非侵入式血液透析動靜脈廔管狹窄偵測方法

Abstract

根據美國腎臟登錄系統年報公佈，全球罹患慢性腎衰竭的病患占總人口比率與每年新增加慢性腎衰竭人口比率逐年增加。血液透析是慢性腎衰竭病患常用的腎臟功能替代方式。在血液透析治療過程之前，會先透過外科手術將病患手上的動脈與靜脈接合，以合成動靜脈廔管，以利血液透析進行。動靜脈廔管的性能與壽命會受到外科手術併發症與重複穿刺的影響。狹窄通常是導致動靜脈廔管功能性喪失及完全閉塞最常見的原因。目前評估動靜脈廔管狹窄是藉由醫生的經驗觸診以及利用聽診器聽診廔管的壓力以及血流量。若有疑似狹窄需要做進一步的評估時，才會進行較客觀的超音波檢查以及血管造影檢查。這些設備不僅價格昂貴，而且需要受過專門訓練的技術人員進行操作。本論文提出血液透析用的動靜脈廔管狹窄偵測系統以及動靜脈廔管狹窄監測演算法，可以搭配血液透析儀使用，其使用非侵入式、無線可攜帶，並且可以長時間監測病患是否發生廔管狹窄的症狀，提供可靠之方式提供臨床判定動靜脈狹窄的一項依據，進而讓醫師提早偵測而進行後續醫療處理，以維持血液透析治療之品質及提昇病患之存活率。本論文的模擬與實驗結果顯示此方法在臨床上具有很高判斷性。

End-Stage renal disease (ESRD) is a chronic renal disease, which is a progressive loss in renal function over a period. The most common treatment for ESRD patients is hemodialysis. For hemodialysis, a steady vascular access has to be created by cardiovascular surgical anastomosis of artery and vein, referred to as arteriovenous fistula (AVF). The vascular thrombosis or stenosis may occur over time to result in the arteriovenous fistulas dysfunction. The angiography and Doppler ultrasonography are costly and need sophisticated operators to manipulate them. The acoustic characteristic of vascular stenosis can also be evaluated roughly by the clinical experience of physicians or nurses via stethoscope auscultation or palpation. The proposed method was using S-transform (ST) to extract the feature of the blood flow sounds. Then, the radial basis function (RBF) neural network was used to classify the feature of the blood flow sound. The algorithm can maintain high performance, thus arteriovenous fistula stenosis detection (AFSD) algorithm can be viewed as a practical method.