Friday , November 27 2020

AI helps detect brain aneurysms on CT angiography



AI helps detect brain aneurysms on CT angiography

Examples of false-positive aneurysms, including (a) bone structures and vessel bifurcation, (b) veins, (c) vessel curvatures and (d) calcified plaques. The red box (d) indicates aneurysms noted by radiologists and the blue boxes indicate aneurysm candidates provided by the algorithm. Credit: Radiological Society of North America

A powerful type of artificial intelligence known as deep learning can help doctors detect potentially fatal brain aneurysms in CT angiography, according to a study published in the journal Radiology.

Brain aneurysms are weakened areas of the blood vessels in the brain. If left untreated, they can leak or rupture, sometimes with fatal results. The detection and characterization of these aneurysms are critical, as the risk of rupture depends on the size, shape and location of the aneurysm.

Angiotomography is usually the first choice for assessing brain aneurysms. The test is highly accurate, but brain aneurysms may go unnoticed in the initial evaluation due to their small size and the complexity of the blood vessels in the brain.

“In our daily work, we always come across cases where some important injuries have been missed by the human eye,” said senior study author Xi Long, Ph.D., from the Department of Radiology at Union Hospital at Tongji Medical College in Wuhan , China. “Brain aneurysms are among the small lesions that can go unnoticed in the routine evaluation of radiological images.”

Deep learning offers enormous potential as a complementary tool for a more accurate interpretation of brain aneurysms. A deep learning system is trained on existing images and learns to recognize abnormalities that can be difficult for a human observer to see. In radiology, deep learning has recently been used in a variety of functions to assist radiologists, such as in the detection of tuberculosis on chest radiographs.

AI helps detect brain aneurysms on CT angiography

A 54-year-old woman with a 2.9 mm maximum diameter aneurysm located in the left internal carotid artery (arrow). (a) Axial CT scan of the skull and (b) volume-rendered three-dimensional reconstruction image. The aneurysm was not detected by the algorithm, possibly because of its small size (<3 mm) and location near the base of the skull. Credit: Radiological Society of North America

In the new study, Dr. Long and colleagues developed a fully automated and highly sensitive algorithm for detecting brain aneurysms in CT angiography images. They used CT angiograms from more than 500 patients to train the deep learning system and then tested it on another 534 CT angiograms that included 649 aneurysms.

The algorithm detected 633 of the 649 brain aneurysms for a sensitivity of 97.5%. He also found eight new aneurysms that were overlooked in the initial assessment.

Statistical analysis revealed that assistance for deep learning improved the performance of radiologists. The improvement was more pronounced in less experienced radiologists.

“The developed deep learning system demonstrated excellent performance in detecting aneurysms,” said Dr. Long. “We found some aneurysms that were overlooked by human readers in the initial reports, but were successfully described by the deep learning system.”

The results suggest that the deep learning algorithm is promising as a support tool for the detection of brain aneurysms with the potential to be used clinically for a second opinion when interpreting cranial angiotomography images. It has a number of advantages in this environment, said Dr. Long, mainly due to the fact that the computer is not influenced by factors such as the level of experience, working time and mood that affect human performance.

AI helps detect brain aneurysms on CT angiography

93-year-old woman with an aneurysm of 2 mm in maximum diameter located in the left posterior cerebral artery (arrow). (a) Axial CT scan of the skull and (b) volume-rendered three-dimensional reconstruction image. The aneurysm was lost in the initial report, but it was successfully detected with the algorithm. Credit: Radiological Society of North America

The system has some limitations, noted Dr. Long. It may not detect very small aneurysms or aneurysms located near structures of similar density, such as bones. It also suffers from false positive results, which means that it incorrectly identifies aneurysm-like structures as aneurysms, which requires careful review of the system’s suggestions by human readers.

“Simply put, the deep learning system is designed to assist human readers, not to replace them,” said Dr. Long.

The system needs additional validation on more heterogeneous data, such as images of people in different parts of the world, which is essential to assess its generalization and applicability to daily clinical work.

“Right now, the role of this deep learning system, which has been trained to recognize aneurysms, is to provide suggestions to the human reader to improve their performance and reduce errors,” said Dr. Long. “The combined work of the human reader and the computer system improves diagnostic accuracy for the patient’s sake.”


Guidelines detail how to manage visceral artery aneurysms


More information:
Jiehua Yang et al. Deep learning to detect brain aneurysms with CT angiography, Radiology (2020). DOI: 10.1148 / radiol.2020192154 “Deep learning for the detection of cerebral aneurysms with CT angiography” Radiology, 2020.

Provided by the Radiological Society of North America

Quote: AI helps to detect brain aneurysms in CT angiography (2020, November 3) recovered on November 3, 2020 at https://medicalxpress.com/news/2020-11-ai-brain-aneurysms-ct-angiography. html

This document is subject to copyright. In addition to any fair dealing for the purpose of study or private research, no part may be reproduced without written permission. The content is provided for informational purposes only.




Source link