OH
Obaidul Haque
Scientific Research & Innovations

Research Publications

Specializing in novel deep learning architectures for Medical Image Analysis, Computer Vision, State-Space Models (Mamba), Vision Transformers, and Biomedical Signal Processing (EEG/ECG).

Total Papers

40+

ACM | IEEE | Elsevier| Springer

1st Author

15+

Lead Architecture Design

Published & Presented

25

11 Published · 14 Presented

Under Review

17

Top Tier Journals & Conf

Filter Domain:

Top 5 Featured Breakthroughs

Showing 5 of 5 highlighted
Accepted1st Author
2026

OBAIS-Mamba: Kernel-Invariant Anatomical State Learning for Dose-Robust Low-Dose CT Reconstruction

ACM International Conference on Computing Advancements (ICCA), 2026

Authors: Obaidul Haque, et al.

Proposes a novel state-space model (SSM) architecture, OBAIS-Mamba, engineered to overcome quantum noise and severe streak artifacts in Low-Dose Computed Tomography (LDCT). By formulating a kernel-invariant continuous state-space formulation, the model learns anatomical continuity across slices while preserving delicate lesion edges without over-smoothing.

Core Novelty & Contributions
  • •Engineered kernel-invariant state transitions achieving higher Peak Signal-to-Noise Ratio (PSNR) under 10% radiation dosage.
  • •Formulated frequency-guided structural preservation loss reducing structural blurring in fine pulmonary nodules.
  • •Demonstrated 2.4x inference speedup over standard 3D diffusion reconstruction baselines on clinical thoracic scans.
#Mamba#State-Space Models (SSM)#Low-Dose CT#Medical Imaging#PyTorch
Published3rd Author
2025

Stroke Prediction Using Ensemble Machine and Deep Learning Models

Biomedical Materials & Devices (BMD), Springer Nature, 2025

Authors: Md. Samiul Alom, Rajib Kumar Halder, Obaidul Haque, Saikat Kumar Gain, Md. Alamin Talukder, Fahim Shakil Tamim, Nusrat Jahan, Tamanna Tabassum, Fatema Akter, Faisal Imran, Rakib Hossen

Comprehensive epidemiological and clinical study benchmarking stacking ensemble classifiers, gradient boosting algorithms, and deep neural networks to predict cerebrovascular stroke risk factors across heterogeneous demographic cohorts.

Core Novelty & Contributions
  • •Benchmarked ensemble stacking models against imbalanced clinical patient datasets.
  • •Identified primary multi-factorial hypertension, glucose, and cardiovascular risk indicators.
  • •Published in Springer Nature Biomedical Materials and Devices with peer-reviewed open access.
#Stroke Prediction#Clinical ML#Ensemble Stacking#Springer Nature Q1#Published DOI
View Paper
Accepted1st Author
2026

An Interpretable Dual-Lead CNN-Transformer Hybrid Framework for ECG Arrhythmia Classification with Gradient-Based Explainability

IEEE SPICSCON, 2026

Authors: Obaidul Haque, Ferdaus Anam Jibon, Abu Bakar Rakib, Md. Abdur Rahim, Rafat Ara, Mariha Tabassum

Presents a hybrid architecture fusing convolutional local receptive fields with multi-head self-attention transformers for dual-lead electrocardiogram (ECG) arrhythmia categorization. Incorporates gradient-weighted class activation mapping (Grad-CAM) to visualize P-Q-R-S-T morphological anomalies.

Core Novelty & Contributions
  • •Dual-lead cross-channel attention mechanism isolating subtle cardiac conduction delays.
  • •Gradient-based saliency mapping providing cardiologists with explainable, beat-by-beat attribution heatmaps.
  • •Rigorous evaluation on MIT-BIH and standard clinical arrhythmia databases demonstrating high specificity.
#ECG Signal Processing#Vision Transformers#Explainable AI (XAI)#Grad-CAM#IEEE
Minor Revision1st Author
2026

Automated Classification of Alzheimer’s Disease Stages from Brain MRI

Discover Neuroscience, Springer Nature, 2026

Authors: Obaidul Haque, Ferdaus Anam Jibon, Utpal Kanti Das, A. H. M. Kamal, Gahangir Hossain

Investigates deep neural networks with spatial-temporal attention mechanisms for four-stage classification of Alzheimer's Disease progression (Non-Demented, Very Mild, Mild, and Moderate Cognitive Impairment) from structural magnetic resonance imaging (sMRI) volumes.

Core Novelty & Contributions
  • •Designed an attention-guided feature extractor highlighting hippocampal and ventricular morphological volume shifts.
  • •Achieved robust multi-class accuracy across public multi-center neuroimaging datasets (ADNI / OASIS).
  • •Provided patient-level cross-validation preventing data leakage between anatomical scan slices.
#Brain MRI#Alzheimer's Stage Detection#Neuroimaging#Springer Nature Q2#PyTorch
Presented1st Author
2026

EfficientMSwin: A YOLO-Guided Multi-Scale Swin Transformer for Fine-Grained Malaria Cell Stage Classification

International Conference on Electronics, Communications and Information Technology (ICONNECT), 2026

Authors: Obaidul Haque, Mariha Tabassum, Ferdaus Anam Jibon, Tania Akter Setu, Shah Mostofa Tariquzzaman

A two-stage computer vision diagnostic pipeline integrating a streamlined YOLO detector for red blood cell localization with a hierarchical Multi-Scale Swin Transformer (MSwin) for fine-grained morphological stage classification of Plasmodium falciparum in microscopic blood smears.

Core Novelty & Contributions
  • •Two-stage cascade decoupling cell localization from fine-grained parasite stage identification.
  • •Shifted-window attention mechanism capturing subtle intra-cellular chromatin and ring-stage variations.
  • •High diagnostic sensitivity verified against stained thin blood smear microscopic imagery.
#Swin Transformer#YOLO#Malaria Diagnostics#Microscopy Analysis#Computer Vision

Complete Research Ledger (40+ Papers)

The remaining publications include 11 published, 14 presented, and 17 under review papers across ICCIT, ICECE (BUET), ICCA (AIUB), ICONNECT, and QPAIN (CUET). Complete BibTeX database available upon request.