I am a 3rd year Ph.D. candidate at VIU Lab, Johns Hopkins University under the supervision of Dr. Vishal M. Patel. My research focuses on computer vision and its application in the medical field as well as generative models, specific emphasis on video generation.
My research lies at the intersection of computer vision and medical image analysis, with a focus on developing deep learning techniques to make healthcare more affordable and accessible globally.
My specific research interests include 2D/3D segmentations, generative networks, representation learning, and addressing bias/ambiguity in medical image problems. I am also open to exploring general vision problems, particularly in the areas of generative networks and representation learning.
I have a strong track record of first-author publications in leading conferences, such as MICCAI and CVPR. Additionally, I am committed to mentoring undergraduate student research groups in machine learning at NSU.
News
April, 2024: Received Postdoc-NeT-AI fellow at DAAD, Germany.
April, 2024: One Paper Accepted at MIDL 2024.
July, 2023: Joined as Applied Scientist Intern at Amazon, Seattle, Bellevue.
February, 2023: One Paper accepted in CVPR 2023.
June, 2022: Received MICCAI Student Travel Award.
June, 2022: Two Papers accepted at MICCAI 2022.
September, 2021: Joined as a PhD student at VIU Lab, Johns Hopkins University.
December, 2020: One Paper is Accepted at Tissue and Cell.
September, 2020: One Paper is Accepted at IJMPC.
July, 2020: One Paper is Accepted at PRIME MICCAI 2020.
November, 2019: One Paper is Accepted at ICME 2019.
September, 2019: One Paper is Accepted at IEEE MoRSE 2019.
Selected Publications
See my Google Scholar profile for the complete and most recent publications.
The paper proposes a new approach for medical image segmentation that utilizes expert groups to generate multiple plausible outputs. The approach outperforms existing state-of-the-art networks in accuracy and diversity. The authors also introduce a new metric aligned with clinical practice
Orientation-Guided Graph Convolutional Network for Bone Surface Segmentation
Authors : Aimon Rahman , WGC Bandara, Jeya Maria Jose Valanarasu, Ilker Hacihaliloglu, Vishal M Patel
MICCAI, 2022
The article proposes a new method for bone surface segmentation using an orientation-guided graph convolutional network to improve connectivity in ultrasound images. The approach also adds supervision on the orientation of the bone surface to further impose connectivity. Validation on over 1000 in vivo US scans shows a 5.01% improvement over state-of-the-art methods in connectivity metric.
Simultaneous Bone and Shadow Segmentation Network Using Task Correspondence Consistency
Authors : Aimon Rahman , Jeya Maria Jose Valanarasu, Ilker Hacihaliloglu, Vishal M Patel
MICCAI, 2022
The article proposes a new approach for segmenting both bone surface and the corresponding acoustic shadow in ultrasound images using a single end-to-end network with a shared transformer-based encoder and task independent decoders. The approach leverages the complementary features between the two tasks and includes a cross-task feature transfer block to learn meaningful feature transfer between the two decoders.
3C-GAN: class-consistent CycleGAN for malaria domain adaptation model
Authors : Aimon Rahman , M Sohel Rahman, M.R.C. Mahdy
Biomedical Physics & Engineering Express, 2021
We introduce a modified loss function in unpaired adversarial translation model to prevent unwanted feature hallucination.
We pretrain a model on videos for human activity recognition which leads to better
representations for the downstream tuberculosis type classification task, especially for under-represented class samples.
Our method achieved 2nd place in the ImageCLEF 2021
Tuberculosis Type Classification Challenge.
This work proposes a blood cell segmentation dataset consisting of multiple cell types.
Additionally, all cell types
do not have equal instances, which encourages researchers to develop algorithms for
learning from imbalanced classes in a few shot learning paradigm. We also provide both learning and non-learning based
methods as baselines.
A Comparative Analysis of Deep Learning Architectures on High Variation Malaria Parasite Classification Dataset
Transformed an object detection dataset to classification dataset, conditional image synthesis is used to generate synthetic dataset and benchmarked several classification algorithms for the task of detecting malaria from microscopic images of red blood cells.
Authors : Hasib Zunair, Aimon Rahman, Nabeel Mohammed, and Joseph Paul Cohen
PRedictive Intelligence In MEdicine (PRIME), Medical Image Computing & Computer Assisted Intervention (MICCAI), 2020 Paper / Code / Slides
Showed that analyzing 3D medical images in a per slice basis is a sub-optimal approach, that can be improved by 3D context. Ranked 5-th in ImageCLEF 2019.
Benchmarked several classification algorithms for the task of detecting malaria from microscopic images of red blood cells. Transfer learning approach worked best in our study.
Trained a U-Net architecture with a pretrained ResNet backbone on knee MRIs at the slice level. The goal was to reconstruct high resolution images from the given undersampled image.
Authors: Hasib Zunair, Aimon Rahman , Nabeel Mohammed
CLEF Working Notes - Conference and Labs of the Evaluation Forum, 2019 Paper / Code
A 3D CNN with a slice selection method employed in the task of chest CT image analysis
for predicting tuberculosis (TB). Our method achieved 10-th place in the ImageCLEF 2019
Tuberculosis SVR - Severity scoring.
