Ibtihaj Ahmad
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Ibtihaj Ahmad has a total of 39 co-authors. Here are the top co-authors by number of shared publications:
- Fuzhong Xue
Department of Medical Dataology, School of Public Health, Cheeloo College of Medicine, Shandong University, China
3 shared publications Zain Ul Islam
Hamad Bin Khalifa University, Education City,ICT Division, College of Science Engineering,Doha,Qatar
3 shared publications- Zain Ul Islam
Division of Information and Computing Technology, College of Science and Engineering, Hamad Bin Khalifa University, Doha, Qatar
3 shared publications - Hengfei Cui
Northwestern Polytechnical University
2 shared publications Muhammad Abbas Hussain
College of EME, National University of Sciences and Technology (NUST), Pakistan
2 shared publicationsSadia Jabbar Anwar
Northwestern Polytechnical University
2 shared publications
Latest Publications
- 24 April 2026
- Published by Springer Nature in Health Information Science and Systems
- Vol. 14 (1) , 61
From EHR chaos to clinical clarity: enhancing foundational encoders with task-specific attention for domain-oriented representation learning in downstream clinical applications
Electronic health records (EHRs) contain rich clinical information; however, they pose challenges for representation learning due to long free-text notes, domain shift, irregular structure, incomplete, and sparse fields. To address these challenges, we propose a lightweight, encoder-agnostic framework that segments each record into clinically meaningful sections, encodes them with a shared foundation-model encoder, stabilizes features via upper-layer mixing, and aggregates sections with a task-specific attention head cast as a permutation-invariant set function. This section-aware design improves truncation from token limits without specialized long-text engineering and focuses the model on task-relevant evidence. We demonstrate consistent improvements across seven baseline encoders and three downstream applications, i.e., disease prediction, clustering, and representational digital-twin (RDT) retrieval. For prediction on large corpora, section-aware aggregation improves accuracy and F1 and shifts per-disease ROC curves toward higher AUC. Clustering quality increases across ARI, homogeneity, completeness, and V-measure, indicating more coherent patient strata. In RDT retrieval, neighborhoods become more label-consistent (higher homogeneity and Concordance) while maintaining high nearest-neighbor similarity. Ablations show that mixing upper transformer layers combined with task-specific section attention improves performance and reduces cross-seed variance with a modest calibration trade-off. Overall, our framework produces portable, interpretable patient embeddings that may help in various downstream analytics and decision support in real-world clinical settings.
- 24 April 2026
- Published by Springer Nature in Health Information Science and Systems
- Vol. 14 (1) , 61
From EHR chaos to clinical clarity: enhancing foundational encoders with task-specific attention for domain-oriented representation learning in downstream clinical applications
Electronic health records (EHRs) contain rich clinical information; however, they pose challenges for representation learning due to long free-text notes, domain shift, irregular structure, incomplete, and sparse fields. To address these challenges, we propose a lightweight, encoder-agnostic framework that segments each record into clinically meaningful sections, encodes them with a shared foundation-model encoder, stabilizes features via upper-layer mixing, and aggregates sections with a task-specific attention head cast as a permutation-invariant set function. This section-aware design improves truncation from token limits without specialized long-text engineering and focuses the model on task-relevant evidence. We demonstrate consistent improvements across seven baseline encoders and three downstream applications, i.e., disease prediction, clustering, and representational digital-twin (RDT) retrieval. For prediction on large corpora, section-aware aggregation improves accuracy and F1 and shifts per-disease ROC curves toward higher AUC. Clustering quality increases across ARI, homogeneity, completeness, and V-measure, indicating more coherent patient strata. In RDT retrieval, neighborhoods become more label-consistent (higher homogeneity and Concordance) while maintaining high nearest-neighbor similarity. Ablations show that mixing upper transformer layers combined with task-specific section attention improves performance and reduces cross-seed variance with a modest calibration trade-off. Overall, our framework produces portable, interpretable patient embeddings that may help in various downstream analytics and decision support in real-world clinical settings.
- 1 September 2025
- Published by Springer Nature
- p. 243-256
Attention Based Tri-Decoder Framework for Segmentation and Classification of Nuclei in Bladder Histology
- 1 September 2025
- Published by Springer Nature
- p. 243-256
Attention Based Tri-Decoder Framework for Segmentation and Classification of Nuclei in Bladder Histology
- Open Access
- 9 April 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-13
Anatomy guided modality fusion for cancer segmentation in PET CT volumes and images
Segmentation in computed tomography (CT) provides detailed anatomical information, while positron emission tomography (PET) provide the metabolic activity of cancer. Existing segmentation models in CT and PET either rely on early fusion, which struggles to effectively capture independent features from each modality, or late fusion, which is computationally expensive and fails to leverage the complementary nature of the two modalities. This research addresses the gap by proposing an intermediate fusion approach that optimally balances the strengths of both modalities. Our method leverages anatomical features to guide the fusion process while preserving spatial representation quality. We achieve this through the separate encoding of anatomical and metabolic features followed by an attentive fusion decoder. Unlike traditional fixed normalization techniques, we introduce novel “zero layers” with learnable normalization. The proposed intermediate fusion reduces the number of filters, resulting in a lightweight model. Our approach demonstrates superior performance, achieving a dice score of 0.8184 and an $$\hbox {HD}^{95}$$ score of 2.31. The implications of this study include more precise tumor delineation, leading to enhanced cancer diagnosis and more effective treatment planning.- 9 April 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-13
Open AccessAnatomy guided modality fusion for cancer segmentation in PET CT volumes and images
Segmentation in computed tomography (CT) provides detailed anatomical information, while positron emission tomography (PET) provide the metabolic activity of cancer. Existing segmentation models in CT and PET either rely on early fusion, which struggles to effectively capture independent features from each modality, or late fusion, which is computationally expensive and fails to leverage the complementary nature of the two modalities. This research addresses the gap by proposing an intermediate fusion approach that optimally balances the strengths of both modalities. Our method leverages anatomical features to guide the fusion process while preserving spatial representation quality. We achieve this through the separate encoding of anatomical and metabolic features followed by an attentive fusion decoder. Unlike traditional fixed normalization techniques, we introduce novel “zero layers” with learnable normalization. The proposed intermediate fusion reduces the number of filters, resulting in a lightweight model. Our approach demonstrates superior performance, achieving a dice score of 0.8184 and an $$\hbox {HD}^{95}$$ score of 2.31. The implications of this study include more precise tumor delineation, leading to enhanced cancer diagnosis and more effective treatment planning. - Open Access
- 20 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 173530-173539
MANS-Net: Multiple Attention-Based Nuclei Segmentation in Multi Organ Digital Cancer Histopathology Images
The segmentation of nuclei is critical in histopathology investigations. The segmentation of images of nuclei is difficult in variable clinical conditions. Some deep learning methods were recently proposed; however, these approaches rarely provide solutions to clinical challenges. Since most of the information in histopathology images is in the color channels, and the remaining is in the spatial patterns, distributions, etc., these problems can be handled by considering all the associated features simultaneously. This work suggests a novel multiple attention-based model, MANS-Net, that utilizes channel, spatial, and transformer-based attention modules to address the problems mentioned above. MANS-Net efficiently learns color-based, spatial, rough, and granular features, improving nuclei segmentation. We report that MANS-Net significantly outperforms state-of-the-art segmentation algorithms. We achieve F1 score of 0.8457 for PanNuke dataset and F1 score of 0.8137 for Kumar dataset. We show that the suggested model is more lightweight than the state-of-the-art approaches. The suggested model will impact future works that depend on semantic segmentation, such as nuclei instance segmentation and nuclei categorization.
- 20 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 173530-173539
Open AccessMANS-Net: Multiple Attention-Based Nuclei Segmentation in Multi Organ Digital Cancer Histopathology Images
The segmentation of nuclei is critical in histopathology investigations. The segmentation of images of nuclei is difficult in variable clinical conditions. Some deep learning methods were recently proposed; however, these approaches rarely provide solutions to clinical challenges. Since most of the information in histopathology images is in the color channels, and the remaining is in the spatial patterns, distributions, etc., these problems can be handled by considering all the associated features simultaneously. This work suggests a novel multiple attention-based model, MANS-Net, that utilizes channel, spatial, and transformer-based attention modules to address the problems mentioned above. MANS-Net efficiently learns color-based, spatial, rough, and granular features, improving nuclei segmentation. We report that MANS-Net significantly outperforms state-of-the-art segmentation algorithms. We achieve F1 score of 0.8457 for PanNuke dataset and F1 score of 0.8137 for Kumar dataset. We show that the suggested model is more lightweight than the state-of-the-art approaches. The suggested model will impact future works that depend on semantic segmentation, such as nuclei instance segmentation and nuclei categorization.
- 23 August 2024
- Published by Springer Nature in Applied Intelligence
- Vol. 54 (21) , 10640-10652
Deep-SEA: a deep learning based patient specific multi-modality post-cancer survival estimation architecture
- 23 August 2024
- Published by Springer Nature in Applied Intelligence
- Vol. 54 (21) , 10640-10652
Deep-SEA: a deep learning based patient specific multi-modality post-cancer survival estimation architecture
- 12 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Complexity Analysis of Industrial Scale Cyber Physical Systems
Cyber-Physical Systems (CPS) are one of the emerging technologies of the time. Current CPS are categorized as small, medium, and large-scale CPS. Small and medium-scale CPS’s design, development, and operation standards are well-defined. However, the large-scale, also known as industrial-scale CPS, are complex, and their design, development, and operation remain challenging. This is due to the involvement of multiple complex technical and non-technical domains. This research discusses the complexity of future industrial-scale CPS in terms of cyber systems, physical systems, Collaborative Information Processing Systems (CIPS), and computer-aided engineering (CAE) systems. We present a detailed overview of the complexities in cyber & physical systems, human and managerial systems, and CAEs and suggest possible solutions. We discuss these complexities via a case study on a CPS-based energy management system. The implications of our discussions include an improved pathway to reduce the complexity of future CPS.
- 12 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Complexity Analysis of Industrial Scale Cyber Physical Systems
Cyber-Physical Systems (CPS) are one of the emerging technologies of the time. Current CPS are categorized as small, medium, and large-scale CPS. Small and medium-scale CPS’s design, development, and operation standards are well-defined. However, the large-scale, also known as industrial-scale CPS, are complex, and their design, development, and operation remain challenging. This is due to the involvement of multiple complex technical and non-technical domains. This research discusses the complexity of future industrial-scale CPS in terms of cyber systems, physical systems, Collaborative Information Processing Systems (CIPS), and computer-aided engineering (CAE) systems. We present a detailed overview of the complexities in cyber & physical systems, human and managerial systems, and CAEs and suggest possible solutions. We discuss these complexities via a case study on a CPS-based energy management system. The implications of our discussions include an improved pathway to reduce the complexity of future CPS.
- 11 March 2023
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 157, 106748
AATSN: Anatomy Aware Tumor Segmentation Network for PET-CT volumes and images using a lightweight fusion-attention mechanism
- 11 March 2023
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 157, 106748
AATSN: Anatomy Aware Tumor Segmentation Network for PET-CT volumes and images using a lightweight fusion-attention mechanism