Merge pull request #22 from CVPD/hugoblox-import-publications

Hugo Blox Builder - Import latest publications

Author: iarganda

content/publication/abou-2025-comparative/cite.bib

@@ -0,0 +1,15 @@
+@inproceedings{abou2025comparative,
+ abstract = {Accurate white blood cell (WBC) classification in peripheral blood smears is essential for diagnosing and monitoring hematological disorders. Traditionally, Convolutional Neural Networks (CNNs) have been employed, but they often demand large datasets and substantial computational resources. Recently, Vision Transformers (ViTs) have shown promise by using selfattention mechanisms to capture global image dependencies. This study provides a comparative analysis of CNNs and ViTs for WBC classification, assessing their performance using a standardized dataset of peripheral blood smear images. We evaluate accuracy, computational efficiency, and robustness to varying image qualities. Our results indicate that while CNNs perform well in feature extraction, ViTs excel in managing complex patterns and achieving higher classification accuracy with fewer samples. Additionally, we explore the effects of data augmentation and hybrid models that combine CNNs with ViTs. These approaches enhance model generalization and performance, making them promising for clinical applications requiring diverse data and high accuracy. This research advances the understanding of deep learning in medical imaging, highlighting ViTs as a viable alternative to CNNs for WBC classification. Future research will focus on optimizing these models for real-time clinical use and exploring their application in other diagnostic fields.},
+ author = {Mohamad Abou Ali and Fadi Dornaika and Ignacio Arganda-Carreras and Abdallah Kassem},
+ booktitle = {2025 Eighth International Conference on Advances in Biomedical Engineering (ICABME)},
+ doi = {10.1109/ICABME66883.2025.11211837},
+ issn = {2377-5696},
+ keywords = {White blood cells, Deep learning, Computer vision, Accuracy, Transformers, Data models, Robustness, Real-time systems, Convolutional neural networks, Artificial intelligence, Medical Image Analysis, Blood Cell Classification, Artificial Intelligence (AI), Deep Learning (DL), Vision Transformers (ViTs), Convolutional Neural Networks (CNNs), Automation},
+ month = {Oct},
+ number = {},
+ pages = {1-6},
+ title = {Comparative Analysis of Convolutional Neural Networks and Vision Transformers for White Blood Cell Classification in Medical Imaging},
+ url = {https://ieeexplore.ieee.org/abstract/document/11211837},
+ volume = {},
+ year = {2025}
+}

content/publication/abou-2025-comparative/index.md

@@ -0,0 +1,55 @@
+---
+title: Comparative Analysis of Convolutional Neural Networks and Vision Transformers
+  for White Blood Cell Classification in Medical Imaging
+authors:
+- Mohamad Abou Ali
+- Fadi Dornaika
+- Ignacio Arganda-Carreras
+- Abdallah Kassem
+date: '2025-10-01'
+publishDate: '2026-04-13T10:35:42.242075Z'
+publication_types:
+- paper-conference
+publication: '*2025 Eighth International Conference on Advances in Biomedical Engineering
+  (ICABME)*'
+doi: 10.1109/ICABME66883.2025.11211837
+abstract: Accurate white blood cell (WBC) classification in peripheral blood smears
+  is essential for diagnosing and monitoring hematological disorders. Traditionally,
+  Convolutional Neural Networks (CNNs) have been employed, but they often demand large
+  datasets and substantial computational resources. Recently, Vision Transformers
+  (ViTs) have shown promise by using selfattention mechanisms to capture global image
+  dependencies. This study provides a comparative analysis of CNNs and ViTs for WBC
+  classification, assessing their performance using a standardized dataset of peripheral
+  blood smear images. We evaluate accuracy, computational efficiency, and robustness
+  to varying image qualities. Our results indicate that while CNNs perform well in
+  feature extraction, ViTs excel in managing complex patterns and achieving higher
+  classification accuracy with fewer samples. Additionally, we explore the effects
+  of data augmentation and hybrid models that combine CNNs with ViTs. These approaches
+  enhance model generalization and performance, making them promising for clinical
+  applications requiring diverse data and high accuracy. This research advances the
+  understanding of deep learning in medical imaging, highlighting ViTs as a viable
+  alternative to CNNs for WBC classification. Future research will focus on optimizing
+  these models for real-time clinical use and exploring their application in other
+  diagnostic fields.
+tags:
+- White blood cells
+- Deep learning
+- Computer vision
+- Accuracy
+- Transformers
+- Data models
+- Robustness
+- Real-time systems
+- Convolutional neural networks
+- Artificial intelligence
+- Medical Image Analysis
+- Blood Cell Classification
+- Artificial Intelligence (AI)
+- Deep Learning (DL)
+- Vision Transformers (ViTs)
+- Convolutional Neural Networks (CNNs)
+- Automation
+links:
+- name: URL
+  url: https://ieeexplore.ieee.org/abstract/document/11211837
+---

content/publication/abou-2025-robust/cite.bib

@@ -0,0 +1,15 @@
+@inproceedings{abou2025robust,
+ abstract = {Detecting Alzheimer's disease (AD) from MRI scans remains a significant challenge in medical imaging due to the subtle, low-contrast anatomical changes associated with neurodegeneration. Real-world scans often suffer from noise, motion artifacts, and scanner variability, all of which can severely degrade deep learning (DL) model performance. In our previous work on MRI-based 4-class brain tumor classification, we developed a robustness methodology that employed Gaussian blur and Gaussian noise as augmentation techniques to counteract the impact of various image degradations-including multiple noise types, blurring methods, and simulated patient motion. In this study, we adapt and apply the same robustness strategy to a 4-class AD classification task using a publicly available Kaggle MRI dataset. To address class imbalance, we employed a Conditional Wasserstein GAN with Gradient Penalty (WGAN-GP) to generate a balanced training set of $\mathbf2, 5 6 0$ images per class, while retaining the original, unaugmented test set. A fine-tuned EfficientNetV2B0 model achieved 99 % accuracy on the augmented training data; however, when evaluated under real-world challenges, performance dropped to $\mathbf4 4. 3 3 %$-significantly lower than the 69.66 % achieved in the brain tumor benchmark. This stark contrast underscores the inherent difficulty of AD classification, exacerbated by low-resolution axial-only scans and minimal interclass visual differences. Although the robustness methodology yielded partial performance recovery, the results highlight the need for higher-resolution, multi-planar imaging and domain-specific preprocessing strategies. This study advances the understanding of robustness in DL-based neuroimaging and underscores the importance of tailored augmentation pipelines for reliable Alzheimer's diagnosis in clinical settings.},
+ author = {Mohamad Abou Ali and Mostafa Mohieldin and Fadi Dornaika and Ignacio Arganda-Carreras and Rejdi Chmouri and Ali Cherry and Ali Hage-Diab},
+ booktitle = {2025 Eighth International Conference on Advances in Biomedical Engineering (ICABME)},
+ doi = {10.1109/ICABME66883.2025.11211705},
+ issn = {2377-5696},
+ keywords = {Deep learning, Training, Visualization, Magnetic resonance imaging, Training data, Generative adversarial networks, Brain modeling, Robustness, Alzheimer's disease, Artificial intelligence, MRI Alzheimer Disease, MRI Brain Cancer, Deep Learning (DL), Artificial Intelligence (AI), Real-World simulation, Robustness, Generative Adversarial Networks (GAN), Gaussian Augmentation},
+ month = {Oct},
+ number = {},
+ pages = {1-5},
+ title = {Robust Deep Learning for MRI-Based Alzheimer's Disease Detection: Overcoming Real-World Imaging Challenges},
+ url = {https://ieeexplore.ieee.org/abstract/document/11211705},
+ volume = {},
+ year = {2025}
+}

content/publication/abou-2025-robust/index.md

@@ -0,0 +1,63 @@
+---
+title: "Robust Deep Learning for MRI-Based Alzheimer's Disease Detection: Overcoming
+  Real-World Imaging Challenges"
+authors:
+- Mohamad Abou Ali
+- Mostafa Mohieldin
+- Fadi Dornaika
+- Ignacio Arganda-Carreras
+- Rejdi Chmouri
+- Ali Cherry
+- Ali Hage-Diab
+date: '2025-10-01'
+publishDate: '2026-04-13T10:35:42.258584Z'
+publication_types:
+- paper-conference
+publication: '*2025 Eighth International Conference on Advances in Biomedical Engineering
+  (ICABME)*'
+doi: 10.1109/ICABME66883.2025.11211705
+abstract: Detecting Alzheimer's disease (AD) from MRI scans remains a significant
+  challenge in medical imaging due to the subtle, low-contrast anatomical changes
+  associated with neurodegeneration. Real-world scans often suffer from noise, motion
+  artifacts, and scanner variability, all of which can severely degrade deep learning
+  (DL) model performance. In our previous work on MRI-based 4-class brain tumor classification,
+  we developed a robustness methodology that employed Gaussian blur and Gaussian noise
+  as augmentation techniques to counteract the impact of various image degradations-including
+  multiple noise types, blurring methods, and simulated patient motion. In this study,
+  we adapt and apply the same robustness strategy to a 4-class AD classification task
+  using a publicly available Kaggle MRI dataset. To address class imbalance, we employed
+  a Conditional Wasserstein GAN with Gradient Penalty (WGAN-GP) to generate a balanced
+  training set of $mathbf2, 5 6 0$ images per class, while retaining the original,
+  unaugmented test set. A fine-tuned EfficientNetV2B0 model achieved 99 % accuracy
+  on the augmented training data; however, when evaluated under real-world challenges,
+  performance dropped to $mathbf4 4. 3 3 %$-significantly lower than the 69.66 % achieved
+  in the brain tumor benchmark. This stark contrast underscores the inherent difficulty
+  of AD classification, exacerbated by low-resolution axial-only scans and minimal
+  interclass visual differences. Although the robustness methodology yielded partial
+  performance recovery, the results highlight the need for higher-resolution, multi-planar
+  imaging and domain-specific preprocessing strategies. This study advances the understanding
+  of robustness in DL-based neuroimaging and underscores the importance of tailored
+  augmentation pipelines for reliable Alzheimer's diagnosis in clinical settings.
+tags:
+- Deep learning
+- Training
+- Visualization
+- Magnetic resonance imaging
+- Training data
+- Generative adversarial networks
+- Brain modeling
+- Robustness
+- Alzheimer's disease
+- Artificial intelligence
+- MRI Alzheimer Disease
+- MRI Brain Cancer
+- Deep Learning (DL)
+- Artificial Intelligence (AI)
+- Real-World simulation
+- Robustness
+- Generative Adversarial Networks (GAN)
+- Gaussian Augmentation
+links:
+- name: URL
+  url: https://ieeexplore.ieee.org/abstract/document/11211705
+---

content/publication/abou-2025-samgan/cite.bib

@@ -0,0 +1,15 @@
+@inproceedings{abou2025samgan,
+ abstract = {Recent advances in deep learning have significantly improved skin cancer classification, yet persistent issues such as extreme class imbalance and limited intra-class diversity hinder clinical generalization. This study builds upon the Naturalize framework by proposing a robust augmentation pipeline targeting rare lesion types in the ISIC 2019 dataset. Leveraging SAM-based segmentation and StyleGAN2-ADA synthesis, we construct two fully balanced datasets with $\text2, 4 0 0$ and $\text7, 2 0 0$ images per class. The method enhances representation of underrepresented classes (DF, VASC, SCC) while preserving contextual realism through background blending and geometric transformations. Across eight pre-trained CNN and ViT models, Xception trained on the 7.2 k dataset achieved the best generalization with 66% test accuracy and macro F1-score of $\text0. 6 4$. Despite challenges in synthetic fidelity (e.g., FID scores of 62 for VASC, 58 for DF), the proposed pipeline marks a significant step toward equitable, high-performing AI diagnostics. Our framework offers a reproducible, end-to-end strategy for scalable skin lesion classification under real-world data constraints.},
+ author = {Mohamad Abou Ali and Ali Rasoul Rifai and Fadi Dornaika and Ignacio Arganda-Carreras and Ali Cherry and Mohamad Hajj-Hasasn and Lara Hamawy},
+ booktitle = {2025 Eighth International Conference on Advances in Biomedical Engineering (ICABME)},
+ doi = {10.1109/ICABME66883.2025.11211545},
+ issn = {2377-5696},
+ keywords = {Deep learning, Image segmentation, Image synthesis, Pipelines, Generative adversarial networks, Transformers, Skin, Lesions, Medical diagnostic imaging, Skin cancer, Skin Cancer Classification, Deep Learning, Data Augmentation, Segment Anything Model (SAM), Generative Adversarial Networks (GAN), StyleGAN2-ADA, Class Imbalance, Medical Image Synthesis, Vision Transformer (ViT), ISIC 2019 Dataset},
+ month = {Oct},
+ number = {},
+ pages = {1-6},
+ title = {SAM-GAN Synthesis for Equitable Skin Cancer Diagnosis: A Deep Learning Approach to Address Class Imbalance in Dermoscopy},
+ url = {https://ieeexplore.ieee.org/abstract/document/11211545},
+ volume = {},
+ year = {2025}
+}

content/publication/abou-2025-samgan/index.md

@@ -0,0 +1,58 @@
+---
+title: 'SAM-GAN Synthesis for Equitable Skin Cancer Diagnosis: A Deep Learning Approach
+  to Address Class Imbalance in Dermoscopy'
+authors:
+- Mohamad Abou Ali
+- Ali Rasoul Rifai
+- Fadi Dornaika
+- Ignacio Arganda-Carreras
+- Ali Cherry
+- Mohamad Hajj-Hasasn
+- Lara Hamawy
+date: '2025-10-01'
+publishDate: '2026-04-13T10:35:42.250369Z'
+publication_types:
+- paper-conference
+publication: '*2025 Eighth International Conference on Advances in Biomedical Engineering
+  (ICABME)*'
+doi: 10.1109/ICABME66883.2025.11211545
+abstract: Recent advances in deep learning have significantly improved skin cancer
+  classification, yet persistent issues such as extreme class imbalance and limited
+  intra-class diversity hinder clinical generalization. This study builds upon the
+  Naturalize framework by proposing a robust augmentation pipeline targeting rare
+  lesion types in the ISIC 2019 dataset. Leveraging SAM-based segmentation and StyleGAN2-ADA
+  synthesis, we construct two fully balanced datasets with $text2, 4 0 0$ and $text7,
+  2 0 0$ images per class. The method enhances representation of underrepresented
+  classes (DF, VASC, SCC) while preserving contextual realism through background blending
+  and geometric transformations. Across eight pre-trained CNN and ViT models, Xception
+  trained on the 7.2 k dataset achieved the best generalization with 66% test accuracy
+  and macro F1-score of $text0. 6 4$. Despite challenges in synthetic fidelity (e.g.,
+  FID scores of 62 for VASC, 58 for DF), the proposed pipeline marks a significant
+  step toward equitable, high-performing AI diagnostics. Our framework offers a reproducible,
+  end-to-end strategy for scalable skin lesion classification under real-world data
+  constraints.
+tags:
+- Deep learning
+- Image segmentation
+- Image synthesis
+- Pipelines
+- Generative adversarial networks
+- Transformers
+- Skin
+- Lesions
+- Medical diagnostic imaging
+- Skin cancer
+- Skin Cancer Classification
+- Deep Learning
+- Data Augmentation
+- Segment Anything Model (SAM)
+- Generative Adversarial Networks (GAN)
+- StyleGAN2-ADA
+- Class Imbalance
+- Medical Image Synthesis
+- Vision Transformer (ViT)
+- ISIC 2019 Dataset
+links:
+- name: URL
+  url: https://ieeexplore.ieee.org/abstract/document/11211545
+---

content/publication/iglesias-2026-synthetic/cite.bib

@@ -0,0 +1,11 @@
+@article{iglesias2026synthetic,
+ abstract = {Autonomous driving is advancing rapidly, fueled by improvements in sensor technology and artificial intelligence (AI). Lidar-captured point clouds are crucial for these systems, offering precise spatial data and detailed environmental representations. Data are essential for developing and validating robust AI-based systems. However, obtaining real-world data is costly, and capturing specific scenarios, especially dangerous or rare cases, is impractical. As a result, synthetic point cloud generation is emerging as a promising alternative. Despite increasing interest and techniques, there is no comprehensive survey on synthetic point cloud generation for autonomous driving. This article addresses this gap by exploring the challenges in generating synthetic point clouds, outlining the methods used, and cataloging available datasets. This survey aims to provide a foundational understanding of the state of the art in synthetic point cloud generation, promoting further research in this critical area of autonomous driving technology.},
+ author = {Aitor Iglesias and Nerea Aranjuelo and Ignacio Arganda-Carreras and Marcos Nieto},
+ doi = {10.1109/MITS.2026.3653178},
+ journal = {IEEE Intelligent Transportation Systems Magazine},
+ pages = {2--24},
+ publisher = {IEEE},
+ title = {Synthetic Point Cloud Generation for Autonomous Driving: A Survey and Taxonomy},
+ url = {https://ieeexplore.ieee.org/abstract/document/11371313},
+ year = {2026}
+}

content/publication/iglesias-2026-synthetic/index.md

@@ -0,0 +1,30 @@
+---
+title: 'Synthetic Point Cloud Generation for Autonomous Driving: A Survey and Taxonomy'
+authors:
+- Aitor Iglesias
+- Nerea Aranjuelo
+- Ignacio Arganda-Carreras
+- Marcos Nieto
+date: '2026-01-01'
+publishDate: '2026-04-13T10:35:42.222429Z'
+publication_types:
+- article-journal
+publication: '*IEEE Intelligent Transportation Systems Magazine*'
+doi: 10.1109/MITS.2026.3653178
+abstract: Autonomous driving is advancing rapidly, fueled by improvements in sensor
+  technology and artificial intelligence (AI). Lidar-captured point clouds are crucial
+  for these systems, offering precise spatial data and detailed environmental representations.
+  Data are essential for developing and validating robust AI-based systems. However,
+  obtaining real-world data is costly, and capturing specific scenarios, especially
+  dangerous or rare cases, is impractical. As a result, synthetic point cloud generation
+  is emerging as a promising alternative. Despite increasing interest and techniques,
+  there is no comprehensive survey on synthetic point cloud generation for autonomous
+  driving. This article addresses this gap by exploring the challenges in generating
+  synthetic point clouds, outlining the methods used, and cataloging available datasets.
+  This survey aims to provide a foundational understanding of the state of the art
+  in synthetic point cloud generation, promoting further research in this critical
+  area of autonomous driving technology.
+links:
+- name: URL
+  url: https://ieeexplore.ieee.org/abstract/document/11371313
+---