MekaNet: A Tiling-Enhanced Semi-Supervised Detection Framework for Megakaryocytes
A submitted MekaNet manuscript focused on tiling and semi-supervised detection for megakaryocytes.
A BMC-style medical-image manuscript on a tiling-enhanced semi-supervised framework for detecting megakaryocytes in pathology imagery.
problem
Megakaryocyte detection in large pathology imagery is sensitive to tile construction, object scale, sparse labels, and semi-supervised data use.
key idea
Use a tiling-enhanced semi-supervised detection framework to better handle the scale and label constraints of megakaryocyte detection.
my role
Technical contributor for WSI/data pipeline, evaluation, and manuscript support.
methods
- • Tiling strategy
- • Semi-supervised detection
- • WSI/pathology preprocessing
- • Detection evaluation
evidence / results
- • Submitted manuscript tracked in the Overleaf project index
- • Complements the broader MekaNet WSI/tiny-object work
why this belongs in the portfolio
- • Documents a concrete medical-AI system, not just a generic model application
- • Adds domain-friction evidence to the applied-AI portfolio
authors
Jae-Hyun Baek et al.
venue / status
BMC / medical AI manuscript
Submitted/draft manuscript; details kept concise for public portfolio use.
tags