Automated neuroradiological support systems for multiple cerebrovascular disease markers - A systematic review and

Jesse Phitidis1, Alison Q O'Neil2, William N Whiteley3

  • 1Centre for Clinical Brain Sciences, University of Edinburgh, 49 Little France Crescent, Edinburgh, EH164SB, United Kingdom; Canon Medical Research Europe, Bonnington Bond, 2 Anderson Place, Edinburgh, EH65NP, United Kingdom.

Insights

Automated systems aid radiologists in detecting cerebrovascular disease (CVD) markers like white matter hyperintensities (WMH) and stroke lesions on brain scans. However, no current system comprehensively analyzes all CVD markers for improved stroke and dementia risk assessment.

Area of Science:

  • Neurology and Radiology
  • Medical Imaging Analysis
  • Artificial Intelligence in Healthcare

Background:

  • Cerebrovascular diseases (CVD) are major causes of death and disability, leading to stroke and dementia.
  • Brain imaging reveals key CVD markers such as white matter hyperintensities (WMH), ischemic stroke lesions (ISL), lacunes, enlarged perivascular spaces (PVS), hemorrhagic lesions, cerebral microbleeds (CMB), and brain atrophy.
  • These markers are crucial for assessing future stroke and dementia risk and guiding patient management.

Purpose of the Study:

  • To systematically review automated systems that assist radiologists in identifying and analyzing CVD markers on brain imaging.
  • To evaluate both commercially available software and research publications for their capabilities in detecting at least two CVD markers.

Main Methods:

  • A systematic review was conducted on commercial products and research publications focused on automated CVD marker detection.
  • Included systems identified a minimum of two CVD markers from brain imaging data.
  • Data were synthesized to categorize system functionalities and the types of CVD markers analyzed.

Main Results:

  • The review included 29 commercial products and 13 research publications.
  • Commercial systems primarily focus on acute stroke lesion detection (CT) for triage or WMH and atrophy measurement (MRI) for neurodegeneration.
  • Research systems most frequently analyze WMH and ISL together using MRI; lacunes, PVS, and CMB are analyzed less frequently.
  • No openly validated systems currently offer comprehensive joint analysis of all major CVD markers.

Conclusions:

  • Existing automated systems partially address the need for CVD marker analysis, with distinct focuses in commercial and research domains.
  • Commercial systems are geared towards emergency stroke detection, while research systems explore broader applications including neurodegenerative diseases.
  • A significant gap exists in the market and research for a comprehensive, validated system capable of jointly analyzing all key CVD markers for improved diagnostic and prognostic accuracy.