利用用正则化Fisher线性判别分析脑电信号诊断自闭症(IJIGSP-V4-N3-6)

problems
facing
rese earchers.
To o
reveal
the
ve pattern between b autis stic and nor rmal discriminativ children via electroenceph halogram (EE EG) analysis is a big challeng ge. The featur re extraction is averaged F Fast Fourier Tran nsform (FFT) with the Regulated R Fis sher Linear Discri iminant (RFL LD) classifier. Gaussinaty c condition for the optimali ity of Regula ated Fisher Linea ar Discriminan nt (RFLD) ha as been achie eved by a well-co onditioned app propriate prep processing of f the data, as wel ll as optimal shrinkage te echnique for the Lambda para ameter. Wins sorised Filtere ed Data gave the best result.
I.J. Image, Graphics an nd Signal Pr rocessing, 20 012, 3, 35-41 1
Published On nline April 20 012 in MECS (http://www.m ( mecs-press.org g/) DOI: 10.5815/ijigsp.2012. .03.06
Index Term ms—
Electro oencephalogra am, R Regularized
Automa ated lin near
sys stem perform mance and be etter predictio ons, but they y req quire a priori information on the syste em and often n nee ed more time and resources s [4]. a increasing g In recent years, there has been an inte erest in apply ying machine e learning me ethods to the e
autistic subjects. They also applied their work at beta band and had the same accuracy classification 82.4% [9]. The significance of classification accuracy was measured by using different machine learning algorithms: the k-nearest neighbors (k-NN), SVM and naïve Bayesian classification (Bayes) algorithms with mMSE as a feature vector which described by William, B., T. Adrienne, and N. Charles [10]. They used Net Station software for acquisition data and Orange software for machine learning classification. Their accuracy classification is over 80% accuracy into control and high risk for autism HRA groups at age 9 months. Classification accuracy for boys was close to 100% at age 9 months and remains high (70% to 90%) at ages 12 and 18 months. For girls, classification accuracy was highest at age 6 months, but declines thereafter. EEGLAB were used to extract evoked EEG features: raw EEG, CSD interpolated data, and backprojected IC features and also signal statistic was used to classify both groups. These data provide the first empirical demonstration of increased neural noise in those with ASD. Channel selection was based on an
aut tomated detec ction of autism m EEG signal ls [5, 6]. EEG G sig gnals analysis based on mac chine learning g methods has s three main steps s: preprocessi ing, feature ex xtraction, and d cla assification. is paper is to t utilize the e The major goal of thi Regularized isher’s Fi Line ear Discriminat (RFLD) )
Abstract— D Diagnosis of autism is one e of the diffi icult
ng on around the world to oday trying to o of research goin use e neuroscienc ce such as EEG study y to identify y ind dividuals with h autism. Hen nce, a need for f automatic c det tection of EE EG signals h has been soug ght by many y res searchers to diagnose d auti istic people. Furthermore, , the ey report diff ferent findings t discriminat t s regarding to pat tterns between n normal and a autism disorders [1, 2]. Many caus ses of autism m have been proposed, p but t und derstanding of the theory o of causation of o autism and d the e other autism spectrum diso orders is incomplete [3]. In n this s case, the phenomenol logical mode els are most t app propriate to be b applied tha an the mechan nistic models. . Me echanistic models m typic cally involve e physically y inte erpretable pa arameters, allo ow deeper insights into o
Facul lty of Comput ting and Infor rmation Techn nology
King Abd dulaziz Univ versity KAU, , Jeddah, Sau udi Arabia
abit, fdehlawi (miali, hmalibary, h Ma alhaddad, drtha i, eaalsaggaf, ahadi008@ka a au.edu.sa)
I.J. Im mage, Graphic cs and Signal P Processing, 2012, 2 3, 35-41
36
EEG based Autism Diagnosis Using Regularized Fisher Linear Discriminant Analysis
analysis in detecting the autistic children based on EEG signal analysis. Thus, optimum preprocessing -which gives the highest classification accuracy- is studied. The artifacts of the recorded EEG signals were removed by visual inspection. Then, different preprocessing techniques were applied such as Re-referencing, Filtering, Winsorizing, Scaling, Single epoch extraction and Feature vector construction. After preprocessing, FFT was used as features. Dimensionality reduction using decimation factor 2 was applied. Finally, the extracted features were classified using RFLD. This research is considered as part of the main BCI project in the King AbdulAziz University that is funded by (King AbdulAziz City for Science and Technology) KACST, 8-NAN106-3. The layout of the paper is as follows. Section 2 focuses on the literature review, the experiments that were performed and the methods used for data preprocessing, feature extraction are described in section 3. Classification is given in section 4. Results are discussed in section 5.
EEG G based d Auti ism Diagnosis Us sing R Regula arized Fish her Lin near D Discrim minant t Anal lysis
Mahmoud I. Kamel , Mohammed M J. J Alhaddad, Hussein M. Malibary, Khalid K Thabit t, Foud Dahlw wi, Ebtehal A. Alsa aggaf, Anas A. A Hadi
diagnosis,
ห้องสมุดไป่ตู้
Autism,
Fisher’s
st Fourier Tran nsform. discriminant analysis, Fas
I. INTRODU UCTION Autism is a disorder rather than an n organic dise ease is of autism is s one of the difficult d proble ems and diagnosi facing resear rchers and th hose interested d in the field d of signal proces ssing and med dicine. Theref fore, there is a lot Copyright © 2012 MECS