Hadi Fayad
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Hadi Fayad has a total of 17 co-authors. Here are the top co-authors by number of shared publications:
- Tarraf Torfeh
Department of Radiation Oncology, National Center for Cancer Care and Research, Hamad Medical Corporation, Doha 3050, Qatar
2 shared publications - Alaa Abd Al-Rahim
Hamad Medical Corporation
1 shared publication Alaa El Khatib
Hamad Medical Corporation, Qatar
1 shared publicationAmer Ghujeh
Hamad Medical Corporation, Qatar
1 shared publication- Amer Ghujeh
Hamad Medical Corporation
1 shared publication Antar Aly
Hamad Medical Corporation, Qatar
1 shared publication
Latest Publications
- Open Access
- 23 October 2026
- Published by Springer Nature in BMC Biomedical Engineering
- Vol. 8 (1) , 19
A web modular AI interface for brain tumor MRI classification using deep learning models
In order to improve patient outcomes and treatment processes, early and precise diagnosis of brain tumors is crucial. Despite the strong performance shown by deep learning models in classifying brain tumors from MRI scans, integrating these models into user-friendly clinical workflows remains challenging. This study presents a web-based AI platform designed for brain tumor classification using MRI images. The platform enables users to upload MRI scans, apply standardized preprocessing, select between different deep learning models, and receive real-time predictions. A custom convolutional neural network (CNN), pretrained VGG19, and pretrained Xception were evaluated using a publicly available T1-weighted contrast-enhanced brain MRI dataset of approximately 7,000 images across four classes: glioma, meningioma, pituitary tumor, and no tumor. The dataset lacks patient demographic and clinical metadata, limiting assessment of clinical generalizability. The models achieved test accuracies of 96.03%, 98%, and 96%, respectively, with VGG19 demonstrating the best performance. The study is limited by the use of a single public, slice-based dataset and the absence of external clinical validation; future work will focus on multi-sequence MRI and independent clinical cohorts.
- 23 October 2026
- Published by Springer Nature in BMC Biomedical Engineering
- Vol. 8 (1) , 19
Open AccessA web modular AI interface for brain tumor MRI classification using deep learning models
In order to improve patient outcomes and treatment processes, early and precise diagnosis of brain tumors is crucial. Despite the strong performance shown by deep learning models in classifying brain tumors from MRI scans, integrating these models into user-friendly clinical workflows remains challenging. This study presents a web-based AI platform designed for brain tumor classification using MRI images. The platform enables users to upload MRI scans, apply standardized preprocessing, select between different deep learning models, and receive real-time predictions. A custom convolutional neural network (CNN), pretrained VGG19, and pretrained Xception were evaluated using a publicly available T1-weighted contrast-enhanced brain MRI dataset of approximately 7,000 images across four classes: glioma, meningioma, pituitary tumor, and no tumor. The dataset lacks patient demographic and clinical metadata, limiting assessment of clinical generalizability. The models achieved test accuracies of 96.03%, 98%, and 96%, respectively, with VGG19 demonstrating the best performance. The study is limited by the use of a single public, slice-based dataset and the absence of external clinical validation; future work will focus on multi-sequence MRI and independent clinical cohorts.
- 2 August 2026
- Published by Springer Nature
- p. 690-704
AI-Driven PET Attenuation Correction: Eliminating the Need for CT
- 2 August 2026
- Published by Springer Nature
- p. 690-704
AI-Driven PET Attenuation Correction: Eliminating the Need for CT
- 30 August 2022
- Published by Society of Nuclear Medicine in Journal of Nuclear Medicine Technology
- Vol. 51 (1) , 63-67
National Diagnostic Reference Levels for Nuclear Medicine in Qatar
Nuclear medicine (NM) started in Qatar in mid 1980s with a one head gamma camera in Hamad General Hospital (HGH). However, Qatar is expanding and now Hamad Medical Corporation (HMC) has two NM departments and one Positron Emission Tomography (PET)/CT Center for Diagnosis and Research with several hybrid single photon emission computed tomography (SPECT)/CT and PET/CT. Furthermore, two new NM departments will be established in Qatar in the coming three years. Therefore, there is a need for optimizing radiation protection in NM imaging and establishing diagnostic reference levels (DRLs) for the first time in Qatar. This need, is not only for the NM part of the examination but also for the computed tomography (CT) part especially in hybrid SPECT/CT and PET/CT. Methods: Data was collected for adult patients from the 3 SPECT/CT machines in the two NM facilities and from the 2 PET/CT machines in the PET/CT Center. The 75th percentile values (also known as the third quartile Q3) were considered as preliminary DRL values and they were consistent with the most commonly administered activities. DRL results for NM imaging in Qatar adults are described including various general NM protocols especially Technetium-99m (99mTc) based radiopharmaceuticals and PET/CT protocols including mainly oncology applications. Results: First DRLs for NM imaging in Qatar adults were established; values are in agreement with other published DRLs as it is the case for example for PET Oncology using 18F- fluorodeoxyglucose (FDG) with DRLs of 258, 230, 370, 400 and 461–710 MBq for Qatar, Kuwait, Korea, UK and USA, respectively. Similarly, for NM Cardiac Stress or Rest Myocardial Perfusion imaging using99mTc Methoxy Isobutyl Isonitrile (MIBI), DRLs were 926, 976, 1110, 800 and 945-1402 MBq for Qatar, Kuwait, Korea, UK and USA, respectively. Conclusion: This study will enable administrated activity optimization for NM procedures in Qatar and this will be of great value especially for the new departments that will adhere to these DRLs.
- 30 August 2022
- Published by Society of Nuclear Medicine in Journal of Nuclear Medicine Technology
- Vol. 51 (1) , 63-67
National Diagnostic Reference Levels for Nuclear Medicine in Qatar
Nuclear medicine (NM) started in Qatar in mid 1980s with a one head gamma camera in Hamad General Hospital (HGH). However, Qatar is expanding and now Hamad Medical Corporation (HMC) has two NM departments and one Positron Emission Tomography (PET)/CT Center for Diagnosis and Research with several hybrid single photon emission computed tomography (SPECT)/CT and PET/CT. Furthermore, two new NM departments will be established in Qatar in the coming three years. Therefore, there is a need for optimizing radiation protection in NM imaging and establishing diagnostic reference levels (DRLs) for the first time in Qatar. This need, is not only for the NM part of the examination but also for the computed tomography (CT) part especially in hybrid SPECT/CT and PET/CT. Methods: Data was collected for adult patients from the 3 SPECT/CT machines in the two NM facilities and from the 2 PET/CT machines in the PET/CT Center. The 75th percentile values (also known as the third quartile Q3) were considered as preliminary DRL values and they were consistent with the most commonly administered activities. DRL results for NM imaging in Qatar adults are described including various general NM protocols especially Technetium-99m (99mTc) based radiopharmaceuticals and PET/CT protocols including mainly oncology applications. Results: First DRLs for NM imaging in Qatar adults were established; values are in agreement with other published DRLs as it is the case for example for PET Oncology using 18F- fluorodeoxyglucose (FDG) with DRLs of 258, 230, 370, 400 and 461–710 MBq for Qatar, Kuwait, Korea, UK and USA, respectively. Similarly, for NM Cardiac Stress or Rest Myocardial Perfusion imaging using99mTc Methoxy Isobutyl Isonitrile (MIBI), DRLs were 926, 976, 1110, 800 and 945-1402 MBq for Qatar, Kuwait, Korea, UK and USA, respectively. Conclusion: This study will enable administrated activity optimization for NM procedures in Qatar and this will be of great value especially for the new departments that will adhere to these DRLs.
- 3 June 2018
- Published by Wiley in Medical Physics
- Vol. 45 (7) , 3043-3051
A 4D global respiratory motion model of the thorax based on CT images: A proof of concept
Purpose: Respiratory motion reduces the sensitivity and specificity of medical images especially in the thoracic and abdominal areas. It may affect applications such as cancer diagnostic imaging and/or radiation therapy (RT). Solutions to this issue include modeling of the respiratory motion in order to optimize both diagnostic and therapeutic protocols. Personalized motion modeling required patient‐specific four‐dimensional (4D) imaging which in the case of 4D computed tomography (4D CT) acquisition is associated with an increased dose. The goal of this work was to develop a global respiratory motion model capable of relating external patient surface motion to internal structure motion without the need for a patient‐specific 4D CT acquisition.Methods: The proposed global model is based on principal component analysis and can be adjusted to a given patient anatomy using only one or two static CT images in conjunction with a respiratory synchronized patient external surface motion. It is based on the relation between the internal motion described using deformation fields obtained by registering 4D CT images and patient surface maps obtained either from optical imaging devices or extracted from CT image‐based patient skin segmentation. 4D CT images of six patients were used to generate the global motion model which was validated by adapting it on four different patients having skin segmented surfaces and two other patients having time of flight camera acquired surfaces. The reproducibility of the proposed model was also assessed on two patients with two 4D CT series acquired within 2 weeks of each other.Results: Profile comparison shows the efficacy of the global respiratory motion model and an improvement while using two CT images in order to adapt the model. This was confirmed by the correlation coefficient with a mean correlation of 0.9 and 0.95 while using one or two CT images respectively and when comparing acquired to model generated 4D CT images. For the four patients with segmented surfaces, expert validation indicates an error of 2.35 ± 0.26 mm compared to 6.07 ± 0.76 mm when using a simple interpolation between full inspiration (FI) and full expiration (FE) CT only; i.e., without specific modeling of the respiratory motion. For the two patients with acquired surfaces, this error was of 2.48 ± 0.18 mm. In terms of reproducibility, model error changes of 0.12 and 0.17 mm were measured for the two patients concerned.Conclusions: The framework for the derivation of a global respiratory motion model was developed. A single or two static CT images and associated patient surface motion, as a surrogate measure, are only needed to personalize the model. This model accuracy and reproducibility were assessed by comparing acquired vs model generated 4D CT images. Future work will consist of assessing extensively the proposed model for radiotherapy applications.
- 3 June 2018
- Published by Wiley in Medical Physics
- Vol. 45 (7) , 3043-3051
A 4D global respiratory motion model of the thorax based on CT images: A proof of concept
Purpose: Respiratory motion reduces the sensitivity and specificity of medical images especially in the thoracic and abdominal areas. It may affect applications such as cancer diagnostic imaging and/or radiation therapy (RT). Solutions to this issue include modeling of the respiratory motion in order to optimize both diagnostic and therapeutic protocols. Personalized motion modeling required patient‐specific four‐dimensional (4D) imaging which in the case of 4D computed tomography (4D CT) acquisition is associated with an increased dose. The goal of this work was to develop a global respiratory motion model capable of relating external patient surface motion to internal structure motion without the need for a patient‐specific 4D CT acquisition.Methods: The proposed global model is based on principal component analysis and can be adjusted to a given patient anatomy using only one or two static CT images in conjunction with a respiratory synchronized patient external surface motion. It is based on the relation between the internal motion described using deformation fields obtained by registering 4D CT images and patient surface maps obtained either from optical imaging devices or extracted from CT image‐based patient skin segmentation. 4D CT images of six patients were used to generate the global motion model which was validated by adapting it on four different patients having skin segmented surfaces and two other patients having time of flight camera acquired surfaces. The reproducibility of the proposed model was also assessed on two patients with two 4D CT series acquired within 2 weeks of each other.Results: Profile comparison shows the efficacy of the global respiratory motion model and an improvement while using two CT images in order to adapt the model. This was confirmed by the correlation coefficient with a mean correlation of 0.9 and 0.95 while using one or two CT images respectively and when comparing acquired to model generated 4D CT images. For the four patients with segmented surfaces, expert validation indicates an error of 2.35 ± 0.26 mm compared to 6.07 ± 0.76 mm when using a simple interpolation between full inspiration (FI) and full expiration (FE) CT only; i.e., without specific modeling of the respiratory motion. For the two patients with acquired surfaces, this error was of 2.48 ± 0.18 mm. In terms of reproducibility, model error changes of 0.12 and 0.17 mm were measured for the two patients concerned.Conclusions: The framework for the derivation of a global respiratory motion model was developed. A single or two static CT images and associated patient surface motion, as a surrogate measure, are only needed to personalize the model. This model accuracy and reproducibility were assessed by comparing acquired vs model generated 4D CT images. Future work will consist of assessing extensively the proposed model for radiotherapy applications.