Stroke Localization Using Multiple Ridge Regression Predictors Based on Electromagnetic Signals
Article
Article Title | Stroke Localization Using Multiple Ridge Regression Predictors Based on Electromagnetic Signals |
---|---|
ERA Journal ID | 213646 |
Article Category | Article |
Authors | Gao, Shang, Zhu, Guohun, Bialkowski, Alina and Zhou, Xujuan |
Journal Title | Mathematics |
Journal Citation | 11 (2), pp. 1-9 |
Article Number | 464 |
Number of Pages | 9 |
Year | 2023 |
Publisher | MDPI AG |
Place of Publication | Switzerland |
ISSN | 2227-7390 |
Digital Object Identifier (DOI) | https://doi.org/10.3390/math11020464 |
Web Address (URL) | https://www.mdpi.com/2227-7390/11/2/464 |
Abstract | Localizing stroke may be critical for elucidating underlying pathophysiology. This study proposes a ridge regression–meanshift (RRMS) framework using electromagnetic signals obtained from 16 antennas placed around the anthropomorphic head phantom. A total of 608 intracranial haemorrhage (ICH) and ischemic (IS) signals are collected and evaluated for RRMS, where each type of signal contains two different diameters of stroke phantoms. Subsequently, multiple ridge regression predictors then give the target distances from the antennas and mean shift is used to cluster the predicted stroke location based on these distances. The test results show that the training time and economic cost are significantly reduced as the average prediction time only takes 0.61 s to achieve an accurate result (average position error = 0.74 cm) using a conventional laptop. It has great potential to be used as an auxiliary standard medical method, or rapid diagnosis of stroke patients in underdeveloped areas, due to its rapidity, good deployability, and low hardware cost. |
Keywords | antenna array; electromagnetic signal; machine learning; mean-shift; ridge regression; stroke; vector network analysis |
ANZSRC Field of Research 2020 | 460299. Artificial intelligence not elsewhere classified |
Byline Affiliations | University of Queensland |
School of Business |
https://research.usq.edu.au/item/v86y8/stroke-localization-using-multiple-ridge-regression-predictors-based-on-electromagnetic-signals
Download files
76
total views38
total downloads1
views this month1
downloads this month