Developing the Lung Graph-Based Machine Learning Model for Identification of Fibrotic Interstitial Lung Diseases
Haishuang Sun1,2, Min Liu3,4, Anqi Liu5,6
1National Center for Respiratory Medicine, State Key Laboratory of Respiratory Health and Multimorbidity; National Clinical Research Center for Respiratory Diseases；Institute of Respiratory Medicine, Chinese Academy of Medical Sciences; Department of Pulmonary and Critical Care Medicine, China-Japan Friendship Hospital, Beijing, 100029, China.
Journal of Imaging Informatics in Medicine
|February 12, 2024
Summary
A new lung graph machine learning model accurately identifies fibrotic interstitial lung disease (f-ILD) from HRCT scans. This AI tool demonstrated superior diagnostic performance compared to radiologists, aiding early f-ILD detection and intervention.
Area of Science:
- Radiology
- Artificial Intelligence
- Medical Imaging
Background:
- Accurate detection of fibrotic interstitial lung disease (f-ILD) is crucial for timely clinical intervention.
- High-resolution computed tomography (HRCT) is a key imaging modality for ILD assessment.
- Current diagnostic methods can be subjective and require expert interpretation.
Purpose of the Study:
- To develop and evaluate a novel lung graph-based machine learning model for identifying f-ILD.
- To compare the diagnostic performance of the machine learning model against experienced radiologists.
Main Methods:
- A lung graph-based machine learning model was developed using HRCT data from 279 patients (156 f-ILD, 123 non-f-ILD).
- Local radiomics features were extracted from a lung geometric atlas to construct lung graph models.
- A Weighted Ensemble model was trained to characterize global radiomics feature distribution for f-ILD diagnosis.
Main Results:
- The machine learning model achieved high diagnostic accuracy at the patient level (0.986).
- The model's classification accuracy significantly surpassed that of three radiologists.
- The area under the curve (AUC) values for the model were statistically higher than those of the radiologists (p < 0.05).
Conclusions:
- The developed lung graph-based machine learning model effectively identifies f-ILD from HRCT.
- The model's diagnostic performance exceeds that of human radiologists, offering objective assessment capabilities.
- This AI tool has the potential to significantly aid clinicians in the objective assessment and diagnosis of f-ILD.
