Detection of Intracranial Aneurysms Using Multiphase CT Angiography with a Deep Learning Model

Area of Science:

• Medical Imaging
• Artificial Intelligence in Radiology
• Neurosurgery

Background:

• Intracranial aneurysms (IAs) pose significant risks, necessitating accurate detection.
• Computed tomography angiography (CTA) is a key imaging modality for IA diagnosis.
• Current detection methods can be limited by single-phase image analysis.

Purpose of the Study:

• To evaluate a multiphase fusion deep-learning model with automatic phase selection for IA detection.
• To compare the sensitivity and recall of this model against single-phase algorithms.
• To assess the model's performance in identifying aneurysm characteristics like position, shape, size, and rupture status.

Main Methods:

• Retrospective analysis of CTA images from 1110 patients (training), 139 (internal validation), and 134 (test).
• Independent validation using CTA images correlated with digital subtraction angiography (DSA).
• Development and comparison of a multiphase fusion deep-learning model with automatic phase selection against a single-phase algorithm.

Main Results:

• The multiphase fusion model showed higher sensitivity across internal, test, and independent validation datasets compared to single-phase analysis.
• Recall for aneurysm position, shape, size, and rupture status was greater or equal with multiphase selection.
• For aneurysm rupture detection in test data, multiphase selection achieved 94.8% recall (presence) and 87.6% (absence), outperforming single-phase (89.6% and 79.4%).

Conclusions:

• A multiphase fusion deep learning model with automatic phase selection enables automated detection of intracranial aneurysms.
• This AI model offers high sensitivity for IA detection from CTA images.
• The approach shows promise for improving diagnostic accuracy in neurovascular imaging.