Abstract Details

Title Accuracy of Artificial Intelligence in Measuring Intracerebral Hemorrhage Volumes and Expansion Compared to Human Estimates

Topic Cerebrovascular Disease and Interventional Neurology

Presentation(s) S9 - Cerebrovascular Disease: Intracerebral Hemorrhage (4:42 PM-4:54 PM)

Poster/Presentation
Number 007

Objective

Artificial Intelligence (AI) can identify characteristics of hemorrhage on non-contrast head CT (NCCT). We aim to evaluate the accuracy and reliability of AI in measuring intracerebral hemorrhage (ICH) volumes and volume expansion on subsequent NCCTs more accurately than human estimates.

Background ICH expansion is an independent predictor of mortality and functional outcome with each milliliter of expansion increasing the chance of functional dependence by up to 7%. Detection of ICH expansion is often subjective and may misguide treatment. AI with convolutional neural networks (CNNs) represent a powerful new technology in image analysis and quantification. We compare the accuracy, sensitivity, and specificity between a CNN optimized for ICH volume quantification and a traditional ABC/2 methods.

Design/Methods We performed a retrospective analysis of ICH patients who have had at least one follow-up NCCT within 24 hours. ICH expansion was defined as an increase >33% in volume. Each ICH was manually segmented, serving as ground truth measurements. Comparison of ICH expansion was made using a traditional ABC/2 estimative and a previously validated hybrid 3D/2D mask ROI-based CNN for ICH evaluation, which was trained on over 10,000 patients. Accuracy, sensitivity, and specificity of the CNN and estimative approaches were compared.

Results A total of 230 patients were included for a total of 460 NCCTs. The true average ICH volume was 44.8 mL, 45.3 mL by CNN (Pearson 0.99), and 60.4 mL by ABC/2 (Pearson 0.81). Accuracy, sensitivity, and specificity for ICH expansion detection was 100%, 100%, and 100% for the CNN and 93.0%, 74.2%, and 96.0% for ABC/2. On visual inspection, cases of false positives by ABC/2 approaches tended to demonstrate eccentric expansion.

Conclusions

A customized deep learning tool is highly accurate in the detection for ICH expansion. This may have important implications clinically for management and surveillance as well as in a clinical trial setting.

Authors/Disclosures
Masaki Nagamine, MD (University of California, Irvine Medical Center) PRESENTER	Dr. Nagamine has nothing to disclose.
	No disclosure on file
Wengui Yu, MD, PhD (UC Irvine, Neurology Dept)	Dr. Yu has nothing to disclose.
Peter Chang	No disclosure on file
	No disclosure on file

��ɫ��

��ɫ��