Lightning Oral Presentation Australian and New Zealand Stroke Organisation Conference 2026

Artificial Intelligence assisted imaging interpretation in acute stroke: a survey of clinician perspectives in Aotearoa New Zealand (138942)

Tom Hornblow 1 , Janice Kang 1 , Alan Barber 2 , Tereki Stewart 3 , Dominique Cadilhac 4 , Lillian Choy 5 , Martin Punter 1 6 , Alan Davis 7 , John Fink 8 , Stefan Brew 9 , Anna Ranta 1 6
  1. Department of Medicine, University of Otago, Wellington, New Zealand
  2. Faculty of Medical and Health Sciences, University of Auckland, Auckland, New Zealand
  3. Te Hononga o Tāmaki me Hoturoa, Auckland, New Zealand
  4. Translational Public Health and Evaluation Research Division, Monash University, Melbourne, Victoria, Australia
  5. Department of Neurology, Waikato Hospital, Hamilton, New Zealand
  6. Department of Medicine, Wellington Hospital, Wellington, New Zealand
  7. Department of Medicine, Whangārei Hospital, Whangārei, New Zealand
  8. Department of Neurology, Christchurch Hospital, Christchurch, New Zealand
  9. Department of Neuroradiology, Auckland City Hospital, Auckland, New Zealand

Aims

AI may assist in the interpretation of computed tomography (CT) scans in the setting of acute stroke care. We aimed to assess the clinician end-user opinions of such technology as part of a larger research programme which aims to assess the overall feasibility of the large-scale implementation of AI-assisted imaging technology throughout Aotearoa New Zealand (NZ), and its impact on workflow and patient care.

Methods

A cross-sectional survey was conducted using a 14-item questionnaire between November 13, 2025, and January 08, 2026. The survey was distributed to lead stroke clinicians (n=69) at NZ stroke capable hospitals with a prompt to further forward to general medicine, geriatrics, radiology, and emergency medicine teams. Survey items included participants’ previous exposure to advanced stroke imaging tools, features most important to clinicians, and the potential impacts of a nationwide AI-assisted tool on reporting times and target demographics. Questions included both a series of bipolar five-point Likert-type questions and free text responses.

Results

Fifty-one responses were included for analysis. Responses confirmed overall positive clinician perspectives on AI-assisted stroke imaging tools. 38 (88.4%) supported a single nationwide tool, and 32 (74.4%) supported a stroke-specific tool. 41 (97.6%) predicted outcome benefit for rural populations, and 40 (95.2%) reduced treatment delays out-of-hours. A clear hierarchy of important features of advanced stroke imaging tools was established, with CT perfusion core/perfusion lesion maps with volumetrics ranked the highest.

Conclusions

Clinicians feel AI assistance positively contributes to addressing inequitable stroke outcomes with demand for a single nationwide advanced stroke-specific imaging tool.