1998 Fiscal Year Final Research Report Summary
FUNDAMENTAL RESEARCH ON SOFT DSP AND ITS APPLICATION TO BIOINFORMATION PROCESSING---FUSION OF SOFT COMPUTING AND DIGITAL SIGNAL PROCESSING---
| Project/Area Number |
09650426
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
情報通信工学
|
|---|
| Research Institution | MEIJI UNIVERSITY |
|---|
Principal Investigator |
ARAKAWA Kaoru MEIJI UNIVERSITY,SCHOOL OF SCIENCE AND TECHNOLOGY,PROFESSOR, 理工学部, 教授 (30183734)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
HARASHIMA Hiroshi UNIVERSITY OF TOKYO,FACULTY OF ENGINEERING,PROFESSOR, 工学部, 教授 (60011201)
|
|---|
| Project Period (FY) |
1997 – 1998
|
| Keywords | digital signal processing / biomedical signal processing / nonlinear signal processing / soft computing / fuzzy logic / lung sound / bio-information processing |
|---|
| Research Abstract |
In order to extract abnormal sounds, crackles, from lung sounds, a new digital signal processing methodology using soft computing is developed. First, the amplitude distribution and frequency characteristics of lung sounds are analyzed. On the basis of the fact that the abnormal sounds are characterized as nonstationary abrupt signals, a new filter structure which is composed of a prediction filter and a decision system based on fuzzy approximate reasoning is proposed. The part of the fuzzy approximate reasoning is realized as a simple nonlinear function. This filter can extract nonstationary abrupt signals by performing fuzzy judgment for the amplitude of the prediction error. However, since this filter cannot extract crackles precisely enough by itself, signal separation based on the signal amplitude and conditional averaging around the position of nonstationary abrupt signals are added to this filter. The obtained filter system is applied to about ten samples of actual lung sound data, and the high performance of this filter system is verified for both the signal waveform and the sound. Moreover, a new DSP (Digital Signal Processor) structure is proposed which is effective to implement this filter system. This crackle extracting system is realized as a combination of a linear weighted-sum and a simple nonlinear function. The conventional DSP, the structure of which is suitable for linear processing, is not effective for the proposed system. Therefore, a new type of nonlinear DSP which is the combination of the conventional linear DSP and a nonlinear function is proposed and the crackle extracting system is shown to be effectively realized by this DSP structure.
|
