- 30 June 2025
- Published by Egyptian Knowledge Bank in International Journal of Intelligent Computing and Information Sciences
- Vol. 25 (2) , 54-62
Open AccessA MACHINE LEARNING CASE STUDY ON EARLY DETECTION OF AUTISM SPECTRUM DISORDER USING PHENOTYPIC DATA
Early Autism Spectrum Disorder (ASD) detection is crucial for promoting cognitive, motor skills, and social development. Artificial intelligence-powered systems present an exciting chance to transform ASD detection. The Autism Brain Imaging Data Exchange (ABIDE) represents a significant repository of brain imaging and phenotypic data collected from nineteen sites, encompassing a total of 1,114 cases of both ASD and typical control individuals. Each case includes 347 descriptive variables. This article demonstrates a case study on detecting ASD based on a machine learning (ML) pipeline utilizing phenotypic data from ABIDE. The ML pipeline involves four primary steps: (1) collecting and integrating data, (2) preprocessing the data, (3) training an ML model, and (4) evaluating the ML model. This article employs seven distinct ML algorithms for training the model and documenting the classification accuracy of each algorithm. During the case study, the accuracy ranged from 80.50% to 95.10%. The model trained using the random forest algorithm achieved the preeminent accuracy for ASD detection using phenotypic data.
- 10 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 292-295
Dimension Reduction of Phenotypic Data for Enhancing Autism Spectrum Disorder Detection
Autism Spectrum Disorder (ASD) is a neurological pathology that affects the human brain and causes problems in human social development, communication, and motor skills. The Autism Brain Imaging Data Exchange (ABIDE) is a prominent collection of brain imaging data and phenotypic data from 19 different sites with a total of 1114 cases ranging between ASD and Typical Control individuals (TC). Each case has 347 variables that describe it. In this paper, these variables are analyzed with the aim of extracting and selecting the most significant features for ASD detection using a machine learning (ML) pipeline. ML pipeline consists of three main steps: (1) creating feature groups either by using grouping technique (feature selection) or by using principal component analysis technique (feature extraction), (2) building ML model using three different ML algorithms for each feature group, and (3) evaluating ML model and recording the classification accuracy obtained from each model. The group of features that scores the best classification accuracy will be selected. Our experiments demonstrate that selecting key features significantly improves classification accuracy and highlight the effect of missing data on the model accuracy even after filling them with the mean imputation technique. During the experiments, the accuracy is scaled from 67% to 94.1% by selecting the most notable group of features and running them on the perfect model.
- 10 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 292-295
Dimension Reduction of Phenotypic Data for Enhancing Autism Spectrum Disorder Detection
Autism Spectrum Disorder (ASD) is a neurological pathology that affects the human brain and causes problems in human social development, communication, and motor skills. The Autism Brain Imaging Data Exchange (ABIDE) is a prominent collection of brain imaging data and phenotypic data from 19 different sites with a total of 1114 cases ranging between ASD and Typical Control individuals (TC). Each case has 347 variables that describe it. In this paper, these variables are analyzed with the aim of extracting and selecting the most significant features for ASD detection using a machine learning (ML) pipeline. ML pipeline consists of three main steps: (1) creating feature groups either by using grouping technique (feature selection) or by using principal component analysis technique (feature extraction), (2) building ML model using three different ML algorithms for each feature group, and (3) evaluating ML model and recording the classification accuracy obtained from each model. The group of features that scores the best classification accuracy will be selected. Our experiments demonstrate that selecting key features significantly improves classification accuracy and highlight the effect of missing data on the model accuracy even after filling them with the mean imputation technique. During the experiments, the accuracy is scaled from 67% to 94.1% by selecting the most notable group of features and running them on the perfect model.
- Open Access
- 9 September 2023
- Published by Springer Nature in Scientific Reports
- Vol. 13 (1) , 1-13
A pre-protective objective in mining females social contents for identification of early signs of depression using soft computing deep framework
Currently, a noteworthy volume of information is available and shared every day through participation and communication of individuals on social media. These enormous contents with the right exploit and research leads to valuable discoveries. In this study, a deep framework of learning accurate detection of women’s depression is proposed. It is beneficially guided by social media content of individual posts and tweets and an essential support from psycho-linguistic for providing the indicator depression signs vocabulary that creates the embedding words necessary for building the applied approach. The presented model is validated using dual datasets extracted from Twitter: the first dataset is general data formed by 700 women from different countries; the second contains only 80 women from KSA. A third benchmark dataset CLPsych 2015 is used for comparative analysis purposes. The model proved its performance on the three datasets and the obtained and reported in this paper results shows its effectiveness.
- 9 September 2023
- Published by Springer Nature in Scientific Reports
- Vol. 13 (1) , 1-13
Open AccessA pre-protective objective in mining females social contents for identification of early signs of depression using soft computing deep framework
Currently, a noteworthy volume of information is available and shared every day through participation and communication of individuals on social media. These enormous contents with the right exploit and research leads to valuable discoveries. In this study, a deep framework of learning accurate detection of women’s depression is proposed. It is beneficially guided by social media content of individual posts and tweets and an essential support from psycho-linguistic for providing the indicator depression signs vocabulary that creates the embedding words necessary for building the applied approach. The presented model is validated using dual datasets extracted from Twitter: the first dataset is general data formed by 700 women from different countries; the second contains only 80 women from KSA. A third benchmark dataset CLPsych 2015 is used for comparative analysis purposes. The model proved its performance on the three datasets and the obtained and reported in this paper results shows its effectiveness.
- Open Access
- 27 December 2022
- Published by MDPI AG in Electronics
- Vol. 12 (1) , 105
Gaussian Mixture with Max Expectation Guide for Stacked Architecture of Denoising Autoencoder and DRBM for Medical Chest Scans and Disease Identification
Artificial intelligence (AI), in particular deep learning, has proven to be efficient in medical diagnosis. This paper introduces a new hybrid deep learning model for pneumonia diagnosis based on chest CT scans. At the core of the model, a Gaussian mixture is combined with the expectation-maximization algorithm (EMGMM) to extract the regions of interest (ROI), while a convolutional denoising autoencoder (DAE) and deep restricted Boltzmann machine (DRBM) are combined for the classification. In order to prevent the model from learning trivial solutions, stochastic noises were added as an input to the unsupervised learning phase. The dataset used in this work is a publicly available dataset of chest X-rays for pneumonia on the Kaggle website; it contains 5856 images with 1583 normal cases and 4273 pneumonia cases, with an imbalance ratio (IR) of 0.46. Several operations including zooming, flipping, shifting and rotation were used in the augmentation phase to balance the data distribution across the different classes, which led to enhancing the IR value to 0.028. The computational analysis of the results show that the proposed model is promising as it provides an average accuracy value of 98.63%, sensitivity value of 96.5%, and specificity value of 94.8%.
- 27 December 2022
- Published by MDPI AG in Electronics
- Vol. 12 (1) , 105
Open AccessGaussian Mixture with Max Expectation Guide for Stacked Architecture of Denoising Autoencoder and DRBM for Medical Chest Scans and Disease Identification
Artificial intelligence (AI), in particular deep learning, has proven to be efficient in medical diagnosis. This paper introduces a new hybrid deep learning model for pneumonia diagnosis based on chest CT scans. At the core of the model, a Gaussian mixture is combined with the expectation-maximization algorithm (EMGMM) to extract the regions of interest (ROI), while a convolutional denoising autoencoder (DAE) and deep restricted Boltzmann machine (DRBM) are combined for the classification. In order to prevent the model from learning trivial solutions, stochastic noises were added as an input to the unsupervised learning phase. The dataset used in this work is a publicly available dataset of chest X-rays for pneumonia on the Kaggle website; it contains 5856 images with 1583 normal cases and 4273 pneumonia cases, with an imbalance ratio (IR) of 0.46. Several operations including zooming, flipping, shifting and rotation were used in the augmentation phase to balance the data distribution across the different classes, which led to enhancing the IR value to 0.028. The computational analysis of the results show that the proposed model is promising as it provides an average accuracy value of 98.63%, sensitivity value of 96.5%, and specificity value of 94.8%.
- Open Access
- 7 July 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 10, 73956-73968
A Dual Soft-Computing Based on Genetic Algorithm and Fuzzy Logic Defect Recognition for Gearbox and Motors: Attempts Toward Optimal Performance
Motor and gearbox are considered the main components in various machines related to its supplying power and transmitting motion role. Operating machines acquire vibration signal that are continuously monitoring by sensors placing close to vibration source. This for processing and identify the machine’ components status. Breakdown of the rotating machine causes significant losses and costs, so the analysis of its vibration signals proved literately avoiding these drawbacks with effective faults diagnosis. This paper proposing two models for gearbox and motor faults identification as an attempt towards finding the optimal performance: The first developed model is a fuzzy logic (FL) based model and the other is genetic algorithm (GA) based model. The intended output of both models reduce time and cost of maintenance. It also indirectly increases the machine component’s life. Additionally, the computational analysis proved that, concerning execution time and accuracy; and with the powerful straight forward representation for uncertainties offered by the Fuzzy Logic; it is indeed reliable, however it presented lower classification accuracy (96% for gear box faults and 93% for motor faults) and lower generalization schema. Yet, the proposed strategy which integrates GA and SVM recorded high performances in optimization and higher classification capabilities (97% for both gear box and motors faults). These factors illustrate the effectiveness and optimal performance of the genetic based model.
- 7 July 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 10, 73956-73968
Open AccessA Dual Soft-Computing Based on Genetic Algorithm and Fuzzy Logic Defect Recognition for Gearbox and Motors: Attempts Toward Optimal Performance
Motor and gearbox are considered the main components in various machines related to its supplying power and transmitting motion role. Operating machines acquire vibration signal that are continuously monitoring by sensors placing close to vibration source. This for processing and identify the machine’ components status. Breakdown of the rotating machine causes significant losses and costs, so the analysis of its vibration signals proved literately avoiding these drawbacks with effective faults diagnosis. This paper proposing two models for gearbox and motor faults identification as an attempt towards finding the optimal performance: The first developed model is a fuzzy logic (FL) based model and the other is genetic algorithm (GA) based model. The intended output of both models reduce time and cost of maintenance. It also indirectly increases the machine component’s life. Additionally, the computational analysis proved that, concerning execution time and accuracy; and with the powerful straight forward representation for uncertainties offered by the Fuzzy Logic; it is indeed reliable, however it presented lower classification accuracy (96% for gear box faults and 93% for motor faults) and lower generalization schema. Yet, the proposed strategy which integrates GA and SVM recorded high performances in optimization and higher classification capabilities (97% for both gear box and motors faults). These factors illustrate the effectiveness and optimal performance of the genetic based model.
- Open Access
- 1 January 2020
- Published by INSTICC
- p. 165-172
Hybrid Shallow Learning and Deep Learning for Feature Extraction and Image Retrieval
- 1 January 2020
- Published by INSTICC
- p. 165-172
Open AccessHybrid Shallow Learning and Deep Learning for Feature Extraction and Image Retrieval
- Open Access
- 18 December 2014
- Published by Foundation of Computer Science in International Journal of Computer Applications
- Vol. 107 (1) , 13-21
Development of an Expert System for Diabetic Type-2 Diet
- 18 December 2014
- Published by Foundation of Computer Science in International Journal of Computer Applications
- Vol. 107 (1) , 13-21
Open AccessDevelopment of an Expert System for Diabetic Type-2 Diet