Machine Learning · Deep Learning · Biomedical Imaging

Machine learning for biomedical imaging, with MR Elastography as a primary focus.

I’m a Ph.D. candidate at The Ohio State University building deep learning, segmentation, diffusion/score-based generative modeling, and optimization pipelines for imaging inverse problems and clinical decision support. Current work spans MRE reconstruction, automated scar segmentation in Myocardial Delayed Enhancement, and Diffusion-driven augmentation for limited-data clinical imaging.
• Internship Work: Myocardial scar segmentation using deep learning, cardiac MRI flow-view classification using self-attention and transfer learning, and diffusion models for data augmentation in cardiac MRI segmentation.
• PhD Work: Design of deep learning-based inversion methodologies for MR Elastography by formulating the inverse problem as a data-driven reconstruction framework.

View Research Download CV
Portrait of Hassan Iftikhar

Hassan Iftikhar

Software Engineer Intern @ Neosoft

Ph.D. Candidate · Biomedical Engineering

The Ohio State University, Columbus

Profile

Research identity: ML/DL methods first, with MR Elastography as a primary translational application domain.

My work integrates modern ML methods with imaging physics to improve robustness, data-efficiency, and clinical feasibility in biomedical imaging.

I design inversion and reconstruction frameworks such as DIME and 3D-DIME to recover high-quality stiffness maps from noisy and highly aliased wavefields. In parallel, I apply segmentation and diffusion-based data strategies to other clinical imaging tasks, so the methodology remains broadly transferable beyond elastography. I am co-advised by Prof. Rizwan Ahmad and Prof. Arunark Kolipaka.

MRE Inverse Problems FEM Simulation Deep Learning Inversion Score-Based Gaussian Models Biomedical Image Segmentation SMS-MRE Reconstruction

Featured Research Projects

DIME project figure

DIME

Deep learning driven inversion framework for shear modulus estimation in MRE using U-Net style models trained on FEM-generated phantoms.

  • Higher correlation with ground-truth stiffness
  • Lower bias than traditional inversion methods
  • Sharper recovery of focal stiffness changes

Published in MAGMA

Paper Code
3D-DIME project figure

3D-DIME

3D extension of DIME that improves cross-slice anatomical consistency using volumetric FEM phantoms for liver and brain applications.

  • Technique demonstrated in ISMRM 2026 abstract
  • Breast tumor 3D-MRE application in progress
  • Journal manuscript in preparation
Abstract
SMS-MRE project figure

SMS-Accelerated MRE Reconstruction

Deep learning reconstruction for SMS-EPI MRE to suppress aliasing and noise while reducing effective scan constraints.

  • Liver MRE: shorter effective breath-hold burden
  • Ongoing lung MRE work at 0.55T and higher multiband factors
  • Submitted to ISMRM 2026

Applied ML Across Biomedical Imaging

Diffusion / Score-Based Gaussian Modeling (Neosoft)

Built score-based diffusion workflows to generate and augment clinically realistic image variations, helping mitigate dataset scarcity and improve model robustness in limited-data medical imaging settings.

Myocardial Scar Segmentation (Neosoft)

Developed automated myocardial scar segmentation on LGE-CMR using hybrid CNN+Transformer models, with diffusion-driven augmentation to address class imbalance and sample scarcity in production-oriented pipelines.

Publications & Talks

Selected Publications

  1. DIME: Deep Learning Driven Inversion Framework for Shear Modulus Estimation in Magnetic Resonance Elastography (2025), MAGMA.
  2. DIME ISMRM 2026 abstract (submitted), Cape Town, South Africa.
  3. 3D-DIME ISMRM 2026 abstract (submitted), Cape Town, South Africa.
  4. SMS-accelerated MRE reconstruction ISMRM 2026 abstract (submitted).
  5. AI stiffness estimation dataset curation, ISMRM 2025, Honolulu.
View prior ISMRM publication record

Talks & Presentations

  • Hayes Advanced Research Forum, 2026 (Columbus, OH)
  • ISMRM 2026: Image Reconstruction and AI Session
  • ISMRM 2025: Acquisition and Reconstruction Session
  • ISMRM 2024: Image Reconstruction and AI Session

Experience

Software Engineer Intern · Neosoft

Milwaukee, WI · May 2026 – Aug 2026

Built a hybrid CNN+Transformer pipeline for automated myocardial scar segmentation from LGE-CMR and integrated it toward production in suiteHeart post-processing workflows. Used diffusion/score-based generative augmentation to mitigate data scarcity and improve segmentation robustness.

Graduate Research Associate · OSU Wexner Medical Center

Columbus, OH · Aug 2022 – Present

Led AI-based MRE stiffness reconstruction and fast cardiac reconstruction research, including domain adaptation, score-based Gaussian priors, and plug-and-play optimization pipelines.

Machine Learning & Software Development Intern · BLOMSO

Columbus, OH · May 2025 – Aug 2025

Developed end-to-end ML systems for soil nutrient prediction with custom hardware and limited-data model optimization.

Skills

PythonC/C++MATLABPyTorch TensorFlowCUDASignal Processing Image ProcessingTransformersU-Net Diffusion/Score ModelsBiomedical Segmentation Data AugmentationClinical AI Pipelines COMSOLEmbedded Systems

Education

Ph.D., Computational Imaging and Machine Learning
The Ohio State University, 2022–2027 · GPA: 3.8/4.0

B.S., Electrical and Computer Engineering
UET Lahore, 2018–2022 · GPA: 3.6/4.0

Leadership & Service

  • Peer Mentor, BMEGSA Peer2Peer Program (2023–2024)
  • Undergraduate Research Mentor (2022–2024)
  • PhD Admissions Guidance Speaker, UET Lahore (2023–2025)
  • Volunteer, International Friendship Inc. (2022–2025)

Honors & Awards

  • Distinction in Signals Processing, OSU (2022)
  • National Talent Scholarship (2018–2022)

Let’s Connect

I’m open to research collaborations in biomedical imaging, MRE, and machine learning for healthcare.

Gmail Outlook LinkedIn GitHub