Welcome to CredenceX AI Research Lab

Building Intelligent Systems That Sense, Reason, and Assist

CredenceX AI Research Lab is an independent research initiative advancing trustworthy, explainable, and socially beneficial artificial intelligence. We develop data-driven solutions for medical imaging to support early warning, clinical decision support, and risk-aware deployment in high-stakes environments. We aim to reduce disparities by ensuring our models perform reliably across devices and diverse patient populations. We also study how to communicate uncertainty and explainability so outputs can be interpreted carefully in research and evaluation settings.

For research and evaluation only; not medical advice or a medical device unless explicitly stated.

Learn More

Core Research Areas at CredenceX

Exploring cutting-edge technologies to build safer, smarter, and more trustworthy AI systems for tomorrow

01

Vision + Language for Healthcare

Link images, reports, and clinical context to support multimodal understanding and decision support.

02

Explainable Imaging AI

Design explanations and evidence signals that align with clinical interpretation and workflow constraints.

03

Trustworthy & Risk-Aware AI

Advance reliability testing, uncertainty-aware outputs, calibration, and auditable decision support.

04

Efficient AI at the Edge

Build lightweight models for real-time use on web, mobile, and resource-limited devices.

05

Human-in-the-Loop Decision Support

Develop decision pipelines that preserve clinician control and communicate risk transparently.

06

Robust Across Sites & Scanners

Focus on cross-hospital generalization to reduce performance degradation in real-world settings.

Featured Projects at CredenceX

Showcasing our cutting-edge research projects that push the boundaries of AI innovation and real-world impact

Trustworthy & Calibrated AI

completed

DepTformer-XAI-SV

2025

Problem + modality: Depression emotion and severity classification from multimodal patient signals.
Core method: Reproducible explainable transformer pipeline with systematic ablations and a minimal Flask demo.
Evaluation emphasis: XAI faithfulness checks and reproducibility benchmarks (research use only).

PyTorchPythonFlaskDocker

Explainable Medical Image Intelligence

completed

Explainable Lung Cancer Diagnosis

2025

Problem + modality: Lung cancer diagnosis from CT scans and histopathology images.
Core method: Lightweight hybrid CNN–Transformer (MobileViT + CBAM attention + texture cues).
Evaluation emphasis: Grad-CAM explainability and robust cross-modality evaluation.

PyTorchMobileViTCBAMGrad-CAM

Multimodal Vision–Language Foundation Models

completed

Multimodal Information Fusion

2025

Problem + modality: Audio-visual object recognition from paired sound and image inputs.
Core method: Modular pipeline with hybrid, tensor, and FiLM-style fusion and flexible feature extraction.
Evaluation emphasis: Noise-robust training and fusion-strategy ablations.

PythonFiLM FusionXceptionxLSTM

Efficient Hybrid Transformers for Edge Deployment

completed

CottonVerse

2025

Problem + modality: Cotton leaf disease, fabric stain defects, and fabric composition from field and lab images.
Core method: Flask web application with efficient hybrid transformers and Grad-CAM overlays.
Evaluation emphasis: Probability calibration charts and visual explanation quality.

FlaskPyTorchtimmpytorch-grad-cam

Research Publications at CredenceX

Cutting-edge research contributions advancing the field of artificial intelligence

Journal

Published

Vision-audio multimodal object recognition using hybrid and tensor fusion techniques

Md Redwan Ahmed, Rezaul Haque, SM Arafat Rahman, Ahmed Wasif Reza, Nazmul Siddique, Hui Wang

Information Fusion (Elsevier)

Audio–visual multimodal object recognition with hybrid + tensor fusion strategies designed for robust real-world performance.

Journal

Published

Ensemble Transformer with Post-hoc Explanations for Depression Emotion and Severity Detection

Sazzadul Islam, Rezaul Haque, Mahbub Alam Khan, Arafath Bin Mohiuddin, Md Ismail Hossain Siddiqui, Zishad Hossain Limon, Katura Gania Khushbu, SM Masfequier Rahman Swapno, Md Redwan Ahmed, Abhishek Appaji

iScience (Cell Press / Elsevier)

DepTformer-XAI-SV: ensemble transformers for depression emotion/severity detection with LIME explanations and a web app.

Journal

Published

Explainable Transformer Framework for Fast Cotton Leaf Diagnostics and Fabric Defect Detection

SM Masfequier Rahman Swapno, Anamul Sakib, Amira Hossain, Jesika Debnath, Abdullah Al Noman, Abdullah Al Sakib, Md Redwan Ahmed, Rezaul Haque, Abhishek Appaji

iScience (Cell Press / Elsevier)

Explainable transformer framework spanning agriculture (cotton leaf) and textile inspection (fabric defect) with practical interpretability.

Journal

Published

LMVT: A hybrid vision transformer with attention mechanisms for efficient and explainable lung cancer diagnosis

Jesika Debnath, Amira Hossain, Anamul Sakib, Hamdadur Rahman, Rezaul Haque, Md Redwan Ahmed, Ahmed Wasif Reza, SM Masfequier Rahman Swapno, Abhishek Appaji

Informatics in Medicine Unlocked (Elsevier)

Hybrid ViT with attention and XAI for efficient and explainable lung cancer diagnosis (deployment-oriented).

Journal

Published

Accelerated and accurate cervical cancer diagnosis using a novel stacking ensemble method with explainable AI

Md Ismail Hossain Siddiqui, Shakil Khan, Zishad Hossain Limon, Hamdadur Rahman, Mahbub Alam Khan, Abdullah Al Sakib, SM Masfequier Rahman Swapno, Rezaul Haque, Ahmed Wasif Reza, Abhishek Appaji

Informatics in Medicine Unlocked (Elsevier)

Stacking ensemble + explainability for reliable cervical cancer diagnosis using Pap smear imaging.

Journal

Published

Hierarchical Swin Transformer Ensemble with Explainable AI for Robust and Decentralized Breast Cancer Diagnosis

Md Redwan Ahmed, Hamdadur Rahman, Zishad Hossain Limon, Md Ismail Hossain Siddiqui, Mahbub Alam Khan, Al Shahriar Uddin Khondakar Pranta, Rezaul Haque, SM Masfequier Rahman Swapno, Young-Im Cho, Mohamed S Abdallah

Bioengineering (MDPI)

Federation-ready Swin-Transformer ensemble with post-hoc explainability for robust breast cancer diagnosis.

Stay Updated with CredenceX

Subscribe to receive the latest research updates, breakthrough discoveries, and exclusive insights from our AI research lab.