Using artificial neural network analysis of global ventilation-perfusion scan morphometry as a diagnostic tool

Area of Science:

• Radiology
• Medical Imaging Analysis
• Artificial Intelligence in Medicine

Background:

• Pulmonary embolism (PE) diagnosis relies on imaging and clinical assessment.
• Radionuclide ventilation-perfusion (V/Q) scans provide functional lung information.
• Automated analysis of V/Q scans may improve diagnostic accuracy.

Purpose of the Study:

• To evaluate the predictive capability of global statistical data from V/Q scans for PE.
• To compare artificial neural network (ANN) performance with clinician performance in diagnosing PE.

Main Methods:

• 161 patients undergoing V/Q scans and pulmonary angiography were studied.
• Morphometric data from V/Q scans were used as input for ANNs.
• ANNs were trained to predict the likelihood of PE.

Main Results:

• ANNs using automated V/Q scan data outperformed clinicians in predicting PE in patients with normal chest radiographs (p < .005 and p < .01).
• No significant difference in performance was observed between ANNs and clinicians for patients with abnormal chest radiographs.

Conclusions:

• Artificial neural networks utilizing standard, user-independent V/Q scan statistics can significantly improve PE diagnostic accuracy.
• This AI-assisted approach is particularly beneficial for patients with normal chest radiographs.