biomedical-imaging

🧬 Implement nuclei segmentation and classification using DINO self-supervision for histopathology images in a streamlined deep learning pipeline.

High-school research portfolio: 2 peer-reviewed publications (IEEE, IJHSR) + 3 active projects spanning quantum-inspired ML, computational biology, biomedical imaging, wildfire RL, and science-education research. Stanford Medicine, UCLA COSMOS & UCSB SRA affiliated.

Research portfolio of Zahed Tavangari: bioengineering, space biomedicine, nanomedicine, computational modelling

Interactive research prototype exploring quantum dot nanoparticles for multi-disease molecular imaging — currently covering mesial temporal sclerosis and oncology (solid tumor imaging), built on a shared, reusable three-phase engine.

🩺 医学人工智能科研指南 · Medical-AI-Guide

This repository provides DINOSim, a method that leverages the DINOv2 foundation model for zero-shot object detection and segmentation in bioimage analysis. DINOSim uses pretrained DINOv2 embeddings to compare patch similarities, allowing it to detect and segment unseen objects in complex datasets with minimal annotations.

Few-shot biomedical image segmentation from ~8 masks: frozen DINOv3 + a bank of classical native-resolution priors, fused in one light head. ISBI 2027 reference implementation.

[SIU'26] BIP-BAP: Biomedical Image Processing BAP (Scientific Research Project) - Piri Reis University

Morphology-first lung immunofluorescence quantification. QuPath reads and tiles whole slides; an unchanged Fiji/ImageJ engine does all measurement. Statistical unit = mice.

A 3D Slicer extension for interactive 3D visualization and quantitative colocalization analysis of multi-channel confocal z-stack images.

[IEEE TMI Best Paper Award] Official Implementation for UNet++

Biomedical computer vision pipeline for nuclei segmentation, instance detection and morphology, benchmarking custom U-Net, StarDist and Cellpose-SAM.

SNGP for uncertainty-aware biomedical image classification and OOD detection.

Open-source digital twin for fluorescence imaging systems, including image-dataset analysis, optical filters, illumination, sensors, electronics, calibration, and hardware optimization.

Open source Python library for building bioimage analysis pipelines

Microscopy nuclei segmentation on BBBC038 — OpenCV watershed baseline + PyTorch U-Net, metrics and quantification

Deep learning-based biomedical image classification using ResNet18 for automated peripheral blood cell recognition with Grad-CAM explainability.

Noise2Void self-supervised denoising applied to proprietary fluorescence microscopy data from INT Milan — training directly on single noisy images without clean targets.

Reproducible framework for evaluating VAE and autoencoder reconstruction strategies for X-ray classification under sample scarcity.

Analisi fasoriale fit-free di dati FLIM: trasformata phasor e unmixing lineare vs fitting NNLS; uv + Streamlit