Mahdi Islam

Mahdi Islam

Lecturer & Research Supervisor, Independent University, Bangladesh | MAIA (Medical Imaging & Applications) Graduate

Posts by Collection

certifications

portfolio

posters

projects

publications

Real-time clinical gait analysis and foot anomalies detection using pressure sensors and convolutional neural network

Published in 2022 7th International Conference on Business and Industrial Research (ICBIR), 2022

This research presents a novel insight on gait disorder detection using transfer learning algorithms on sensor-acquired data.

Recommended citation: Mahdi Islam, Musarrat Tabassum, Mirza Muntasir Nishat, Fahim Faisal, Muhammad Sayem Hasan. "Real-time clinical gait analysis and foot anomalies detection using pressure sensors and convolutional neural network", 2022 7th International Conference on Business and Industrial Research (ICBIR), Pages 717-722, IEEE.

Semi-Supervised Transformer-Based Cervical Segmentation: FUGC 2025 Challenge

Published in IEEE 22nd International Symposium on Biomedical Imaging (ISBI), 2025

Semi-supervised transformer-based approach for cervical segmentation submitted to the Fetal Ultrasound Grand Challenge (FUGC) 2025, part of the team that ranked 5th internationally.

Recommended citation: Mahdi Islam, Musarrat Tabassum, Marta Elbatel, Agnes Mayr, Christian Kremser, Markus Haltmeier, Enrique Almar-Muñoz. "Semi-Supervised Transformer-Based Cervical Segmentation: FUGC 2025 Challenge." IEEE 22nd International Symposium on Biomedical Imaging (ISBI), 2025.

Uncertainty-Guided Active Learning for Access Route Segmentation and Planning in Transcatheter Aortic Valve Implantation

Published in Journal of Imaging, MDPI, 2025

This study introduces an annotation-efficient active learning pipeline for 3D segmentation and automated diameter quantification in CMR-based TAVI planning.

Recommended citation: Mahdi Islam, Musarrat Tabassum, Agnes Mayr, Christian Kremser, Markus Haltmeier, Enrique Almar-Munoz. "Uncertainty-Guided Active Learning for Access Route Segmentation and Planning in Transcatheter Aortic Valve Implantation." *Journal of Imaging*, vol. 11, no. 9, 318, 2025. https://doi.org/10.3390/jimaging11090318.

Automated Aortoiliofemoral Segmentation and Transfemoral Access Planning Using Non-Contrast MR Angiography for Transcatheter Aortic Valve Replacement Guidance

Published in Radiology: Cardiothoracic Imaging, 2026

Automated pipeline for aortoiliofemoral segmentation and quantitative transfemoral access planning from non-contrast MR angiography, developed to support TAVR procedure guidance.

Recommended citation: Enrique Almar-Muñoz, Mahdi Islam, Musarrat Tabassum, Markus Haltmeier, Martin Pamminger, Leo Neumann, Christian Kremser, Martin Reindl, Sebastian J. Reinstadler, Bernhard Metzler, Agnes Mayr. "Automated Aortoiliofemoral Segmentation and Transfemoral Access Planning Using Non-Contrast MR Angiography for Transcatheter Aortic Valve Replacement Guidance." Radiology: Cardiothoracic Imaging, 2026 (under review).

FUGC: Benchmarking Semi-Supervised Learning Methods for Cervical Segmentation

Published in IEEE Transactions on Medical Imaging, 2026

Benchmark study evaluating semi-supervised learning methods for cervical segmentation in fetal ultrasound, based on the Fetal Ultrasound Grand Challenge (FUGC).

Recommended citation: Jun Bai, Yiqun Tang, Zongwei Zhou, Mahdi Islam, Musarrat Tabassum, Enrique Almar-Muñoz, et al. "FUGC: Benchmarking Semi-Supervised Learning Methods for Cervical Segmentation." IEEE Transactions on Medical Imaging, 2026.

Uncertainty-Guided Active Learning for Access Route Segmentation and Planning in Transcatheter Aortic Valve Implantation

Published in ISMRM Annual Meeting 2026, Cape Town — Digital Poster, Congenital Heart Disease, Valves, and Vessels, 2026

Digital poster abstract (#05614) presented at the ISMRM Annual Meeting 2026 in Cape Town, on an annotation-efficient CMR-based pipeline for aorto-iliac segmentation and TAVI access route quantification.

Recommended citation: Enrique Almar-Muñoz, Mahdi Islam, Musarrat Tabassum, Christian Kremser, Markus Haltmeier, Agnes Mayr. "Uncertainty-Guided Active Learning for Access Route Segmentation and Planning in Transcatheter Aortic Valve Implantation." ISMRM Annual Meeting 2026, Cape Town, Abstract #05614.

Brain Metastases Segmentation for BraTS 2026 Task 1: A Multi-Architecture Comparison

Published in BraTS-METS, MICCAI 2026, 2026

Comparison of a plain nnU-Net baseline, a Residual Encoder Large (ResEncL) variant, region-based training, and a Primus (PrimusV3S) transformer model for brain metastasis segmentation, submitted to the BraTS-METS 2026 Task 1 challenge.

Recommended citation: Mahdi Islam, Musarrat Tabassum. "Brain Metastases Segmentation for BraTS 2026 Task 1: A Multi-Architecture Comparison." BraTS-METS, MICCAI 2026.

talks

teaching