Skip to content
DraconiticResearch
Search Sources API
203,234 studies31 sourcesUpdated 2 min ago
Peer-reviewedCC BYJournal of clinical medicine

Diagnostic Accuracy of GPT-5 in Detecting and Localizing Acute Ischemic Stroke on Isolated Diffusion-Weighted Imaging: A Retrospective Case–Control Study

  • Emre Gökçen1
  • Mikail Kuşdoğan1
  • Mustafa Fatih Erkoc2
  • Levent Albayrak1
  • Güneş Seda Albayrak3
  • Deniz Tanrıverdi4
4 affiliations
  1. 1Department of Emergency Medicine, Faculty of Medicine, Yozgat Bozok University, 66100 Yozgat, Turkey
  2. 2Department of Radiology, Faculty of Medicine, Yozgat Bozok University, 66100 Yozgat, Turkey
  3. 3Department of Neurology, Faculty of Medicine, Yozgat Bozok University, 66100 Yozgat, Turkey
  4. 4Department of Emergency Medicine, Dr. Nevruz Erez State Hospital, 76000 Iğdır, Turkey
Read full textPDFDOICite

Abstract

Background/Objectives: Multimodal large language models are emerging tools for medical image interpretation, but their performance on isolated diffusion-weighted imaging (DWI) for acute ischemic stroke remains largely unexplored. We evaluated GPT-5 for stroke detection and vascular territory classification on preselected representative DWI images using a structured two-step prompting protocol. Methods: This single-center, retrospective, balanced case–control study included 220 cases (110 stroke patients and 110 stroke-negative controls). Four representative axial DWI slices per case were provided without apparent diffusion coefficient maps. Step 1 queried stroke presence and vascular territory; Step 2 queried the middle cerebral artery (MCA) segment-associated territory (M1–M4) following an MCA response. The reference standard was radiologist–neurologist consensus based on full multimodal magnetic resonance imaging, blinded to GPT-5 outputs. Diagnostic performance metrics and Cohen’s kappa (κ) were calculated. Results: Stroke detection showed a sensitivity of 100% (95% confidence interval [CI], 96.6–100%), specificity of 83.6% (95% CI, 75.6–89.4%), accuracy of 91.8%, and κ = 0.836. Accuracy was 93.6% for anterior/posterior circulation (κ = 0.838) and 90.9% for vascular territory (κ = 0.799). Conditional MCA segment-associated territory accuracy was 71.6% (53/74; κ = 0.602); end-to-end accuracy was 66.3% (53/80). Exploratory analysis of the 18 false-positive cases identified four overlapping error categories: artifactual and physiologic signal misinterpretation (72.2%), chronic lesion misclassification (27.8%), terminological inconsistency or unsupported report statements (77.8%), and etiological scenarios or unhedged definitive diagnoses (38.9%). Conclusions: GPT-5 achieved high sensitivity and a specificity of 83.6% for stroke classification on preselected representative DWI images using a vascular territory-based diagnostic framework. However, the 16.4% false-positive rate indicates that positive findings require radiologist confirmation. These findings support further evaluation of GPT-5 as a potential clinician-assisted image-interpretation tool but do not establish a clinical decision-support role.

Related

  1. Diagnostic Performance of Color Doppler Ultrasound for Detecting Nuchal Cord and Associated Immediate Neonatal Outcomes at Term in Uganda: A Prospective Diagnostic‐Accuracy Cohort Study
    Fiona Asiimwe, Theoneste Hakizimana, Marie Pascaline Sabine Ishimwe +10
    Health science reportsPeer-reviewedCC BYFull textGeneral medicineEurope PMC
  2. Diagnostic accuracy of urine test strips for CSF analysis: high sensitivity in detecting blood, low sensitivity but high specificity in detecting inflammation
    Lina Marie Albers, Franz Felix Konen, Erda Atakan +6
    Frontiers in NeurologyPeer-reviewedCC BY 4.0Full textPhysiologyCrossrefDOAJ
  3. Diagnostic accuracy of glycated albumin and fructosamine for detecting diabetes and prediabetes in adults: a systematic review and meta-analysis protocol.
    Kavindya Fernando, Dulani Kottahachchi, Lochana Pasan +4
    Systematic reviewsPeer-reviewedCC BY-NC-NDFull text1 citedEndocrinology & metabolismEurope PMCDOAJ
  4. Trimester-Specific Diagnostic Accuracy of Amniotic Fluid α-Fetoprotein for Detecting Preterm Premature Rupture of Membranes Versus AmniSure
    Homeira Vafaei, Vida Amiri, Naeimehossadat Asmarian +6
    Galen Medical JournalPeer-reviewedCC BY 4.0Full textEpidemiology & public healthPMC
  5. Diagnostic accuracy of ultrasound in detecting pathological complete response to neoadjuvant chemotherapy in breast cancer: a systematic review and diagnostic test accuracy meta-analysis.
    Liping Fan, Wentao Qu, Lintao Liu +3
    Gland surgeryPeer-reviewedCC BY-NC-NDFull textMedical imaging & radiologyEurope PMC

Cite

BibTeXRIS
BibTeX
@article{gokcen2026diagnostic,
  title = {{Diagnostic Accuracy of GPT-5 in Detecting and Localizing Acute Ischemic Stroke on Isolated Diffusion-Weighted Imaging: A Retrospective Case–Control Study}},
  author = {Emre Gökçen and Mikail Kuşdoğan and Mustafa Fatih Erkoc and Levent Albayrak and Güneş Seda Albayrak and Deniz Tanrıverdi},
  journal = {Journal of clinical medicine},
  year = {2026},
  month = oct,
  doi = {10.3390/jcm15197716},
  url = {https://europepmc.org/article/PMC/PMC13666144},
  note = {CC BY},
}
RIS
TY  - JOUR
TI  - Diagnostic Accuracy of GPT-5 in Detecting and Localizing Acute Ischemic Stroke on Isolated Diffusion-Weighted Imaging: A Retrospective Case–Control Study
AU  - Emre Gökçen
AU  - Mikail Kuşdoğan
AU  - Mustafa Fatih Erkoc
AU  - Levent Albayrak
AU  - Güneş Seda Albayrak
AU  - Deniz Tanrıverdi
PY  - 2026
DA  - 2026/10/01
JO  - Journal of clinical medicine
DO  - 10.3390/jcm15197716
UR  - https://europepmc.org/article/PMC/PMC13666144
AB  - Background/Objectives: Multimodal large language models are emerging tools for medical image interpretation, but their performance on isolated diffusion-weighted imaging (DWI) for acute ischemic stroke remains largely unexplored. We evaluated GPT-5 for stroke detection and vascular territory classification on preselected representative DWI images using a structured two-step prompting protocol. Methods: This single-center, retrospective, balanced case–control study included 220 cases (110 stroke patients and 110 stroke-negative controls). Four representative axial DWI slices per case were provided without apparent diffusion coefficient maps. Step 1 queried stroke presence and vascular territory; Step 2 queried the middle cerebral artery (MCA) segment-associated territory (M1–M4) following an MCA response. The reference standard was radiologist–neurologist consensus based on full multimodal magnetic resonance imaging, blinded to GPT-5 outputs. Diagnostic performance metrics and Cohen’s kappa (κ) were calculated. Results: Stroke detection showed a sensitivity of 100% (95% confidence interval [CI], 96.6–100%), specificity of 83.6% (95% CI, 75.6–89.4%), accuracy of 91.8%, and κ = 0.836. Accuracy was 93.6% for anterior/posterior circulation (κ = 0.838) and 90.9% for vascular territory (κ = 0.799). Conditional MCA segment-associated territory accuracy was 71.6% (53/74; κ = 0.602); end-to-end accuracy was 66.3% (53/80). Exploratory analysis of the 18 false-positive cases identified four overlapping error categories: artifactual and physiologic signal misinterpretation (72.2%), chronic lesion misclassification (27.8%), terminological inconsistency or unsupported report statements (77.8%), and etiological scenarios or unhedged definitive diagnoses (38.9%). Conclusions: GPT-5 achieved high sensitivity and a specificity of 83.6% for stroke classification on preselected representative DWI images using a vascular territory-based diagnostic framework. However, the 16.4% false-positive rate indicates that positive findings require radiologist confirmation. These findings support further evaluation of GPT-5 as a potential clinician-assisted image-interpretation tool but do not establish a clinical decision-support role.
SN  - 2077-0383
ER  -