Machine learning applied on chest x-ray can aid in the diagnosis of COVID-19: a first experience from Lombardy, Italy
Problem
Rapid, reliable differentiation of COVID-19 from non-COVID-19 patients using bedside chest x-rays was needed during the Lombardy outbreak.
Why it matters
Chest x-ray is widely available; AI support could help diagnose COVID-19 where PCR capacity is strained.
What the AI did
An ensemble of ten convolutional neural networks was trained on 250 COVID-19 and 250 non-COVID-19 chest x-rays and tested on an independent group of 110 patients against an RT-PCR reference standard.
Results
At 10-fold cross-validation: 0.78 sensitivity (95% CI 0.74–0.81), 0.82 specificity (abstract truncated)
Impact
Classifier aided COVID-19 diagnosis on an independent test cohort; clinical deployment not stated.
Record
| Venue | European Radiology Experimental |
|---|---|
| Published | 2021 |
| Work type | article |
| Authors | Isabella Castiglioni, Davide Ippolito, Matteo Interlenghi, Caterina Beatrice Monti, Christian Salvatore, Simone Schiaffino, Annalisa Polidori, Davide Giacomo Gandola et al. |
| Citations | 145 (OpenAlex, at import time) |
| References | 17 |
| Model or method | ensemble of ten convolutional neural networks (CNNs) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1186/s41747-020-00203-z |
| Open access | gold |
| Keywords | COVID-19, chest x-ray, deep learning, CNN ensemble, diagnosis, radiology |
Importance rating
major — millions of lives, an industry or a field
COVID-19 diagnosis during the Lombardy outbreak; abstract frames pandemic-scale need, strained PCR capacity context, multi-hospital cohorts and radiologist comparison.