Applied research in medical imaging

Alongside engineering work, I do applied research in medical imaging. One manuscript is currently under review; this section covers that work and what follows from it.

Manuscripts

Under review (Elsevier)

Model Accuracy

93.2%

Multi-Class Classification

Research Role

First Author

DHNN Study

Data

MRI

Large-Scale Preprocessing Pipeline

Under Review

Multi-Class Brain Tumor Classification and Grade Estimation using Dual-Head Neural Network

Elsevier Results in Engineering (Manuscript No. RINENG-D-26-00182)First Author2026

Proposed a novel Dual-Head Neural Network leveraging EfficientNetB0 for simultaneous multi-class brain tumor classification and grade prediction from MRI scans.

Key Findings

›Achieved 93.2% multi-class classification accuracy
›Simultaneous tumor type and grade prediction in one architecture
›Large-scale MRI preprocessing with cleaning, labeling, and augmentation
›Compact model (~4.0M parameters, ~15ms inference) for practical deployment

Methodology

›EfficientNetB0 backbone with transfer learning
›Dual-head architecture for multi-task learning
›MRI preprocessing pipeline using OpenCV and Pandas
›Rigorous training/validation for multi-class classification settings

Impact

Demonstrates a robust and efficient workflow for clinically relevant tumor typing and grading support in medical imaging contexts.

TensorFlowKerasPythonOpenCVMedical AI

View PDF

Research Interests

Where the research connects to the engineering — the questions I care about most sit at the intersection of applied AI and real, constrained systems:

›Time-series anomaly detection in industrial systems
›Retrieval grounding for domain-specific LLMs
›Multi-task architectures for efficient inference on constrained hardware

Interested in collaboration?

Open to research collaborations and joint work in medical imaging and applied AI for industrial systems.

Get in Touch→