Deep Learning-Based Decision-Tree Classifier for COVID-19 Diagnosis From Chest X-ray Imaging
Problem
COVID-19 testing demand overwhelmed capacity; RT-PCR is definitive but chest X-ray is fast and affordable, so AI detection of COVID-19 from CXR images was investigated.
Why it matters
Fast, cheap COVID-19 screening could ease pandemic testing bottlenecks.
What the AI did
Three binary decision trees, each trained by a PyTorch-based convolutional neural network, classify CXR images as normal/abnormal, tuberculosis, or COVID-19.
Results
98% accuracy for normal/abnormal tree, 80% for tuberculosis tree; COVID-19 classifier accuracy truncated in abstract, not stated in source.
Impact
Feasibility demonstrated, but no external validation stated and RT-PCR remains the definitive test.
Record
| Venue | Frontiers in Medicine |
|---|---|
| Published | 2020 |
| Work type | article |
| Authors | Seung Hoon Yoo, Hui Geng, Tin Lok Chiu, Siu Ki Yu, Dae Chul Cho, Jin Uk Heo, Minsung Choi, Il Hyun Choi et al. |
| Citations | 287 (OpenAlex, at import time) |
| References | 38 |
| Model or method | CNN-based decision-tree classifier (PyTorch) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.3389/fmed.2020.00427 |
| Open access | gold |
| Keywords | COVID-19, chest X-ray, deep learning, decision tree, CNN, diagnosis |
Importance rating
major — millions of lives, an industry or a field
Abstract frames COVID-19 as a global pandemic with increased demand for testing, diagnosis, and treatment; fast CXR pre-screening addresses pandemic-scale testing bottlenecks.