JesusOmarDev1/NodeQuantAI
Sistema de segmentación volumétrica de ganglios linfáticos mediastínicos en tomografías computarizadas mediante una arquitectura 3D Attention U-Net, con análisis radiómico, clasificación automática y dashboard interactivo.
What's novel
Sistema de segmentación volumétrica de ganglios linfáticos mediastínicos en tomografías computarizadas mediante una arquitectura 3D Attention U-Net, con análisis radiómico, clasificación automática y dashboard interactivo.
Code Analysis
11 files read · 3 roundsA clinical decision support platform that extracts radiomic features from CT lymph node segmentations, predicts tumor volume via stacking regression, and classifies malignancy risk into ordinal levels using a Frank-Hall ordinal wrapper around a multi-model stacking classifier, served through a Strea
Strengths
Genuinely sophisticated ML pipeline with ordinal classification, calibrated probabilities, Optuna optimization, and domain-informed feature engineering; the dual-venv subprocess architecture is a pragmatic engineering solution to PyRadiomics compatibility constraints.
Weaknesses
No automated test suite, hardcoded Windows paths break portability, significant code duplication (ordinal classifier and feature engineering functions exist in multiple locations), and the README incorrectly describes a Django frontend when the actual app is Streamlit.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
48
Contributors
2
Files
38
Active weeks
3
Repository
Language
Python
Stars
2
Forks
0
License
MIT