The choice on surgery timing, early and delayed, is a noteworthy problem in aneurysmal SAH patients. Through the analysis of large patient populations where therapeutic strategy and surgical results are compared, the controversy is still open and only partially solved. Aim of the events sequence characterizing the evolution of SAH condition by mean of a Neural Network. Our patients with SAH due to the rupture of a supratentorial aneurysm who underwent surgery were considered. Timing was selected by an Expert and an 'a priori' surgery timing was never taken into account. Each patient (pattern) was described according to diagnostic clinical-instrumental parameters: Hunt-Hess scale; Ct scan according to the Fisher scale; Transcranial Doppler, through Vm and Gosling index, was used to demonstrate the vasospasm presence. The input nodes of patterns were elaborated by a self-associative neural network of Multilayer Perceptrone type which, through the classic back-propagation technique, elaborated the timing of surgery. Data from the network analysis lead us to state the followings: 1) there is a difference between the Expert and Network classification; 2) the coincidence between the two classification is greater in early surgery, whereas decreases in delayed surgery ones. It also varies in relation to the site of aneurysms. The analysis through an expert system of multifactorial and complex biological realities, though initial, could represent an innovative approach and contribute to a better knowledge and facilitate his solution.
LA SCELTA DEL 'TIMING' CHIRURGICO NEI PAZIENTI CON EMORRAGIA SUBARACNOIDEA DA FISSURAZIONE DI ANEURISMA DEL CIRCOLO CEREBRALE SOPRATENTORIALE: ANALISI MEDIANTE RETE NEURALE
ARENA, Paolo Pietro;
1994-01-01
Abstract
The choice on surgery timing, early and delayed, is a noteworthy problem in aneurysmal SAH patients. Through the analysis of large patient populations where therapeutic strategy and surgical results are compared, the controversy is still open and only partially solved. Aim of the events sequence characterizing the evolution of SAH condition by mean of a Neural Network. Our patients with SAH due to the rupture of a supratentorial aneurysm who underwent surgery were considered. Timing was selected by an Expert and an 'a priori' surgery timing was never taken into account. Each patient (pattern) was described according to diagnostic clinical-instrumental parameters: Hunt-Hess scale; Ct scan according to the Fisher scale; Transcranial Doppler, through Vm and Gosling index, was used to demonstrate the vasospasm presence. The input nodes of patterns were elaborated by a self-associative neural network of Multilayer Perceptrone type which, through the classic back-propagation technique, elaborated the timing of surgery. Data from the network analysis lead us to state the followings: 1) there is a difference between the Expert and Network classification; 2) the coincidence between the two classification is greater in early surgery, whereas decreases in delayed surgery ones. It also varies in relation to the site of aneurysms. The analysis through an expert system of multifactorial and complex biological realities, though initial, could represent an innovative approach and contribute to a better knowledge and facilitate his solution.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.