Background: 13C-Octanoic ac. breath test provides accurate measurement of gastric emptying time with good correlation to scintigraphy. However, long breath sampling time and complicated mathematics are requested. We have shown that neural networks analysis of breath test 13C excretion data is as valid, and accurate as Ghoos’s mathematical model. Aim: To verify if the neural network method is also able to reduce the frequency and number of breath samples, we processed the data from 13C-Octanoic ac. breath test by both neural network and Ghoos’s mathematical model. Materials and Methods: 20 healthy volunteers and 40 functional dyspeptic patients (38F, 22M, mean age 41 yrs) performed 13C-octanoic ac. breath test. Breath samples were collected before and every 15 minutes for 6 hours after a standard test meal with 100 mg 13C-octanoic ac. (Cortex Italia, Milan, Italy) incorporated into egg yolk and analysed by AP 2003 isotope ratio mass spectrometer. On the basis of T½ parameter of gastric emptying according to Ghoos et al., all subjects were subdivided in the following sub-groups: A) normal (T½ 70±20) B) slightly delayed (>2 SD), C) mildly delayed (>4 SD), D) severely delayed (>6 SD). Furthermore, the %dose/h of 13C of each breath sample has been appropriately normalised and given to an unsupervised neural network. The network used for this study has an output organised in a 9x9 matrix, implemented within Matlab 4.3, on Pentium PC. We have performed both mathematical and neural network analysis on the whole number of breath samples, as well as on progressively reduced numbers of them. Results: The neural network showed a more stable behaviour than the mathematical analysis in response to input data reduction. In relation to the T1/2 obtained by Ghoos’ method with the whole data set, used as gold standard, the neural network classification agreed at 98.33% (1 error). By reducing the number of breath samples to six (30, 60, 90, 120, 180 e 240 minutes) the diagnostic agreement was kept excellent (92%=5 errors). On the contrary, the same Ghoos’ analysis agreed only at 68.33% (19 errors) when examine only the 6 above breath samples (p <0.01) Conclusion: Neural networks analysis of 13C excretion data simplifies 13C-octanoic ac. breath test providing accurate measurement of gastric emptying with only 6 breath samples.

13C-octanoic ac. Breath test: neural network analysis of 13C excretion data

MANSI, CARLO;GIACOMINI, MAURO;DULBECCO, PIETRO;SAVARINO, VINCENZO
1999

Abstract

Background: 13C-Octanoic ac. breath test provides accurate measurement of gastric emptying time with good correlation to scintigraphy. However, long breath sampling time and complicated mathematics are requested. We have shown that neural networks analysis of breath test 13C excretion data is as valid, and accurate as Ghoos’s mathematical model. Aim: To verify if the neural network method is also able to reduce the frequency and number of breath samples, we processed the data from 13C-Octanoic ac. breath test by both neural network and Ghoos’s mathematical model. Materials and Methods: 20 healthy volunteers and 40 functional dyspeptic patients (38F, 22M, mean age 41 yrs) performed 13C-octanoic ac. breath test. Breath samples were collected before and every 15 minutes for 6 hours after a standard test meal with 100 mg 13C-octanoic ac. (Cortex Italia, Milan, Italy) incorporated into egg yolk and analysed by AP 2003 isotope ratio mass spectrometer. On the basis of T½ parameter of gastric emptying according to Ghoos et al., all subjects were subdivided in the following sub-groups: A) normal (T½ 70±20) B) slightly delayed (>2 SD), C) mildly delayed (>4 SD), D) severely delayed (>6 SD). Furthermore, the %dose/h of 13C of each breath sample has been appropriately normalised and given to an unsupervised neural network. The network used for this study has an output organised in a 9x9 matrix, implemented within Matlab 4.3, on Pentium PC. We have performed both mathematical and neural network analysis on the whole number of breath samples, as well as on progressively reduced numbers of them. Results: The neural network showed a more stable behaviour than the mathematical analysis in response to input data reduction. In relation to the T1/2 obtained by Ghoos’ method with the whole data set, used as gold standard, the neural network classification agreed at 98.33% (1 error). By reducing the number of breath samples to six (30, 60, 90, 120, 180 e 240 minutes) the diagnostic agreement was kept excellent (92%=5 errors). On the contrary, the same Ghoos’ analysis agreed only at 68.33% (19 errors) when examine only the 6 above breath samples (p <0.01) Conclusion: Neural networks analysis of 13C excretion data simplifies 13C-octanoic ac. breath test providing accurate measurement of gastric emptying with only 6 breath samples.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11567/395916
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? 0
social impact