Kamrul Hasan
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
Kamrul Hasan has a total of 89 co-authors. Here are the top co-authors by number of shared publications:
Soo Yeol Lee
Department of Biomedical Engineering
6 shared publicationsSharmin R. Ara
5 shared publications- Shoaib Akhter Rafi5 shared publications
- Soo Yeol Lee
Department of Biomedical Engineering, Kyung Hee University, Yongin, Korea
5 shared publications - Uday Kamal
Department of Electrical and Electronic Engineering, Bangladesh University of Engineering and Technology, Dhaka, Bangladesh
5 shared publications Abdul Muntakim Rafi
University of Windsor
4 shared publications
Latest Publications
- Open Access
- 28 September 2026
- Published by Elsevier in SSRN Electronic Journal
A Deep Learning-Enabled Rapid Battery Virtual Testing Framework for State-of-Health and Remaining Useful Life Estimation
The increasing deployment of lithium-ion batteries in electric vehicles (EVs) and energy storage systems necessitates rapid, reliable, and scalable methods for assessing battery health. Conventional techniques for estimating state of health (SoH) and remaining useful life (RUL) typically require full-cycle data and extended testing durations, limiting their practical applicability in real-world scenarios. This work presents a rapid battery virtual testing framework with improved cross-dataset robustness that enables accurate SoH estimation and feasibility-level RUL assessment using short-duration charging data. The proposed framework integrates an attention-based bidirectional long short-term memory (Att-BiLSTM) model with a unified data representation strategy to support learning across heterogeneous battery datasets. The model is trained and evaluated using publicly available datasets from NASA and Oxford, encompassing diverse operating conditions and degradation behaviors. Results demonstrate that reliable SoH estimation can be achieved using charging segments as short as 5--10 minutes, with mean absolute percentage errors of approximately 1--2% for Oxford-type datasets and 3--5% for NASA-type datasets under homogeneous and combined training scenarios. The use of combined datasets further improves robustness under cross-dataset testing conditions. To assess practical applicability, a prototype experimental measurement system is developed and tested on lithium-ion cells. Despite additional variability in experimental conditions, the framework maintains reasonable prediction accuracy, with errors on the order of ~6%, supporting proof-of-concept applicability beyond controlled datasets. A simple extrapolation-based approach is further employed to estimate RUL from predicted SoH trajectories, illustrating the feasibility of rapid lifecycle assessment. The findings highlight the potential of the proposed framework as a scalable approach for rapid battery health screening, with applications in EV maintenance, battery management systems, and second-life battery evaluation.
- 28 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessA Deep Learning-Enabled Rapid Battery Virtual Testing Framework for State-of-Health and Remaining Useful Life Estimation
The increasing deployment of lithium-ion batteries in electric vehicles (EVs) and energy storage systems necessitates rapid, reliable, and scalable methods for assessing battery health. Conventional techniques for estimating state of health (SoH) and remaining useful life (RUL) typically require full-cycle data and extended testing durations, limiting their practical applicability in real-world scenarios. This work presents a rapid battery virtual testing framework with improved cross-dataset robustness that enables accurate SoH estimation and feasibility-level RUL assessment using short-duration charging data. The proposed framework integrates an attention-based bidirectional long short-term memory (Att-BiLSTM) model with a unified data representation strategy to support learning across heterogeneous battery datasets. The model is trained and evaluated using publicly available datasets from NASA and Oxford, encompassing diverse operating conditions and degradation behaviors. Results demonstrate that reliable SoH estimation can be achieved using charging segments as short as 5--10 minutes, with mean absolute percentage errors of approximately 1--2% for Oxford-type datasets and 3--5% for NASA-type datasets under homogeneous and combined training scenarios. The use of combined datasets further improves robustness under cross-dataset testing conditions. To assess practical applicability, a prototype experimental measurement system is developed and tested on lithium-ion cells. Despite additional variability in experimental conditions, the framework maintains reasonable prediction accuracy, with errors on the order of ~6%, supporting proof-of-concept applicability beyond controlled datasets. A simple extrapolation-based approach is further employed to estimate RUL from predicted SoH trajectories, illustrating the feasibility of rapid lifecycle assessment. The findings highlight the potential of the proposed framework as a scalable approach for rapid battery health screening, with applications in EV maintenance, battery management systems, and second-life battery evaluation.
- 3 August 2026
- Published by IOP Publishing in Physics in Medicine & Biology
- Vol. 71 (15) , 155012
GCL-RCA-Net: an RCA-RemUNet-enhanced graph-contrastive framework for hybrid data-driven and dual-branch physics-guided parallel MRI reconstruction
Objective. Parallel magnetic resonance imaging (pMRI) with Cartesian equispaced undersampling accelerates acquisition but introduces structured aliasing that degrades reconstruction quality, especially at high acceleration factors. Although recent self-supervised and contrastive reconstruction methods have shown promise, most operate slice-wise and do not capture cross-slice anatomical continuity or align learned representations with physics-guided reconstruction. Approach. To address these limitations, we propose GCL-RCA-Net, an RCA-RemUNet-enhanced graph-contrastive framework for pMRI reconstruction under Cartesian equispaced undersampling. The method combines measurement-consistent data-driven initialization, a physics-guided dual-branch module, anatomy-aware graph contrastive priors, and multi-stage cross-domain refinement. The initial stage stabilizes reconstruction by directly restoring acquired k -space samples and learning missing data. A dual-branch module then exploits complementary physics through calibration-based generalized autocalibrating partially parallel acquisitions interpolation in k -space and sensitivity-encoding-based conjugate gradient refinement in the image domain, followed by adaptive branch selection and fusion. To model anatomical context, a self-supervised graph contrastive module encodes cross-slice continuity and non-local structural relationships, providing descriptors that guide the cascaded refinement network. Finally, a multi-stage lattice-style cross-domain fusion module jointly refines the data-driven estimate and fused physics output through coupled image- and k -space pathways, followed by final data consistency enforcement. Results. On fastMRI brain data with acceleration factors from 2 × to 8 × , and in cross-anatomy evaluation on fastMRI knee data, GCL-RCA-Net consistently outperforms recent baselines under equispaced Cartesian undersampling. Relative to the strongest competing method, it achieves PSNR gains of 2.66–4.35 dB on brain and 0.31–3.03 dB on knee reconstruction, with corresponding improvements in structural similarity. Significance. The proposed framework shows that combining measurement-consistent data-driven estimation, physics-guided dual-branch refinement, and anatomy-aware graph contrastive priors yields robust pMRI reconstruction under severe undersampling, with strong potential for faster, higher-quality, and clinically reliable MRI.
- 3 August 2026
- Published by IOP Publishing in Physics in Medicine & Biology
- Vol. 71 (15) , 155012
GCL-RCA-Net: an RCA-RemUNet-enhanced graph-contrastive framework for hybrid data-driven and dual-branch physics-guided parallel MRI reconstruction
Objective. Parallel magnetic resonance imaging (pMRI) with Cartesian equispaced undersampling accelerates acquisition but introduces structured aliasing that degrades reconstruction quality, especially at high acceleration factors. Although recent self-supervised and contrastive reconstruction methods have shown promise, most operate slice-wise and do not capture cross-slice anatomical continuity or align learned representations with physics-guided reconstruction. Approach. To address these limitations, we propose GCL-RCA-Net, an RCA-RemUNet-enhanced graph-contrastive framework for pMRI reconstruction under Cartesian equispaced undersampling. The method combines measurement-consistent data-driven initialization, a physics-guided dual-branch module, anatomy-aware graph contrastive priors, and multi-stage cross-domain refinement. The initial stage stabilizes reconstruction by directly restoring acquired k -space samples and learning missing data. A dual-branch module then exploits complementary physics through calibration-based generalized autocalibrating partially parallel acquisitions interpolation in k -space and sensitivity-encoding-based conjugate gradient refinement in the image domain, followed by adaptive branch selection and fusion. To model anatomical context, a self-supervised graph contrastive module encodes cross-slice continuity and non-local structural relationships, providing descriptors that guide the cascaded refinement network. Finally, a multi-stage lattice-style cross-domain fusion module jointly refines the data-driven estimate and fused physics output through coupled image- and k -space pathways, followed by final data consistency enforcement. Results. On fastMRI brain data with acceleration factors from 2 × to 8 × , and in cross-anatomy evaluation on fastMRI knee data, GCL-RCA-Net consistently outperforms recent baselines under equispaced Cartesian undersampling. Relative to the strongest competing method, it achieves PSNR gains of 2.66–4.35 dB on brain and 0.31–3.03 dB on knee reconstruction, with corresponding improvements in structural similarity. Significance. The proposed framework shows that combining measurement-consistent data-driven estimation, physics-guided dual-branch refinement, and anatomy-aware graph contrastive priors yields robust pMRI reconstruction under severe undersampling, with strong potential for faster, higher-quality, and clinically reliable MRI.
- Open Access
- 26 February 2026
- Published by Elsevier in Healthcare Analytics
- Vol. 9
An explainable artificial intelligence framework for trustworthy arrhythmia detection from heart rhythm data
- 26 February 2026
- Published by Elsevier in Healthcare Analytics
- Vol. 9
Open AccessAn explainable artificial intelligence framework for trustworthy arrhythmia detection from heart rhythm data
- Open Access
- 10 February 2026
- Published by Elsevier in SSRN Electronic Journal
Multivariate feature-disentangled graph-enforced domain adaptation with multi-objective deep learning for remaining useful life prediction of lithium-ion batteries
Accurate estimation of the remaining useful life (RUL) of lithium-ion batteries is widely recognized as a cornerstone for ensuring reliability, safety, and cost-effectiveness in next‑generation energy storage systems, yet remains challenging due to the complex, nonlinear nature of battery degradation. Existing approaches to battery RUL prognostics are hindered by three persistent domain gaps: (i) models often isolate macroscopic state‑of‑health (SOH) trajectories from short‑term sensor dynamics, failing to capture their complementarity; (ii) sensor‑ and health‑driven embeddings evolve at different rates, leading to structural misalignment and fragmented latent spaces; and (iii) feature distributions shift across diverse battery chemistries, restricting transferability. To address these challenges, we present a multivariate feature‑disentangled, graph‑enforced domain adaptation framework with multi‑objective deep learning for RUL prediction. At its core, SHER‑Net fuses SOH embeddings with high‑frequency sensor representations to capture both long‑term and short‑term dynamics. Graph‑based enforced consistency learning (GECL) provides training‑time structural alignment, mitigating intra‑ and inter‑modal drift without inference overhead. A dual‑stream disentangled fusion network (DDFN) with split embedding‑based domain adaptation (SEDA) partitions the latent space into domain‑invariant and domain‑specific subspaces, thereby eliminating negative transfer and over‑alignment. Finally, a multi‑objective loss synchronizes predictive accuracy, structural consistency, and feature disentanglement, completing the adaptation cycle. Extensive evaluation on public and proprietary datasets demonstrates that this framework consistently surpasses state‑of‑the‑art methods, achieving normalized RMSE of 1.10-3.33% and establishing a new benchmark for transferable, inference‑efficient, and accurate battery prognostics.
- 10 February 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessMultivariate feature-disentangled graph-enforced domain adaptation with multi-objective deep learning for remaining useful life prediction of lithium-ion batteries
Accurate estimation of the remaining useful life (RUL) of lithium-ion batteries is widely recognized as a cornerstone for ensuring reliability, safety, and cost-effectiveness in next‑generation energy storage systems, yet remains challenging due to the complex, nonlinear nature of battery degradation. Existing approaches to battery RUL prognostics are hindered by three persistent domain gaps: (i) models often isolate macroscopic state‑of‑health (SOH) trajectories from short‑term sensor dynamics, failing to capture their complementarity; (ii) sensor‑ and health‑driven embeddings evolve at different rates, leading to structural misalignment and fragmented latent spaces; and (iii) feature distributions shift across diverse battery chemistries, restricting transferability. To address these challenges, we present a multivariate feature‑disentangled, graph‑enforced domain adaptation framework with multi‑objective deep learning for RUL prediction. At its core, SHER‑Net fuses SOH embeddings with high‑frequency sensor representations to capture both long‑term and short‑term dynamics. Graph‑based enforced consistency learning (GECL) provides training‑time structural alignment, mitigating intra‑ and inter‑modal drift without inference overhead. A dual‑stream disentangled fusion network (DDFN) with split embedding‑based domain adaptation (SEDA) partitions the latent space into domain‑invariant and domain‑specific subspaces, thereby eliminating negative transfer and over‑alignment. Finally, a multi‑objective loss synchronizes predictive accuracy, structural consistency, and feature disentanglement, completing the adaptation cycle. Extensive evaluation on public and proprietary datasets demonstrates that this framework consistently surpasses state‑of‑the‑art methods, achieving normalized RMSE of 1.10-3.33% and establishing a new benchmark for transferable, inference‑efficient, and accurate battery prognostics.
- 16 December 2025
- Published by IOP Publishing in Physics in Medicine & Biology
- Vol. 71 (1)
Robust CNN multi-nested-LSTM framework with compound loss for patch-based multi-push ultrasound shear wave imaging and segmentation
Objective: Ultrasound shear wave imaging enables noninvasive, quantitative assessment of tissue pathology with mechanical elasticity measurements. However, shear wave elastography (SWE) reconstructions are challenged by noise sensitivity, inefficient multi-push strategies for scalable ROI coverage, and limited annotated data, leading to suboptimal reconstruction and unreliable inclusion segmentation. Approach: In this work, we present a novel two-stage deep learning framework that addresses these limitations through a CNN-based Multi-Nested-LSTM reconstruction network followed by a compound-loss-driven CNN-denoiser. The reconstruction stage begins with a ResNet3D-encoder that extracts spatiotemporal features from sequential multi-push acoustic radiation force (ARF) data. These features are temporally windowed with Nested CNN-LSTM, converted from 3D to 2D with temporal attention module (TAM), and enhanced by FFT-based frequency attention. The resulting 2D maps are subsequently decoded into primary 2D elasticity reconstructions. To mitigate data-scarcity and improve generalization, a patch-based training regime is also proposed. The second stage introduces a dual-decoder denoising network that separately processes inclusion and background stiffness features, followed by a fusion module that produces a denoised modulus map and a segmentation mask. A multi-objective compound loss is designed to accommodate the denoising, fusing, and mask generation. The method is validated on sequential multi-push (simulation and experimental) SWE motion data with multiple overlapping regions. Results: The method was tested on simulated and CIRS phantom datasets with four overlapping push regions, yielding 26.33 dB PSNR, 30.73 dB CNR, and 0.813 IoU in simulation, and 22.44 dB PSNR, 36.88 dB CNR, and 0.781 IoU experimentally. Evaluation on an ex vivo swine liver confirmed elasticity estimates within reported biological stiffness ranges. Compared to DSWE-Net and Spatio-Temporal CNNs, our approach shows superior reconstruction, segmentation, and noise insensitivity. Significance: This framework provides a robust approach to SWE reconstruction and inclusion segmentation, demonstrating strong potential for clinical translation.
- 16 December 2025
- Published by IOP Publishing in Physics in Medicine & Biology
- Vol. 71 (1)
Robust CNN multi-nested-LSTM framework with compound loss for patch-based multi-push ultrasound shear wave imaging and segmentation
Objective: Ultrasound shear wave imaging enables noninvasive, quantitative assessment of tissue pathology with mechanical elasticity measurements. However, shear wave elastography (SWE) reconstructions are challenged by noise sensitivity, inefficient multi-push strategies for scalable ROI coverage, and limited annotated data, leading to suboptimal reconstruction and unreliable inclusion segmentation. Approach: In this work, we present a novel two-stage deep learning framework that addresses these limitations through a CNN-based Multi-Nested-LSTM reconstruction network followed by a compound-loss-driven CNN-denoiser. The reconstruction stage begins with a ResNet3D-encoder that extracts spatiotemporal features from sequential multi-push acoustic radiation force (ARF) data. These features are temporally windowed with Nested CNN-LSTM, converted from 3D to 2D with temporal attention module (TAM), and enhanced by FFT-based frequency attention. The resulting 2D maps are subsequently decoded into primary 2D elasticity reconstructions. To mitigate data-scarcity and improve generalization, a patch-based training regime is also proposed. The second stage introduces a dual-decoder denoising network that separately processes inclusion and background stiffness features, followed by a fusion module that produces a denoised modulus map and a segmentation mask. A multi-objective compound loss is designed to accommodate the denoising, fusing, and mask generation. The method is validated on sequential multi-push (simulation and experimental) SWE motion data with multiple overlapping regions. Results: The method was tested on simulated and CIRS phantom datasets with four overlapping push regions, yielding 26.33 dB PSNR, 30.73 dB CNR, and 0.813 IoU in simulation, and 22.44 dB PSNR, 36.88 dB CNR, and 0.781 IoU experimentally. Evaluation on an ex vivo swine liver confirmed elasticity estimates within reported biological stiffness ranges. Compared to DSWE-Net and Spatio-Temporal CNNs, our approach shows superior reconstruction, segmentation, and noise insensitivity. Significance: This framework provides a robust approach to SWE reconstruction and inclusion segmentation, demonstrating strong potential for clinical translation.