biomedical signal processing

A research perspective on biomedical Signal processing

1. BIO MEDICAL SIGNAL
PROCESSING:
RESEARCH PERSPECTIVE
Dr. Meenakshi Sood
Department of Electronics & Communication Engg
Jaypee University of Information Technology
[email protected], [email protected]

2. Biomedical SIGNAL PROCESSING
•Application of engineering principles and techniques to
the medical field to close the gap between engineering
and medicine.
•Guide the medicine to use innovative educational tools
such as humanistic models, realistic simulations, web-
based online resources, etc.
•It combines the design and problem solving skills of
engineering with medical and biological sciences to
improve healthcare diagnosis and treatment.
11/6/2020
M Sood JaypeeUniversity of
Information Technology
2

3. The Need…
•Medical decision-making requires the clinician to apply
accumulated knowledge to a specific amount of patient
information to produce a result that may be
a diagnosis,
prognosis,
course of therapy, or
the selection of further tests.
•Limited resources -increased demand
•Insufficient time available for diagnosis and treatment.
•Need for systems that can improve health care processes
and their outcomes in this scenario
11/6/2020 3

4. Sub-disciplines
•Bioinstrumentation
•Biomaterials
•Biomechanics
•BioInformatics
•Biomedical computing & signal processing
•Medical Imaging
•OrthopaedicBioengineering
•Rehabilitation Engineering
•Minimally invasive surgery
11/6/2020
Jaypee University of Information
Technology
4

5. Application Areas

6. Examples of Biomedical Signals
Electroneurogram (ENG)
Electromyogram (EMG)
Electrocardiogram (ECG)
Electroencephalogram (EEG)
Electrogastrogram (EGG)
Electroocculogram (EOG)
Phonocardiogram (PCG)
Vibromyogram (VMG)
Vibroarthogram (VAG)
11/6/2020
Jaypee University of Information
Technology
6

7. ECG(electrocardiograph)
•electrical activity of the heart
11/6/2020
Jaypee University of Information
Technology
7
EMG
EMG (Electromyogram)
signal generated by muscle cells

8. EGG(Electrogastrogram)
•The electrical activity of thestomach consists of rhytmicwaves of
depolarization and repolarizationof its constituent smooth muscle
cells.
11/6/2020
Jaypee University of Information
Technology
8
EOG(Electroocculogram)
•Eye movements produce a moving (rotating) dipole
source and, accordingly, signals that are a measure of
the movement are obtained .

9. IMAGES-2D SIGNALBenign Malignant
CT Images
US Images
Microscopic Images of Blood
MRI Images
DNA sequence signal

10. RESEARCH FIELDS
User/Patient
Interface
Technology
Signal Analysis
Pre processing
Data Acquisition
Statistical Analysis:
Pattern Recognition
Initiate a warning or a variety
of therapies
Stimulator
Drug
Optimization:
Feature Extraction/
Clustering
NurseFeature 1
Feature 2
Feature 3
11/6/2020
Jaypee University of Information
Technology
10

11. RESEARCH GAPS
11/6/2020
Jaypee University of Information
Technology
11
Signal Conditioning
Transformation and reduction of the signals
Identification of Diagnostic features
Non Stationarity
Non Linearity
Optimization
Classifiers
Pattern Recognition

12. Signal Conditioning
11/6/2020
Jaypee University of Information
Technology
12

13. Signal transformation
•Noisecomponent:
–duetotheelectronicsinthemeasuringdevice,
–artifactsrelatedtothepatient’smovements,or
–otherbackgroundsignalsrecordedsimultaneously
•Moredatathanactuallyneededtoderiveparametersoffering
semanticinformation
11/6/2020
Jaypee University of Information
Technology
13

14. Parameter/ Featureselection
•Usually, relevant information is not the direct result
of a sample or recording of a signal.
•Parametersbearing resemblance to the signs and
symptoms that are used to make diagnosis are
extracted from the signal.
11/6/2020
Jaypee University of Information
Technology
14

15. Feature extraction

16. Feature selection

17. Feature Selection
Steps
•Feature selection is an
optimizationproblem.
11/6/2020
JaypeeUniversity of Information
Technology
17
Searchstrategies
Optimum
Heuristic
Randomized
Evaluationstrategies
-Filter methods
-Wrapper methods

18. Optimization Techniques
11/6/2020
Jaypee University of Information
Technology
18

19. Classification methods
•Classificationalgorithmslearnfromthegiveninput
dataandusesthislearningtoclassifynewsetof
observations.
•Differentclassificationalgorithmsthatcanbeused
are:Classification Methods
Nearest
Neighbor
Naïve
Bayes
Decision
Tree
Random
Forest
ANNPNNSVMANFC

20. 11/6/2020
M Sood Jaypee University of
Information Technology
20
Signal classification:
Computer Aided Diagnostic system

21. Transition…….
11/6/2020
M Sood Jaypee University of
Information Technology
21

22. 11/6/2020
M Sood Jaypee University of
Information Technology
22
Machine Learning approach:

23. 11/6/2020
M Sood Jaypee University of
Information Technology
23

24. Deep Learning approach
11/6/2020
M Sood Jaypee University of
Information Technology
24

25. 11/6/2020
M Sood Jaypee University of
Information Technology
25

26. 11/6/2020
M Sood Jaypee University of
Information Technology
26
ConvolutionalNeural Network

27. EXAMPLE
11/6/2020
M Sood Jaypee University of
Information Technology
27

28. Research and Application
•PhyologicalResearch
•Neurological Research
•Medical Research
•BioInformaticsResearch
•Educational Research and Application
•Therapeutic Application
11/6/2020
Jaypee University of Information
Technology
28