Bài giảng Biomedical signal processing and modeling - Pattern Classification and Diagnostic Decision

Nội dung text: Bài giảng Biomedical signal processing and modeling - Pattern Classification and Diagnostic Decision

1. Nguyễn Công Phương BIOMEDICAL SIGNAL PROCESSING AND MODELING Pattern Classification and Diagnostic Decision
2. Contents I. Introduction II. Concurrent, Coupled, and Correlated Processes III. Filtering for Removal of Artifacts IV. Detection of Events V. Analysis of Waveshape and Waveform Complexity VI. Frequency Domain Characterization VII.Modeling Biomedical Systems VIII.Analysis of Nonstationary and Multicomponent Signals IX. Pattern Classification and Diagnostic Decision sites.google.com/site/ncpdhbkhn 2
3. Pattern Classification and Diagnostic Decision 1. Introduction 2. Pattern Classification 3. Supervised Pattern Classification 4. Unsupervised Pattern Classification 5. Probabilistic Models and Statistical Decision 6. Logistic Regression Analysis 7. Neural Networks sites.google.com/site/ncpdhbkhn 3
4. Introduction • An important final purpose of biomedical signal analysis is to classify a given signal into one of a few known categories and to arrive at a diagnostic decision regarding the condition of the patient. • A physician or medical specialist may achieve this goal via visual or auditory analysis of the signal presented. • However, when parameters such as AR-model coefficients and spectral statistics are derived, a human analyst is not likely to be able to comprehend and analyze the features. • Furthermore, as the number of the computed features increases, the associated diagnostic logic may become too complicated and unwieldy for human analysis. • Computer methods would then be desirable to realize the classification and decision process. sites.google.com/site/ncpdhbkhn 4
5. Pattern Classification and Diagnostic Decision 1. Introduction 2. Pattern Classification 3. Supervised Pattern Classification 4. Unsupervised Pattern Classification 5. Probabilistic Models and Statistical Decision 6. Logistic Regression Analysis 7. Neural Networks sites.google.com/site/ncpdhbkhn 5
6. Pattern Classification (1) • Pattern recognition or classification may be defined as categorization of input data into identifiable classes via the extraction of significant features or attributes of the data from a background of irrelevant detail. • In biomedical signal analysis, after quantitative features have been extracted from the given signals, each signal may be represented by a feature vector T x = [ x1, x2, , xn] • which is also known as a measurement vector or a pattern vector. • When the values xi are real numbers, x is a point in an n-dimensional Euclidean space. Vectors of similar objects may be expected to form clusters. sites.google.com/site/ncpdhbkhn 6
7. Pattern Classification (2) • For efficient pattern classification, measurements that could lead to disjoint sets or clusters of feature vectors are desired. • This point underlines the importance of appropriate design of the preprocessing and feature extraction procedures. • Features or characterizing attributes that are common to all patterns belonging to a particular class are known as intraset or intraclass features . • Discriminant features that represent differences between pattern classes are called interset or interclass features . • The pattern classification problem is that of generating optimal decision boundaries or decision procedures to separate the data into pattern classes based on the feature vectors. sites.google.com/site/ncpdhbkhn 7
8. Pattern Classification and Diagnostic Decision 1. Introduction 2. Pattern Classification 3. Supervised Pattern Classification 1. Discriminant and decision functions 2. Fisher linear discriminant analysis 3. Distance functions 4. The nearest-neighbor rule 4. Unsupervised Pattern Classification 5. Probabilistic Models and Statistical Decision 6. Logistic Regression Analysis 7. Neural Networks sites.google.com/site/ncpdhbkhn 8
9. Discriminant and decision functions (1) = + ++ + = T d()x w1122 x w x ... wn x n w n + 1 w x >0, if x ∈C d(x ) = wT x 1 ≤ ∈ 0, if x C2 >0, if x ∈C = T i di(x ) w i x ≤ 0, otherwise sites.google.com/site/ncpdhbkhn 9
10. Discriminant and decision Ex. 1 functions (2) 2.5 pw= a × pl + b 2 1. 5=a × 1 + b 0=a × 4 + b 1.5 →=−a0. 5 , b = 2 PW 1 Setosa → =− + Versicolor pw0. 5 pl 2 Virginica 0.5 →pw +05. pl −= 2 0 0 →=d pw +0. 5 pl − 2 1 2 3 4 5 6 7 PL >0, ifx ∈ Versicolor d(x )= pw +0 . 5 pl − 2 ≤0, ifx ∈ Setosa sites.google.com/site/ncpdhbkhn 10