Research on predicting radiographic exposure time in imaging based on neural network prediction models

Area of Science:

• Radiology
• Medical Imaging
• Vascular Interventions

Background:

• Cerebral angiography via the radial artery is a common procedure.
• Optimizing radiographic exposure time is crucial for patient safety and image quality.

Purpose of the Study:

• To identify anatomical and clinical predictors of radiographic exposure duration in radial artery cerebral angiography.
• To develop a predictive model for fluoroscopy time in this procedure.

Main Methods:

• Analysis of 210 patients undergoing radial artery cerebral angiography.
• Evaluation of anatomical (e.g., aortic arch diameter, innominate artery dimensions) and clinical (e.g., hypertension) factors.
• Development and validation of a neural network prediction model.

Main Results:

• Hypertension, double subclavian-innominate artery curves, and proximal left common carotid artery loops increased fluoroscopy time.
• Age, aortic arch diameter, innominate artery width, and its angle with the aortic arch were positively correlated with fluoroscopy time.
• Innominate artery length showed a negative correlation with fluoroscopy time. The neural network model achieved 80.0% accuracy on the training set and 76.9% on the test set (AUC=0.821).

Conclusions:

• Several anatomical variations and clinical factors significantly influence fluoroscopy time during radial artery cerebral angiography.
• A predictive model incorporating these factors demonstrates good accuracy in estimating exposure duration.