Transcript Chapter7_Cli_Eng_Anaesthesia

Hierarchical Monitoring and Fuzzy
Logic Control in Anaesthesia
What is anaesthesia ?
* Art or Science
* Loss of Awareness
* Loss of all Sensation (Pain, Temp., Position)
* The process should be reversible
* Modern general anaesthesia (TRIAD)
Unconsciousness
Balanced
Anaesthesia
Analgesia
Muscle Relaxation
Automatic Control in Anaesthesia
The main problem : measurement of clinical
signs for on-line input to the system
* Muscle Relaxation:
direct measurement: EMG
control algorithm: PID, GPC, FLC, etc.
* Unconsciousness (Depth of Anaesthesia):
no direct measurement
Indirect measurement:
(a) SAP, HR
(b) SW, LA, PR, MO
(c) End Tidal of AA, MAC,
(d) Plasma concentration
(e) Heart Rate Variability
(f) Brain Signals
EEG, AEP, SEP
Etc.
Interpretative algorithm:
(a) Aperiodic Analysis
(b) Fourier Transform Analysis
(c) Auto-Regression Algorithm
(d) Fuzzy Logic
(e) Neural Networks
Etc.
Control algorithm:
(a) PID, GPC
(b) Fuzzy Logic
(c) Neural Networks
Etc.
* Analgesia (Pain Relief)
no direct measurement
subjective indirect measurement
brain signals (i.e. SEP)
Postoperative conditions
Cancer pain
PCA (Patient-Controlled Analgesia)
Fuzzy logic features:
* Can reason with imprecise data
* Leads to "soft computing"
* Concept of "machine IQ"
* Cope with non-linear, complex,
unknown processes
* Link with neural networks
* Link with genetic algorithms
* Lack of stability theory
Anaesthetists use “rules of thumb”
“imprecise, personal rules”
Ex:
IF T1% is greater than the set point by
a LARGE AMOUNT
THEN set the atracurium infusion rate to
a HIGH LEVE
This rules contains imprecise terms:
a LARGE AMOUNT
a HIGH LEVEL
In Clinical Engineering:
1988
1988
1992
1994
1995
Linkens & Mahfouf (UK)
Muscle Relaxation (Simulation)
Sheppard & Ying (USA)
MAP, SNP (Sodium Nitroprusside)
Hacisalihzade et al. (Switzerland)
MAP, Isoflurane
Tsutsui & Arita (Japan)
SAP, Enflurane
Shieh et al. (UK)
DOA, Propofol & Isoflurane
1995 Zbinden & Hacisalihzade (Switzerland)
MAP, Isoflurane, Human
1996 Schaublin & Zbinden (Switzerland)
End Tidal CO2, Ventilation (Fre., Vol.)
1996 Curatolo & Zbinden (Switzerland)
Fi (Iso.) & O2 conc., min. flow
1996 Mason et al. (UK)
Muscle relaxation (Atracurium)
1996 Shieh et al. (ROC)
Muscle Relaxation (Atr.)
1997 Mason et al. (UK)
Muscle relaxation (Atracurium), SOFLC
1997 Shieh et al. (ROC)
Muscle Relaxation (Miv.)
1998 Shieh et al. (ROC)
Unconsciousness (Desflurane)
2000 Shieh et al. (ROC)
Muscle Relaxation (Roc.)
Computer Monitoring and Control
in Muscle Relaxation
Introduction:
Short-acting non-depolarizing relaxants
Advance of modern computer technology
The purpose of this approach:
Small, handy and easy to use
Clinical Methods of Nerve Stimulation:
• Single-twitch Stimulation
• Train-of-four Stimulation
• Double-burst Stimulation
• Tetanic Stimulation
Recording of Evoked Responses:
• Mechanical Responses
• Electromyographic Responses
• Accelerative Responses
Mechanical
Responses:
Electromyographic
Responses:
Accelerative
Responses:
Requirements for the ideal neuromuscular
blocking agent:
1. Non-depolarizing mechanism of action
2. Rapid onset of action
3. Short duration of action
4. Rapid recovery
5. Non cumulative
6. No cardiovascular side effects
7. No histamine release
8. Reversible by cholinesterase inhibitors
9. High potency
10. Pharmacologically inactive metabolites.
The characteristics of Atracurium, Mivacurium
and Rocuronium :
1. An aqueous for intravenous injection
2. Conc.: 10 (A), 2 (M), 10 (R) mg/ml
3. Non-depolarizing neuromuscular blocking drug
4. Onset time: 1.5 (A), 2.5 (M), 1 (R) min
5. Duration of action: Intermediate (A, R), Short(M)
6. Recovery: Intermediate (A, R), Rapid (M)
7. Cardiovascular effect: Yes(A, M), No (R)
8. Histamine release: Yes (A, M), No (R)
9. Metabolites.
Atracurium: Laudanosine (Central effects)
(Cisatracurium: 1/3 Laudanosine)
Mivacurium: Enzyme
Rocuronium : Liver, Kidney
(Vecuronium: causing prolonged block)
The Wellcome Foundation (A, M): J. Savarese
Organon Teknika ( R )
Hierarchical Monitoring of EMG via filters
* Built-in filter:
Noise
High Frequency Disturbance
* Pharmacological filter:
T4/T1
T2/T1, T3/T2, T4/T3
* Median filter
x1, x2, x3,x4,x5 => Median Value
Example: T1%
7, 17, 11, 10, 8 => 10
Hierarchical Monitoring and
Fuzzy Logic Control
S.P.+ E,CE
-
Emergency
Table
Median
Filter
Coarse
Table
Pharmacological
Filter
Self-Tuning
Instrument
Filter
FLC
(Fine Table)
Patient
EMG
EMG (F)
Manual Control for Atracurium:
100
80
60
40
20
0
0
360
720
1080
Time (Samples of 10 s)
(b)
100
T1(%, F)
Pt. Weight: 54 kg;
Pt. Age: 34 yr;
Sex: Female;
Clinical Diagnosis:
Lipoma
retroperitoneal
tumour
Operation:
Debaking
T1(%)
(a)
80
60
40
20
0
0
360
720
Time (Samples of 10 s)
1080
0
180
360
540
Time (Sec.*10)
720
360
540
720
50
40
30
20
10
0
50
40
30
20
10
0
0
180
360
540
Time (Sec.*10)
720
DC
T1 (F)
180
Time (Sec.*10)
IR
0
0.2
0.1
0
-0.1
-0.2
0
100
50
2
1.5
1
0.5
0
STP
Pt. Weight = 58.5 kg;
Pt. Age = 45 yr;
Sex : Female
Clinical Diagnosis: CPS;
Operation: FESS
MIR
Automatic Control for Mivacurium:
Summary: (Muscle Relaxation)
Clinically, this system was useful:
* Provided stable surgical operating conditions
* Minimized the amount of neuromuscular
blocker required by each patient
* Reduced the need for the anaesthetist to
spend time controlling neuromuscular block
* Allowed reliable antagonism of
neuromuscular block at the end of surgery
Automatic Control of Anaesthesia
with Desflurane Using Hierarchical
Structure
Unconsciousness
(Depth of Anaesthesia ) :
no direct measurement
indirect measurement:
SAP, HR
SW, LA, PR, MO
End Tidal of AA, MAC
Plasma concentration
Brain Signals (i.e., EEG, AEP, SEP)
Etc.
An Automatic Control System for
Inhalational Anaesthesia
PATIENT
Vaporizer
Datex AS/3
Stepping
Motor
SAP, HR, AA
COMPUTER
Control Box
Supervision
by Anaesthetist
200
150
100
50
0
0
50
100
150
200
250
200
250
Time (Sec.*30)
12
100
AA (%)
Contuoller Output
Pt. Weight: 52 kg;
Pt. Age: 19 yr;
Sex: Female;
Clinical Diagnosis: CPS
Operation: FESS
BP (mmHg)
A Clinical Trial (Unconsciousness)
75
50
25
8
4
0
0
0
50
100
150
Time (Sec.*30)
200
250
0
50
100
150
Time (Sec.*30)
Modelling of Anaesthesia for Unconsciousness
Fiaa
Gender
Age
Weight
Neural
Networks
Patient
Model
HR
SAP
Fuzzy
Model
SDOA
BIS
Etaa
Fuzzy
Model
PDOA
Neural Networks
Vaporizer Model
Fuzzy
Controller
Rule-based
Fuzzy
Model
DOA
Automatic Monitoring and Control in the
Operating Theatre
AS/3 Monitor
Vaporizer Control
NoteBook
Graseby 3500
Aspect 1050
Monitoring EEG Signals of Awakening
During Propofol Anaesthesia
Problem: Detecting awareness
In Paralyzed patient remained unsolved.
Cardiovascular signs are not always reliable.
The aim of this study:
To investigate the changes of different waves in
EEG signals during different anaesthetic stages
To identify the parameter for early detection
awareness
Summary: (Awareness, EEG)
* The Beta, Alpha, Sub-alpha and Delta
waves have significant differences
during different stages
* The mean(SD) of the mean percentage of
beta wave in 8 patients’ EEG signals during
induction, maintenance, and recovery stages
were 73.43(4.84)%, 42.26(8.31)% and
72.14(7.19)%, respectively.
Detection of Awareness
Method I:
Questions asked during structured interview
1. What was last thing you remember before you
went to sleep for your operation ?
2. What was the first thing you remember after
your operation ?
3. Can you remember anything in between these
two periods ?
4. Did you dream during your operation ?
5. What was the worst thing about your operation ?
Method II:
Tape-recorder using earphones
(church bells, farmyard noises, light orchestral
music, piano music, market voices, bird song,
pop music and choir music).
Method III:
Forearm Technique
EEG signals in Drowsy States
Fundamental EEG Signals Research in
Drowsy States
EEG electrode positions on the scalp
Harmonie Digital EEG Systems- An Awake State
Harmonie Digital EEG Systems- A Sleep State
Brain Signal Processing and Analysis
1. Aperiodic Theory
2. 95% Spectrum Edge Frequency
3. Variety Analysis
unit：0.01 μV/
N
C( n )  
k 1
(W
n
( k 1)
 W(nk ) )  (W(nk )  W(nk 1) )
(t ) 2  N

N : sampling number
n : segment number
2
sec
C (n ) : variety
Wkn : brain signals potential
A Genetic Fuzzy System Applied in Analysis of
Brain Signals
Volunteers
Image
CCD
Experts
E
Output
R
Brain Signals
R
Signal P.
Fuzz.
Inf. Eng.
Defuzz.
CL
O
Rules
R
Decode
No
Mutation
Crossover
Reproduction
Encode
A Best Group
of Rules
Yes
Commercial Products for Mechanical, Sound,
and Electrical Stimulation for Preventing
Drowsy States
(a) Pulse massage (b) Sound wave that move your mind
(c) Nod suppresser
The Prototype of Monitoring and Controlling
Brain Signals for Deterring Driver Drowsiness
Pain Control and Evaluation using
Fuzzy Logic Control
&
Patient Controlled Analgesia
for Shock Wave Lithotripsy
Problem: Detecting pain
Subjective & no direct measurement
Clinical: Visual Analogue Scale (VAS)
Where will cause the pain:
Endoscope
Operating room:ESWL; Prostate
Post OP.: PCA
300/2000 per month
$ 4,500 , nearly 1.5 million NT
ICU
Cancer Pain
How to study the Pain:
Provide a constant pain
Not too long or too short for experiment
Easy communication with the patient
Why do we choose the ESWL to study
the Pain ?
Provide a constant pain using ultrasonic waves
Used for destroying calculi in the upper urinary
tract and gallstones (OP time < 1 hr)
Patients are consciousness
Current Drugs:
1) Pain killer: fentanyl (0.0785 mg/ml)
Preventing vomit: droperidol (2.5 mg/ml)
2) Loading dose:
Fentanyl: 2 ml; droperidol: 0.25 ml
3) NA to add further dose of fentanyl if the
patient complain.
4) If NA can not handle, call anaesthetist.
Pain Control Using PCA
PCA (Patient-Controlled Analgesia) :
Management of pain in
(1) postoperative patients
(2) cancer patients
Function:
(1) administer small bolus doses of
pain-control drugs
(2) at fixed intervals
(3) controlled by the patient with
the push of a button
PCA in ESWL
Drug: alfentanil
Conc.: 0.5 mg/ml
Loading dose: 0.5 ml
Bolus if needed: 0.4 ml
Lockout time: 1 min
Infusion rate: 120 ml/hr
Pain Control Using Fuzzy Logic Control
Fuzzy Logic Control:
– bolus + continuous infusion
– pain feedback controlled by patient
– input variable : pain, chan_pain
– output variable: chan_inf
pain: BP, SP, ZP
chan_pain: NB, NS, ZR, PS, PB
chan_inf: BI, SI, ZO, SD, BD
Patient-Controlled Analgesia
Syringe
Pump
Notebook
(Monitoring &
Control)
Pain signals
controlled by
the patient
Catheter
Patient
Clinical Evaluation of the Pain
Current Commercial Machine
(e)
100
(e)
100
90
90
80
80
70
70
Pain Intensity
Pain Intensity
Fuzzy Control Machine
60
50
40
60
50
40
30
30
20
20
10
10
0
0
0
10
20
30
40
50
60
Drug Conc. (ng/ml)
70
80
90
0
10
20
30
40
50
Drug Conc. (ng/ml)
60
70
Prostate Research