Summary
Smart Info-Media Systems in Asia
2019
Session Number:SS3
Session:
Number:SS3-2
Study on CNN-based Stroke Diagnosis System
Su-min Jung, Taeg-keun Whangbo,
pp.58-62
Publication Date:2019/9/4
Online ISSN:2188-5079
Summary:
A stroke is the most common, very dangerous single-organ disease and aggravates social burden in the aging society. The stroke can be tested through a variety of imaging methods, among which a test method using CT imaging is known to deal promptly with an emergency patient in the early stage of stroke. The Alberta Stroke Program Early CT Score (ASPECTS) is widely used to assess the progress of stroke based on the CT imaging. It is problematic because of its presence of a scoring variability depending on a specialist’s career and individual variations. To resolve this issue, the scoring system proposes the deep learning system which can estimate ASPECTS automatically based on the CT imaging in order to help reduce the occurrence of scoring variability between specialists, as well as to improve decision making. The system uses NCCT brain scan images as inputs and creates outputs which estimated patient’s ASPECTS through three phases of imaging ? preprocessing, segmentation and deep learning. Each phase is designed to imitate experienced specialists’ stroke identification techniques by standardizing dataset and applying appropriate feature extractions on the neural network, based on image processing and deep learning.