The current diagnosis methods of arterial stenosis need to rely on large-scale equipment, which cannot provide early warning for patients with arterial stenosis and make them miss the best treatment opportunity. Therefore, we present an algorithm for the diagnosis of arterial stenosis, which is based on neural network. The input of neural network is peripheral pulse wave signal, and the output is the degree and location of arterial stenosis. The acquisition of peripheral pulse wave signal is convenient, and patients can detect it at home. In this paper, we extend firstly the existing human arterial transmission line model to include cerebral vessels, so that we can study cerebral artery stenosis. Then, we use the extended transmission line model to verify the effect of arterial stenosis (including carotid stenosis, renal artery stenosis, middle cerebral artery stenosis) on peripheral pulse wave signal. Finally, a database of peripheral pulse wave was established to simulate the actual conditions of artery stenosis based on the transmission line model. In addition, a diagnosis model of artery stenosis was constructed by using artificial neural network (ANN). The diagnostic results showed that when the degree of stenosis varied from 10 to 90%, the diagnostic accuracy was 88.7%. For moderate (50%) and severe (90%) degrees of stenosis, the diagnostic accuracy was 90.5% and 98.7%, respectively. Moreover, when the degree of stenosis was greater than 50%, the diagnostic accuracy of narrow positioning is greater than 95%. In general, the proposed algorithm does not rely on large equipment and has high accuracy, which can provide patients with early warning of arterial stenosis.
Read full abstract