kut-paper-template.dvi

Size: px

Start display at page:

Download "kut-paper-template.dvi"

ほだかつねざき
4 years ago
Views:

1 26 Discrimination of abnormal breath sound by using the features of breath sound

2 ,,,,,,,,,,,,, i

3 Abstract Discrimination of abnormal breath sound by using the features of breath sound SATO Ryo Auscultation of breath sounds is one of the most important method to examine in the scene of home nursing care If abnormal breath sound is found, it needs to be treated immediately because there is a high possibility of serious illness The system to integrate auscultatory sound is planned Nurses sends breath sound data to hospitals, and doctors diagnoses it But doctors can t diagnose a large amount of breath sound data enough This is a possibility that these problems are solved by the automatic detection system of abnormal breath sound In this paper, I have proposed the system for automatic discrimination of abnormal breath sound by using the Support Vector Machine(SVM) SVM is one of the pattern recognition algorithm and two-class classifier In proposed method, wavelet transform applies breath sound to get features quentity First, training database is built by training data with features quentity Next, classification criteria defines features quentity Finally, breath sound is discriminated normal breath sound from abnormal breath sound by classification criteria of SVM classifier As a result of the discrimination I confirmed success that discriminate actually breath sound data by using SVM classifier key words Support Vector Machine, breath sound, home nursing care, auscultation ii

4 iii

5 iv

6 v

7 31 ( ) ( ) ( ) ( ) (1) (2) (3) (4) (1) (2) (3) (4) 26 vi

8 1 11,, [1],, 11,,,,,,,,,,, 1

9 12, ,, 3,,, 4, 2

10 2,, 21 2,, 21 3

11 21,, 21, [2], 22 (1) (2) 22, 4

12 22 22 [3][4] (SVM: Support Vector Machine) 1990 Vladimir N Vapnik,, 2, D L = {(t i,x i )} (i = 1,,N) t i = { 1,+1}, x i R d (R d : d ) 23 k, w T x i +b k (21) k w b, t i = +1 w T x i +b +1 (22) t i = 1 w T x i +b 1 (23), t i (w T x i +b) 1 (24) 5

13 22, w w T x ρ(w,b) = min x C y=+1 w max w T x x C y= 1 w = 1 b w 1 b w = 2 w w0x+b t 0 = 0, ρ(w 0,b 0 ) = maxρ(w,b) (25) w D max 1/2, t i (w T x i +b) 1 (i = 1,,N), w w 0 = min w (26) 2 SVM, , [2] d, d SVM, d SVM, 6

15 3,,,

16 SVM 3 4 5,

17 32 32,, ( 33) 33 2 [5][6] ( 250msec ) ( ) ( ), [5][6], 10

18 33 ( ) ( ),, 33 A-1 A-7 B-1 B-7 CD [5] CD [6],, [2] ( 36-39) A-1 A-7, B-1 B-7 33 SVM, 34 m, m 1 n = 1, n = m Leave-one-out, [4],, Leave-one-out, Leave-one-out Leave-one-out SVM 11

19 SVM, A-1 A-7 B-1 B-7 SVM 31, 57% A-3 A-6, B-5 B A-3 A-6,,, B-4 B-6, B-5 B-7 B-5 B-7 12

20 33 31 ( ) (1) ( ) A-1 A-2 A-3 A-4 A-5 A-6 A (2) ( ) B-1 B-2 B-3 B-4 B-5 B-6 B (3)Leave-one-out 0570, [6],,, 333 A-3 A-6 Leave-one-out, 32, 13

21 34 32 ( ) (1) ( ) A-1 A-2 A-4 A-5 A (2) ( ) B-1 B-2 B-3 B-4 B-5 B-6 B (3)Leave-one-out ,,, 34, 2 C1-C5 14

22 34 D1-D , C D, C-6, 35, ( ) 341 SVM 33, 15

23 34 33 (1) ( ) C-1 C-2 C-3 C-4 C (2) ( ) D-1 D-2 D-3 D-4 D-5 D ,,,,, SVM , 16

24 SVM, ( ) (1) ( ) A-1 A-2 A-4 A-5 A (2) ( ) B-1 B-2 B-3 B-4 B-5 B-6 B (3)Leave-one-out 0862, , 344, 0, SVM, 17

25 35 35 ( ) (1) ( ) C-1 C-2 C-3 C-4 C (2) ( ) D-1 D-2 D-3 D-4 D-5 D SVM,,,,,, SVM, 18

26 (1) A A A A A A A B B B B B B B

27 36 37 (2) A A A A A A A B B B B B B B

28 36 38 (3) A A A A A A A B B B B B B B

29 36 39 (4) A A A A A A A B B B B B B B

30 (1) C C C C C D D D D D D

31 (2) C C C C C D D D D D D

32 (3) C C C C C D D D D D D

33 (4) C C C C C D D D D D D

34 4 41,, SVM SVM,, SVM,, SVM,,, 42 16,, 2 27

35 42,,,, 28

36 ,,,,,,,,, NOC FreeBSD,,, 1 4,,,,,,, 4, 29

37 [1],,, 2009 [2],, 2015 [3],,, 2007 [4],,, 2012 [5],,,, CD,, 2014 [6],,,,, CD,,

25 Removal of the fricative sounds that occur in the electronic stethoscope

25 Removal of the fricative sounds that occur in the electronic stethoscope 1140311 2014 3 7 ,.,.,.,.,.,.,.,,.,.,.,.,,. i Abstract Removal of the fricative sounds that occur in the electronic stethoscope