Tunneling Nanotubes (TNTs) Detection using Deep Learning

Duration: July 2023 – January 2024 Mentor: Prof. Sushmita Jha, IIT Jodhpur Collaboration: AIIMS Jodhpur

🧩 Overview

This project focuses on detecting Tunneling Nanotubes (TNTs) between tumor cells to analyze the stage of brain tumor progression using deep learning and image-based analysis. Our aim was to create a diagnostic system that could predict tumor stage directly from cellular images—eliminating the need for chemotherapy-based stage confirmation, which can take up to a month and delay treatment decisions.

⚙️ Core Contributions

Developed a deep learning-based TNT detection system capable of identifying and counting TNTs between tumor cells.

Built a pixel-level TNT labeling and preprocessing model, improving detection precision and reducing false positives.

Utilized Python, C++, and FIJI (ImageJ) for image analysis, data preprocessing, and visualization.

Collaborated with AIIMS Jodhpur to collect and annotate real cellular image datasets for model training.

Conducted comparative evaluation of various CNN architectures for high-accuracy TNT detection.

🧪 Technologies & Tools

Python • C++ • TensorFlow / PyTorch • FIJI (ImageJ) • OpenCV • NumPy • Matplotlib

📊 Outcomes

Achieved A* grade for project excellence at IIT Jodhpur.

Demonstrated potential to reduce diagnosis time for tumor stage classification by >80%.

Proposed a novel pixel-wise TNT labeling framework for biomedical image segmentation tasks.