學研新創 - 創新醫材與診斷技術
手提缺血性腦中風診斷系統
2020-12-23
蔡德明教授團隊/國立交通大學

蔡德明、陳湘頻、蔡侑展、蔡蕙安

EIT電阻抗成像(Electrical Impedance Tomography, EIT)是一種非侵入式(noninvasive) 重建人體內部組織影像之技術,透過量測人體內部的電阻分布以及演算法重建影像,可即時顯示人體內部組織之影像。交大團隊的新一代 EIT 技術,利用電流導引技術產生虛擬電極,並且加入 Machine Learning 與 Deep Learning,提高 EIT電阻抗成像影像解析度及準確性,將此新一代EIT技術應用於中風影像診斷。

中風疾病在全世界每年奪走六百多萬性命,五百萬人因此失能。對於失能者來說,由於日常生活功能衰退,生活及身體照顧需要部分或全部仰賴他人提供長期性協助。有98%照顧者表示因照顧者常是長年無休,導致身心疲憊、精神耗竭,照顧工作使他們感到沉重的負荷。本團隊研發之中風影像偵測帽 Neural Cap Imaging (NCI),是一個可攜帶的頭戴式影像偵測設備,可放置在救護車、醫院或診所內,並能有效協助醫護人員: 即時治療、迅速溝通。本團隊採用EIT電阻抗成相技術,通過給予安全的驅動電流在人體測量訊息,並重建人體內部的電阻分布。初步臨床數據顯示 sensitivity 與 specificity 接近 CT/MRI 相同水平。相較於NCCT (平均20~60分鐘) 和 DWI-MRI (平均20~40分鐘)在醫院的診察時間, NCI裝置僅需6分鐘以內的時間, 即可在電腦螢幕上成像, 供醫護人員判斷中風種類。如中風種類為缺血性, 醫護人員可立即做相關評估, 且可為適合的缺血性中風病患施打rtPA, 而不需等到醫院做 NCCT 或 DWI-MRI後, 再做後續治療。

評審推薦
1.開發的EIT技術含 signal processing system & machine learning system,可於7分鐘內判斷是缺血性中風或出血性中風,有利於後續處置,具創新性。
2.發展以非侵入式頭戴神經影像設備,即時偵測腦中風之狀況,可應用於救護車、醫院或診所,具臨床價值。
3.倍頻顯微術的關鍵技術,已取得美、中、台專利。
2023年度精進成果
Typically, acute stroke patients need to be imaged in medical centers using MRA (magnetic resonance angiography) or CTA (computed tomography angiography) to be diagnosed accurately. MRA and CTA are usually not available in ambulances due to the scanner size and weight. On the other hand, it is possible to use portable devices in ambulances to image the stroke brains and provide diagnoses before acute ischemic stroke patients arrive at the medical centers. A portable device with electrical impedance tomography capability can inject small current levels (e.g., 140µA at 1KHz sinusoidal signal) into stroke patients' brains and simultaneously measure potential distribution based on an electroencephalogram (EEG) cap. Those measurements can be transmitted and processed in an online or cloud computing center. The online computing center or the cloud computing center can use the following method to process the measurements to create an image of the stroke patient's brain conductivity distribution. This would allow the size and location of the ischemic stroke target to be identified. This would allow the emergency room or stroke center physicians more time to consider injecting tissue plasminogen (tPA) (within 4.5 hours of acute ischemic stroke onset) and extracting blood clots within 6 to 24 hours of stroke onset with mechanical thrombectomy. The method introduced in this method used the finite element analysis, genetic algorithm and divide-and-conquer method to study and analyze the electrical conductivity profile of simulated stroke patients based on a 45 dB signal-to-noise ratio (SNR) synthesized measurement. Clinical measurement data obtained from stroke patients with electrical impedance tomography was reported to be approximately 45 dB SNR. Weighting factors in conjunction with the divide-and-conquer method and adaptive genetic algorithm were used. The adaptive genetic algorithm was applied with the weighting factors to identify the conductivity distribution inside the brain by minimizing the difference between the simulation and synthesized measured voltages under stroke conditions. The result shows a pretty good reconstruction of the electrical conductivity profile of the stroke head based on 45 dB SNR simulated measurement of stroke data. This algorithm can be added to the original algorithm to improve further the overall result of the stroke patient's images.
團隊簡介
蔡德明  
學歷 PhD (McGill University)
BS (University of California, Berkeley) 
現職 國立交通大學電機工程系暨生醫工程研究所/電控工程研究所教授
經歷 Higher Education Academy, Senior Fellow
國立交通大學資訊工程系教授
國立交通大學電機資訊國際學位學程主任
國立交通大學電機資訊學院國際化與發展辦公室共同主任
University of California, Berkeley, Electrical Engineering and Computer Science Department, Visiting Professor
University of California, Santa Cruz, Computer Engineering Department, Visiting Professor
Stanford University, California Ear Institute, Visiting Research Scientist
Philips Technology Center, Principal Engineer

 

本網站中所有資料(包括影音.文字.圖表.數據等) ,均屬於本中心或各該新創企業團隊之專屬財產,如有引用,請確實註明出處來源。 <完整資訊>
財團法人生技醫療科技政策研究中心 版權所有
Copyright © 2012 - 2024 Research Center for Biotechnology and Medicine Policy (RBMP). All Rights Reserved