SlideShare a Scribd company logo
1 of 34
A PATIENT
POSITIONING
SYSTEM FOR
MASKLESS HEAD
AND NECK
RADIOTHERAPY
Electrical Engineering Department, UT Dallas, TX
4/23/2015
1
Presented by Olalekan Ogunmolu. April 23,
2015
Olalekan Ogunmolu: Bio Snippet
o Semester of PhD Enrolment: Fall 2014 GPA:
4.00/4.00
o Advisor: Nicholas Gans, Assistant Professor
o MSc (in Engineering) in Control Systems 2011 -
2012
The University of Sheffield, England, UK
MS Thesis: “Autonomous Navigation of a Rotorcraft Unmanned Aerial Vehicle Using
Machine Vision”, The University of Sheffield, South Yorkshire, England, Sept. 2012.
Advisor: Tony J. Dodd, Professor of Autonomous Systems Engineering, ACSE Dept.,
UoS.
o Research Thrusts
o Vision-based Control: Features tracking, Active Appearance Models, Neural
Networks
o Control Systems and Automation: Classical Control, Non-linear Systems,
Switched Systems
o Publication
o Olalekan Ogunmolu et. al, “A Real-Time Soft Robotic Patient Positioning System for
Maskless Head and Neck Cancer Radiotherapy”, IEEE Conference on Automation
and Systems Engineering [Submitted], Gothenburg, Sweden, August 2015Electrical Engineering Department, UT Dallas, TX 4/23/2015
2
Overview of Presentation
 A first-stage systematic examination of a novel method
at automating patient positioning systems during cancer
head and neck cancer radiotherapy (RT)
 Visual-servoing control using a radio-transparent soft
robot on the flexion/extension cranial motion of a patient
during mask-less radiotherapy
 Proof-of-concept validation of one degree of freedom
patient’s motion control with designed soft robot system
Electrical Engineering Department, UT Dallas, TX 4/23/2015
3
Background and Motivation
 Head and Neck (H&N) Cancers contribute nearly 3% to all cancer
developments in the United States [Jemal, A.]
 Very dangerous to benign tissues/organs nearby malign cancerous formations in
H&N region
 Treatment often involves intensity modulated radiotherapy (IMRT)
 Clinical studies have shown that small perturbations cause high
sensitivity to IMRT treatment dose [Xing., L]
 State-of-the-art robotic positioning treatments assume immobilized
patient motion on a 6D couch which is unrealistic
Electrical Engineering Department, UT Dallas, TX 4/23/2015
4
Recent Research Results
 Frameless and Maskless Cranial SRS, Cervino et al
 Anthropomorphic head phantoms employed in checking the
accuracy of a 3D surface imaging system (AlignRT System)
 Compared results from an infra-red optical tracking system with
the AlignRT software system
 For different couch angles, the difference between phantom
positions recorded by the two systems were within 1mm
displacement and 1° rotation
Electrical Engineering Department, UT Dallas, TX 4/23/2015
5
Recent Research Approaches
 Recent Research Thrusts – 6D Motion Frameless SRS
Approach, Wiersma et al
Electrical Engineering Department, UT Dallas, TX 4/23/2015
6
Typical Frame-based Radiotherapy
Electrical Engineering Department, UT Dallas, TX 4/23/2015
7
Research Overview
 Proof-of-concept study and experiment demonstrating a 1-DOF
flexion/extension control of patient cranial motion during H&N Cancer RT
 Testbed is a Mannequin head lying in a supine position on an inflatable air
bladder (IAB)
 Soft-robot consists of the IAM, two two-port SMC Pnematics Co.
proportional valves, and silicone tubes for conveying air from a pressurized
air canister
 A Kinect RGB-D camera is employed for head motion sensing and feedback
to a classical control network implemented on an NI myRIO hardware
 Work in partnership with the Drs. Xuejun Gu and Steve Jiang of the
Radiation Oncology Department of UT Southwestern, Dallas, TX, USA
Electrical Engineering Department, UT Dallas, TX 4/23/2015
8
Recent Research Results
 Average max. volunteer head motion in the head mold during the
20 minutes interval in any direction was 0.7mm (range 0.4 –
1.1mm)
 Patient motion due to couch motion was less than 0.2mm
 Drawbacks
 Minimal immobilization provided by head mold results in poor
positioning
 Relies heavily on patient cooperation to achieve immobilization
 Inter-fractional motions often ignored during pre-treatments
Electrical Engineering Department, UT Dallas, TX 4/23/2015
9
Research Aims and Objectives
 Aims
 Accurate and automatic patient positioning system (pre-
treatment)
 In-treatment automatic and accurate patient positioning with
patient drift compensation
 Objectives
 Surface-image control of the cranial flexion/extension motion of a
patient during simulated H&N RT (pre-treatment)
 Use of radio-transparent soft robot system for
positioning/manipulation tasks
Electrical Engineering Department, UT Dallas, TX 4/23/2015
10
Current System Set-up
Electrical Engineering Department, UT Dallas, TX 4/23/2015
11
Kinect RGBD Camera Sensor
Hair packed with medic neck
pillow to reduce effect of
infrared wavelengths
scattering and minimize
dropped pixels
Inflatable Air Bladder (IAB)
Mannequin
Head
NI myRIO microcontroller
Current Source (Air Flow)
Regulator Circuit
Inlet/Outlet Silicone
Tubes
Torso Ball Joint
Simulator
Inlet Proportional Flow
Control Valve
Data Processing System
24V DC Power
Supply
Vision Sensing System
 Kinect RGB-D Camera employed for position-based visual
servoing: Better depth image and alignment; Skeleton tracking
 Real-time Human Pose Recognition in Parts from Single Depth
Images. Jamie Shotton, et. al, CVPR 2011, Best Paper Award
 Real-time 3D face-tracking based on active appearance model
constrained by depth data. Nikolai Smolyanski et. al, Image and
Vision Computing, 2014, MS SDK v1.5.2
 Online Resources: OpenNI, ofxKinect
 OpenNI Process
 Generates 640 × 480 image at 30 fps with depth resolution of 40
centimeters
Electrical Engineering Department, UT Dallas, TX 4/23/2015
12
Kinect Depth Imaging
 Works with low light levels; Color and texture invariant
 Subtraction of background simplified
 Easy synthesis of realistic depth images of objects
 Cheap computational cost of building a large training dataset
 3 trees, 20 deep, 300k training images per tree, 2000 training example
pixels per image, 2000 candidate features θ, and 50 candidate
thresholds τ
Electrical Engineering Department, UT Dallas, TX 4/23/2015
13
Depth Constrained 3D Face Tracking
 MS Kinect SDK v1.5.2
 Uses depth data from the Kinect RGB-D Commodity camera to
enable 3D tracking using an active appearance method
 image difference patterns corresponding to changes in each
model parameter are learned and used to modify a model
estimate
 Resolves the monocular tracking problem by fitting an energy
term into the 2D+3D AAM fitting that minimizes a distance
between 3D face model vertices and depth data coming from a
RGBD camera.
Electrical Engineering Department, UT Dallas, TX 4/23/2015
14
Depth Constrained 3D Face Tracking
 Process Flow in Face Tracking System
 Find a face rectangle in a video frame using a face detector
 Use a neural network to find five points inside the face area – eye
centers, mouth corners, and tip of the nose
 Precompute scale of tracked face from the five points un-
projected to 3D camera space and scale 3D camera space
appropriately.
 Initialize next frame’s 2D face shape based on the
correspondences found by a robust local feature matching
between that frame and the previous frame.
 Results
 With the depth constrained 2D+3D AAM fitting, we found good
position-estimation results on a human subject when object is at a
distance of 1 to 2.5m from the Kinect System
Electrical Engineering Department, UT Dallas, TX 4/23/2015
15
Face Tracking Results
 Generalization errors and hence incorrect position estimation errors with
respect to the mannequin head due to inconsistency in depth data
 An ongoing investigation
 So we mostly relied on the OpenNI depth map centroids which we
computed for our position estimation
Electrical Engineering Department, UT Dallas, TX 4/23/2015
16
Tracking of a Human Face Tracking of a Mannequin Head
Modeling of Soft Robot System
 Modeling procedure: Overview
 Collect Data Set, 𝒁 𝑵
, of input signal, 𝑢 𝑡 , and output measurement,
𝑦(𝑡), respectively where
 𝒁 𝑵
= {𝑢 1 , 𝑦 1 , ⋯ , 𝑢 𝑁 , 𝑦 𝑁 }, 1 ≤ 𝑡 ≤ 𝑁 (1)
 Obtain a continuous-time parametric model structure similar to a one-
step ahead predictor
 𝑦 𝑛
= −𝑎 𝑛−1 𝑦 𝑛−1
− ⋯ − −𝑎0 𝑦 + 𝑏 𝑚 𝑢 𝑚
+ 𝑏 𝑚−1 𝑢 𝑚−1
+ ⋯ + 𝑏1 𝑢 + 𝑏0 𝑢 (2)
 Introduce vectors
𝜽 = 𝑎 𝑛−1, ⋯ , 𝑎0 𝑏 𝑚−1, ⋯ , 𝑏0 and 𝝓 𝒕 = −𝑦 𝑛−1
⋯ − 𝑦 − 𝑦 𝑢 𝑚
⋯ 𝑢
(3)
Since the model output depends on past data (2), we call the estimated value
𝒚 𝑡 𝜃 . Therefore, 𝒚 𝑡 𝜃 = 𝝓 𝑻
𝑡 𝜽 .
 Identification Goal: identify the best model, χ, in the set guided by the frequency
distribution analysis
 Choice of Excitation Input Signal: Sawtooth Waveform
 Integral and differential of a sawtooth waveform preserves the sawtooth
waveform with only phase and amplitude shifts
Electrical Engineering Department, UT Dallas, TX 4/23/2015
17
Modeling of Soft Robot System
 Spectrum contains both even and odd harmonics of the fundamental
frequency i.e. it contains all integer harmonics
 𝑥 𝑠𝑡ℎ 𝑡 =
𝐴
2
−
𝐴
𝜋 𝑘=1
8800 sin 2𝜋𝑘𝑓𝑡
𝑘
(4)
where A is the amplitude, A = 180mA and f is the signal frequency, 𝑓 =
10𝐻𝑧
Fig 1. Excitation Signal and Corresponding Head Position (Kinect
Measurement)
Electrical Engineering Department, UT Dallas, TX 4/23/2015
18
Data Pre-Processing
 Data Pre-processing
 We removed means and linear trends in collected data to minimize
disturbances above interest to desired system dynamics, mitigate
measurement outliers and non-continuous records in collected data
 Means removal
 𝒖 𝒂𝒗𝒆(𝒕) = 𝒖 𝑡 − 𝒖(𝑡) ; 𝒚 𝒂𝒗𝒆 𝑡 = 𝒚 𝑡 − 𝒚(𝑡); (5)
where 𝒖 𝒂𝒗𝒆(𝒕) and 𝒚 𝒂𝒗𝒆(𝒕) are the respective averages of the input and output
signals;
𝒖(𝒕) =
1
𝑁
𝑡=1
𝑛
𝑢 𝑡 𝑎𝑛𝑑 𝒚 (𝒕) =
1
𝑁
𝑡=1
𝑛
𝑦 𝑡 ;
 n is the discrete data length and N is the total data length
 Data Pre-processing
 Data detrending
 𝒖 𝒅 = 𝒖 𝒂𝒗𝒆 − 𝑨𝜽 𝒖, 𝒚 𝒅 = 𝒚 𝒂𝒗𝒆 − 𝑨𝜽 𝒚, (6)
 𝜽 𝒖 𝑎𝑛𝑑 𝜽 𝒚 are solutions to the least square fit equations
 𝑨 𝑻
𝑨 𝜽 𝑢 = 𝑨 𝑻
𝒖, 𝑨 𝑻
𝑨 𝜽 𝒚 = 𝑨 𝑇
𝒚
Electrical Engineering Department, UT Dallas, TX 4/23/2015
19
Data Pre-Processing
 and
 𝑨 𝑻
=
1 1 ⋯ 1 1
1
𝑁
1
𝑁
⋯
1
𝑁
1
𝑁
(7)
Electrical Engineering Department, UT Dallas, TX 4/23/2015
20
 Fig 2. Data Preprocessing: Means Removal
Cross-Correlation Analysis
 The cross-correlation function provides an estimate of the
system impulse response and is defined as:
 𝜓 𝑡 = 𝑡=𝜏+1
𝑁 𝑢 𝑡 −𝜏 − 𝑢 𝑦 𝑡 −𝜏 −𝑦(𝑡)
√ 𝑡=1
𝑁 𝑢 𝑡 − 𝑢
2
√ 𝑡=1
𝑁 𝑦 𝑡 −𝑦(𝑡)
2, 𝜏 = 0, ±1, Λ, ± 𝑁 − 1 (8)
 The cross-correlation function (CCF) is given by the convolution of the
system impulse response and the process auto-correlation function
(Wiener-Hopf equation)
 𝜓 𝑢𝑦(𝜏)=∫ ℎ 𝜈 𝔼 [𝑢 𝑡 𝑢(𝑡 + 𝜏 − 𝜈)d𝜈 =∫ ℎ 𝜈 𝜓 𝑢𝑢(𝜏 − 𝜈)d𝜈
 The cross correlation function between the output and test input is proportional to
the system impulse response when the input is white noise
 𝒖 𝑡 = 𝒖 𝒘 𝑡
1
𝑭(𝑧−1)
where 𝒖 𝒘 𝑡 is a zero mean white input sequence and
𝑭 𝑧−1
is an autoregressive model filter thus defined: 𝑭 𝑧−1
= 1 + 𝜆1 𝑧−1
+ Λ𝜆 𝜆 𝑧−𝜆
Electrical Engineering Department, UT Dallas, TX 4/23/2015
21
Cross-Correlation Analysis
 We estimate the parameters 𝜆𝑗 𝑗 = 1, 2, Λ, 𝜆 by fitting an AR model to 𝑢(𝑡) using a
least squares algorithm.
 The estimates of the filter, 𝑭 𝒛−𝟏
, are used to filter the input and output signals as
follows
𝑢 𝑤 𝑡 = 𝑭 𝒛−𝟏
𝑢 (𝑡)
𝑦 𝑤 𝑡 = 𝑭 𝒛−𝟏
𝑦 (𝑡)
(9)
 By estimating the filter 𝑭 𝒛−𝟏 𝑤𝑖𝑡ℎ 𝜆 = 20, the estimate of the cross-
correlation function was generated. The auto-correlation function tells of the
quality of the filter 𝑭 𝒛−𝟏 and is defined as
 𝜙11 𝜏 = 𝑡=𝜏+1
𝑁 𝒖 𝒘 𝑡 − 𝑢 𝑤 𝑡 [ 𝑢 𝑤 𝑡−𝜏 − 𝑢 𝑤]
𝑡=1
𝑁 [ 𝑢 𝑤 𝑡 − 𝑢 𝑤]2 , 𝜏 = 0, ±1, Λ, ± 𝑁 − 1
(10)
Electrical Engineering Department, UT Dallas, TX 4/23/2015
22
Correlation Analysis
Electrical Engineering Department, UT Dallas, TX 4/23/2015
23
Fig 3. ACF of detrended excitation signal
Correlation Analysis
Electrical Engineering Department, UT Dallas, TX 4/23/2015
24
Fig. 4. CCF between pre-whitened input and
output
Correlation of Residuals
 To determine the model structure of the system, we used the original
detrended data
 We chose a linear, second-order grey-box model set whose quality is
measurable by a mean-square error (MSE)
 Our choice, garnered from erstwhile analysis, ensures cost of model is not
too high in solving for 𝜽 𝑁;
 a high − order complex model is more difficult to use for simulation and
control design. If it is not marginally better than a simpler model, it may
not be worth the higher price [Llung, §16.8]
Electrical Engineering Department, UT Dallas, TX 4/23/2015
25
Sub-model Selection & Model Validation
4/23/2015Electrical Engineering Department, UT Dallas, TX
26
 A control system will perform well with an optimal linear sub-
model, tolerate disturbances and nonlinearities.
 We pick the linear frequency range: 0.00232 rad/sec to 6.85
rad/sec to represented the model of the soft robot system
 We carried out the canonical correlation to gain insight into the
desired model characteristics
 The correlation is measured between the measured head
position, 𝒚(𝑡), and estimated position by the auto-regressive
model we chose, i.e. the residuals
 𝝐 𝑡, 𝜽 𝑁 = 𝒚 𝑡 − 𝒚 𝑡 𝜽 𝑁
 The prediction errors (model error model) in the obtained model are
computed as a frequency response from the input to the residuals
Model Validation
4/23/2015Electrical Engineering Department, UT Dallas, TX
27
 Fig. 6. Correlation analysis of residuals between output, 𝒚(𝒕), and its estimate,
𝒚(𝒕)
Fig.5.Bodeplotofinputandoutput
Fig.6.Correlationfunctionofresiduals
Fig. 7. Estimates of standard variation of model from validation data set
Model Validation
4/23/2015Electrical Engineering Department, UT Dallas, TX
28
 The confidence interval compares the estimate with the estimated standard deviation
from the validation dataset
 A 99% confidence region (yellow bands) encloses the model response informing us
we have a reliable model [Llung (1999), §16.6]
 Comparing Model Structure
 Evaluation of different model structures and comparing quality of offered models
 Best fit:
 a second-order process model with delay and a RHP zero
 𝐺 𝑠 =
−0.006(𝑠 −1.7137)
(𝑠+0.01)(𝑠+0.1028)
𝑒−2𝑠
(11)
 The model has an 87.35% fit to original data with a mean square error of
0.05498𝑚𝑚2
and a final prediction error of 1.672.
 The open-loop step response of the identified transfer function is shown in Fig. 8.
Soft Robot Step Response
Electrical Engineering Department, UT Dallas, TX 4/23/2015
29
Fig. 8. Open Loop Step Response of Identified System
Control Analysis
 We see that the system is non-minimum phase with very slow transient
response.
 We require a controller that will increase the response time, guarantee
closed-loop stability whilst balancing robustness and controller
aggressiveness.
 Approximating the delay with the second-order Pade function,
𝐻 𝑠 =
𝑠2 − 3𝑠 + 3
𝑠2 + 3𝑠 + 3
 we introduce a PI controller,
𝐺𝑐 𝑠 = 3.79 +
0.0344
s
 nested within a PID controller, 𝐺 𝑝𝑖𝑑 = 3.4993 +
0.054765
𝑠
+ 55.8988𝑠, as in Fig. 9.
Electrical Engineering Department, UT Dallas, TX 4/23/2015
30
Close Loop Diagram and Step
Responses
4/23/2015Electrical Engineering Department, UT Dallas, TX
31
Fig 10. Closed Loop Step Response of Simulated
System
Fig. 9. Block Diagram of Control Network
Fig. 11. Experimental Results: Constant
Set-point
Experimental Results: Video
 Video 1. Experimental Results: Set-points Tracking
Electrical Engineering Department, UT Dallas, TX 4/23/2015
32
Conclusions and Future Work
 Deviations from desired positions during H&N Cancer RT cause dose variations
and degenerate treatment efficacy
 We have presented and demonstrated accurate control of cranial
flexion/extension motion of a patient during maskless H&N RT
 Our soft robot system can accurately track desired trajectory within 14 seconds
after start-up with the aid of a PID/PI feedforward network
 Future efforts include
 Extending results to deformable motions of the upper torso, and H&N.
 Improved bladder control: Adaptive Control, Gain Scheduling e.t.c.
 Incorporation of multi-bladders to accommodate multi-axis positioning
 Benefits
 Comprehensive and accurate control of the patient’s position
 Elimination of anatomical deformations as a result of positioning errors
Electrical Engineering Department, UT Dallas, TX 4/23/2015
33
References
 Cervino, L. I., et al. Frame-less and mask-less cranial stereotactic
radiosurgery: a feasibility study. 2010, Physics In Medicine And Biology
55(7): 1863-1873.
 Jemal A, Siegel R, Xu J, Ward E. Cancer statistics, 2010. CA: A Cancer
Journal for Clinicians2010; 60(5):277–300.
 L. Llung, System Identification Theory for the User, 2nd Edition, Upper
Saddle River, NJ, USA. Prentice Hall, 1999.
 Xing, L. Dosimetric effects of patient displacement and collimator and
gantry angle misalignment on intensity modulated radiation therapy.
Radiotherapy & Oncology, 2000. 56(1): p. 97 - 108
Electrical Engineering Department, UT Dallas, TX 4/23/2015
34

More Related Content

What's hot

(2017/06)Practical points of deep learning for medical imaging
(2017/06)Practical points of deep learning for medical imaging(2017/06)Practical points of deep learning for medical imaging
(2017/06)Practical points of deep learning for medical imagingKyuhwan Jung
 
Link prediction 방법의 개념 및 활용
Link prediction 방법의 개념 및 활용Link prediction 방법의 개념 및 활용
Link prediction 방법의 개념 및 활용Kyunghoon Kim
 
Brain Tumor Segmentation using Enhanced U-Net Model with Empirical Analysis
Brain Tumor Segmentation using Enhanced U-Net Model with Empirical AnalysisBrain Tumor Segmentation using Enhanced U-Net Model with Empirical Analysis
Brain Tumor Segmentation using Enhanced U-Net Model with Empirical AnalysisMD Abdullah Al Nasim
 
Mental Health and Machine Learning in companies
Mental Health and Machine Learning in companiesMental Health and Machine Learning in companies
Mental Health and Machine Learning in companiesRajviShah86
 
Big Data in Medicine
Big Data in MedicineBig Data in Medicine
Big Data in MedicineNasir Arafat
 
Fairness-aware Machine Learning: Practical Challenges and Lessons Learned (WS...
Fairness-aware Machine Learning: Practical Challenges and Lessons Learned (WS...Fairness-aware Machine Learning: Practical Challenges and Lessons Learned (WS...
Fairness-aware Machine Learning: Practical Challenges and Lessons Learned (WS...Krishnaram Kenthapadi
 
Topological data analysis
Topological data analysisTopological data analysis
Topological data analysisSunghyon Kyeong
 
Artificial Intelligence, Machine Learning and Deep Learning
Artificial Intelligence, Machine Learning and Deep LearningArtificial Intelligence, Machine Learning and Deep Learning
Artificial Intelligence, Machine Learning and Deep LearningSujit Pal
 
Big Data Stockholm v 7 | "Federated Machine Learning for Collaborative and Se...
Big Data Stockholm v 7 | "Federated Machine Learning for Collaborative and Se...Big Data Stockholm v 7 | "Federated Machine Learning for Collaborative and Se...
Big Data Stockholm v 7 | "Federated Machine Learning for Collaborative and Se...Dataconomy Media
 
Tools and techniques for data science
Tools and techniques for data scienceTools and techniques for data science
Tools and techniques for data scienceAjay Ohri
 
Prospects of Deep Learning in Medical Imaging
Prospects of Deep Learning in Medical ImagingProspects of Deep Learning in Medical Imaging
Prospects of Deep Learning in Medical ImagingGodswll Egegwu
 
Deep learning health care
Deep learning health care  Deep learning health care
Deep learning health care Meenakshi Sood
 
Principal component analysis and lda
Principal component analysis and ldaPrincipal component analysis and lda
Principal component analysis and ldaSuresh Pokharel
 
HEALTH PREDICTION ANALYSIS USING DATA MINING
HEALTH PREDICTION ANALYSIS USING DATA  MININGHEALTH PREDICTION ANALYSIS USING DATA  MINING
HEALTH PREDICTION ANALYSIS USING DATA MININGAshish Salve
 

What's hot (20)

(2017/06)Practical points of deep learning for medical imaging
(2017/06)Practical points of deep learning for medical imaging(2017/06)Practical points of deep learning for medical imaging
(2017/06)Practical points of deep learning for medical imaging
 
Link prediction 방법의 개념 및 활용
Link prediction 방법의 개념 및 활용Link prediction 방법의 개념 및 활용
Link prediction 방법의 개념 및 활용
 
Brain Tumor Segmentation using Enhanced U-Net Model with Empirical Analysis
Brain Tumor Segmentation using Enhanced U-Net Model with Empirical AnalysisBrain Tumor Segmentation using Enhanced U-Net Model with Empirical Analysis
Brain Tumor Segmentation using Enhanced U-Net Model with Empirical Analysis
 
1.Introduction to deep learning
1.Introduction to deep learning1.Introduction to deep learning
1.Introduction to deep learning
 
Mental Health and Machine Learning in companies
Mental Health and Machine Learning in companiesMental Health and Machine Learning in companies
Mental Health and Machine Learning in companies
 
Big Data in Medicine
Big Data in MedicineBig Data in Medicine
Big Data in Medicine
 
Medical image analysis
Medical image analysisMedical image analysis
Medical image analysis
 
Fairness-aware Machine Learning: Practical Challenges and Lessons Learned (WS...
Fairness-aware Machine Learning: Practical Challenges and Lessons Learned (WS...Fairness-aware Machine Learning: Practical Challenges and Lessons Learned (WS...
Fairness-aware Machine Learning: Practical Challenges and Lessons Learned (WS...
 
Topological data analysis
Topological data analysisTopological data analysis
Topological data analysis
 
Self-organizing map
Self-organizing mapSelf-organizing map
Self-organizing map
 
Data science Big Data
Data science Big DataData science Big Data
Data science Big Data
 
Artificial Intelligence, Machine Learning and Deep Learning
Artificial Intelligence, Machine Learning and Deep LearningArtificial Intelligence, Machine Learning and Deep Learning
Artificial Intelligence, Machine Learning and Deep Learning
 
Big Data Stockholm v 7 | "Federated Machine Learning for Collaborative and Se...
Big Data Stockholm v 7 | "Federated Machine Learning for Collaborative and Se...Big Data Stockholm v 7 | "Federated Machine Learning for Collaborative and Se...
Big Data Stockholm v 7 | "Federated Machine Learning for Collaborative and Se...
 
Algorithm based medicine
Algorithm based medicineAlgorithm based medicine
Algorithm based medicine
 
Tools and techniques for data science
Tools and techniques for data scienceTools and techniques for data science
Tools and techniques for data science
 
Prospects of Deep Learning in Medical Imaging
Prospects of Deep Learning in Medical ImagingProspects of Deep Learning in Medical Imaging
Prospects of Deep Learning in Medical Imaging
 
Robustness in deep learning
Robustness in deep learningRobustness in deep learning
Robustness in deep learning
 
Deep learning health care
Deep learning health care  Deep learning health care
Deep learning health care
 
Principal component analysis and lda
Principal component analysis and ldaPrincipal component analysis and lda
Principal component analysis and lda
 
HEALTH PREDICTION ANALYSIS USING DATA MINING
HEALTH PREDICTION ANALYSIS USING DATA  MININGHEALTH PREDICTION ANALYSIS USING DATA  MINING
HEALTH PREDICTION ANALYSIS USING DATA MINING
 

Viewers also liked

Ph.D. Qualifying Exam Presentation (McGill University, Department of Biology))
Ph.D. Qualifying Exam Presentation (McGill University, Department of Biology))Ph.D. Qualifying Exam Presentation (McGill University, Department of Biology))
Ph.D. Qualifying Exam Presentation (McGill University, Department of Biology))nouji87
 
Commissioning AlignRT with Minimal Disruption
Commissioning AlignRT with Minimal DisruptionCommissioning AlignRT with Minimal Disruption
Commissioning AlignRT with Minimal DisruptionSGRT Community
 
Qualifying exam presentation 05 20_audio
Qualifying exam presentation 05 20_audioQualifying exam presentation 05 20_audio
Qualifying exam presentation 05 20_audioebassell
 
Preliminary Examination Proposal Slides
Preliminary Examination Proposal SlidesPreliminary Examination Proposal Slides
Preliminary Examination Proposal SlidesManas Tungare
 
Fusion of Skeletal and Silhouette-based Features for Human Action Recognition...
Fusion of Skeletal and Silhouette-based Features for Human Action Recognition...Fusion of Skeletal and Silhouette-based Features for Human Action Recognition...
Fusion of Skeletal and Silhouette-based Features for Human Action Recognition...Francisco (Paco) Florez-Revuelta
 
CVPR2016 Fitting Surface Models to Data 抜粋
CVPR2016 Fitting Surface Models to Data 抜粋CVPR2016 Fitting Surface Models to Data 抜粋
CVPR2016 Fitting Surface Models to Data 抜粋sumisumith
 
Kinectic vision looking deep into depth
Kinectic vision   looking deep into depthKinectic vision   looking deep into depth
Kinectic vision looking deep into depthppd1961
 
PhD Research Proposal - Qualifying Exam
PhD Research Proposal - Qualifying ExamPhD Research Proposal - Qualifying Exam
PhD Research Proposal - Qualifying ExamParamita Mirza
 
Silhouette Rimko
Silhouette RimkoSilhouette Rimko
Silhouette RimkoRimkoc
 
Graphic design pdf
Graphic design pdfGraphic design pdf
Graphic design pdfJack Lineker
 
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 3)
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 3)SIGGRAPH 2014 Course on Computational Cameras and Displays (part 3)
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 3)Matthew O'Toole
 
Preliminary Exam Presentation
Preliminary Exam PresentationPreliminary Exam Presentation
Preliminary Exam Presentationggarner
 
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 4)
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 4)SIGGRAPH 2014 Course on Computational Cameras and Displays (part 4)
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 4)Matthew O'Toole
 
Towards fine-precision automated immobilization in maskless radiosurgery
Towards fine-precision automated immobilization in maskless radiosurgeryTowards fine-precision automated immobilization in maskless radiosurgery
Towards fine-precision automated immobilization in maskless radiosurgeryOlalekan Ogunmolu
 
Motion capture in the media
Motion capture in the mediaMotion capture in the media
Motion capture in the medialorenmazzon
 
Motion Capture Technology
Motion Capture TechnologyMotion Capture Technology
Motion Capture TechnologyGameyan Studio
 
Liu Ren at AI Frontiers: Sensor-aware Augmented Reality
Liu Ren at AI Frontiers: Sensor-aware Augmented RealityLiu Ren at AI Frontiers: Sensor-aware Augmented Reality
Liu Ren at AI Frontiers: Sensor-aware Augmented RealityAI Frontiers
 
Image enhancement ppt nal2
Image enhancement ppt nal2Image enhancement ppt nal2
Image enhancement ppt nal2Surabhi Ks
 

Viewers also liked (20)

Ph.D. Qualifying Exam Presentation (McGill University, Department of Biology))
Ph.D. Qualifying Exam Presentation (McGill University, Department of Biology))Ph.D. Qualifying Exam Presentation (McGill University, Department of Biology))
Ph.D. Qualifying Exam Presentation (McGill University, Department of Biology))
 
Commissioning AlignRT with Minimal Disruption
Commissioning AlignRT with Minimal DisruptionCommissioning AlignRT with Minimal Disruption
Commissioning AlignRT with Minimal Disruption
 
Qualifying exam presentation 05 20_audio
Qualifying exam presentation 05 20_audioQualifying exam presentation 05 20_audio
Qualifying exam presentation 05 20_audio
 
Preliminary Examination Proposal Slides
Preliminary Examination Proposal SlidesPreliminary Examination Proposal Slides
Preliminary Examination Proposal Slides
 
Fusion of Skeletal and Silhouette-based Features for Human Action Recognition...
Fusion of Skeletal and Silhouette-based Features for Human Action Recognition...Fusion of Skeletal and Silhouette-based Features for Human Action Recognition...
Fusion of Skeletal and Silhouette-based Features for Human Action Recognition...
 
CVPR2016 Fitting Surface Models to Data 抜粋
CVPR2016 Fitting Surface Models to Data 抜粋CVPR2016 Fitting Surface Models to Data 抜粋
CVPR2016 Fitting Surface Models to Data 抜粋
 
Kinectic vision looking deep into depth
Kinectic vision   looking deep into depthKinectic vision   looking deep into depth
Kinectic vision looking deep into depth
 
PhD Research Proposal - Qualifying Exam
PhD Research Proposal - Qualifying ExamPhD Research Proposal - Qualifying Exam
PhD Research Proposal - Qualifying Exam
 
Silhouette Rimko
Silhouette RimkoSilhouette Rimko
Silhouette Rimko
 
Graphic design pdf
Graphic design pdfGraphic design pdf
Graphic design pdf
 
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 3)
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 3)SIGGRAPH 2014 Course on Computational Cameras and Displays (part 3)
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 3)
 
Preliminary Exam Presentation
Preliminary Exam PresentationPreliminary Exam Presentation
Preliminary Exam Presentation
 
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 4)
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 4)SIGGRAPH 2014 Course on Computational Cameras and Displays (part 4)
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 4)
 
Towards fine-precision automated immobilization in maskless radiosurgery
Towards fine-precision automated immobilization in maskless radiosurgeryTowards fine-precision automated immobilization in maskless radiosurgery
Towards fine-precision automated immobilization in maskless radiosurgery
 
Motion capture in the media
Motion capture in the mediaMotion capture in the media
Motion capture in the media
 
Motion Capture Technology
Motion Capture TechnologyMotion Capture Technology
Motion Capture Technology
 
Motion picture - for art app class sy 2011-2012
Motion picture - for art app class sy 2011-2012Motion picture - for art app class sy 2011-2012
Motion picture - for art app class sy 2011-2012
 
Liu Ren at AI Frontiers: Sensor-aware Augmented Reality
Liu Ren at AI Frontiers: Sensor-aware Augmented RealityLiu Ren at AI Frontiers: Sensor-aware Augmented Reality
Liu Ren at AI Frontiers: Sensor-aware Augmented Reality
 
Urin Analysis
Urin AnalysisUrin Analysis
Urin Analysis
 
Image enhancement ppt nal2
Image enhancement ppt nal2Image enhancement ppt nal2
Image enhancement ppt nal2
 

Similar to PhD Qualifying Exam Slides

Hairong Qi V Swaminathan
Hairong Qi V SwaminathanHairong Qi V Swaminathan
Hairong Qi V SwaminathanFNian
 
Retinal Vessel Segmentation using Infinite Perimeter Active Contour with Hybr...
Retinal Vessel Segmentation using Infinite Perimeter Active Contour with Hybr...Retinal Vessel Segmentation using Infinite Perimeter Active Contour with Hybr...
Retinal Vessel Segmentation using Infinite Perimeter Active Contour with Hybr...IRJET Journal
 
Comparative analysis of multimodal medical image fusion using pca and wavelet...
Comparative analysis of multimodal medical image fusion using pca and wavelet...Comparative analysis of multimodal medical image fusion using pca and wavelet...
Comparative analysis of multimodal medical image fusion using pca and wavelet...IJLT EMAS
 
IEEE Bio medical engineering 2016 Title and Abstract
IEEE Bio medical engineering  2016 Title and Abstract IEEE Bio medical engineering  2016 Title and Abstract
IEEE Bio medical engineering 2016 Title and Abstract tsysglobalsolutions
 
Medical Image Processing in Nuclear Medicine and Bone Arthroplasty
Medical Image Processing in Nuclear Medicine and Bone ArthroplastyMedical Image Processing in Nuclear Medicine and Bone Arthroplasty
Medical Image Processing in Nuclear Medicine and Bone ArthroplastyIOSR Journals
 
Segmentation and Classification of MRI Brain Tumor
Segmentation and Classification of MRI Brain TumorSegmentation and Classification of MRI Brain Tumor
Segmentation and Classification of MRI Brain TumorIRJET Journal
 
International Refereed Journal of Engineering and Science (IRJES)
International Refereed Journal of Engineering and Science (IRJES)International Refereed Journal of Engineering and Science (IRJES)
International Refereed Journal of Engineering and Science (IRJES)irjes
 
A statistical path loss model
A statistical path loss modelA statistical path loss model
A statistical path loss modelNguyen Minh Thu
 
A statistical path loss model
A statistical path loss modelA statistical path loss model
A statistical path loss modelNguyen Minh Thu
 
Brain Tumor Detection and Classification using Adaptive Boosting
Brain Tumor Detection and Classification using Adaptive BoostingBrain Tumor Detection and Classification using Adaptive Boosting
Brain Tumor Detection and Classification using Adaptive BoostingIRJET Journal
 
IRJET- Texture Analysis and Fracture Identification of Bones X-Ray Images...
IRJET-  	  Texture Analysis and Fracture Identification of Bones X-Ray Images...IRJET-  	  Texture Analysis and Fracture Identification of Bones X-Ray Images...
IRJET- Texture Analysis and Fracture Identification of Bones X-Ray Images...IRJET Journal
 
IRJET - Heart Health Classification and Prediction using Machine Learning
IRJET -  	  Heart Health Classification and Prediction using Machine LearningIRJET -  	  Heart Health Classification and Prediction using Machine Learning
IRJET - Heart Health Classification and Prediction using Machine LearningIRJET Journal
 
IRJET- Image Fusion using Lifting Wavelet Transform with Neural Networks for ...
IRJET- Image Fusion using Lifting Wavelet Transform with Neural Networks for ...IRJET- Image Fusion using Lifting Wavelet Transform with Neural Networks for ...
IRJET- Image Fusion using Lifting Wavelet Transform with Neural Networks for ...IRJET Journal
 
Algorithm performance comparison for earthquake signal recognition on smartph...
Algorithm performance comparison for earthquake signal recognition on smartph...Algorithm performance comparison for earthquake signal recognition on smartph...
Algorithm performance comparison for earthquake signal recognition on smartph...TELKOMNIKA JOURNAL
 
Hybrid Transforms Based Watermarking, Encryption and Compression of Bio-Medic...
Hybrid Transforms Based Watermarking, Encryption and Compression of Bio-Medic...Hybrid Transforms Based Watermarking, Encryption and Compression of Bio-Medic...
Hybrid Transforms Based Watermarking, Encryption and Compression of Bio-Medic...IRJET Journal
 
Pulse Rate Monitoring Using Image Processing
Pulse Rate Monitoring Using Image ProcessingPulse Rate Monitoring Using Image Processing
Pulse Rate Monitoring Using Image ProcessingIRJET Journal
 
Automatic Segmentation of Brachial Artery based on Fuzzy C-Means Pixel Clust...
Automatic Segmentation of Brachial Artery based on Fuzzy  C-Means Pixel Clust...Automatic Segmentation of Brachial Artery based on Fuzzy  C-Means Pixel Clust...
Automatic Segmentation of Brachial Artery based on Fuzzy C-Means Pixel Clust...IJECEIAES
 

Similar to PhD Qualifying Exam Slides (20)

Hairong Qi V Swaminathan
Hairong Qi V SwaminathanHairong Qi V Swaminathan
Hairong Qi V Swaminathan
 
Retinal Vessel Segmentation using Infinite Perimeter Active Contour with Hybr...
Retinal Vessel Segmentation using Infinite Perimeter Active Contour with Hybr...Retinal Vessel Segmentation using Infinite Perimeter Active Contour with Hybr...
Retinal Vessel Segmentation using Infinite Perimeter Active Contour with Hybr...
 
Comparative analysis of multimodal medical image fusion using pca and wavelet...
Comparative analysis of multimodal medical image fusion using pca and wavelet...Comparative analysis of multimodal medical image fusion using pca and wavelet...
Comparative analysis of multimodal medical image fusion using pca and wavelet...
 
IEEE Bio medical engineering 2016 Title and Abstract
IEEE Bio medical engineering  2016 Title and Abstract IEEE Bio medical engineering  2016 Title and Abstract
IEEE Bio medical engineering 2016 Title and Abstract
 
160.pptx
160.pptx160.pptx
160.pptx
 
Medical Image Processing in Nuclear Medicine and Bone Arthroplasty
Medical Image Processing in Nuclear Medicine and Bone ArthroplastyMedical Image Processing in Nuclear Medicine and Bone Arthroplasty
Medical Image Processing in Nuclear Medicine and Bone Arthroplasty
 
Segmentation and Classification of MRI Brain Tumor
Segmentation and Classification of MRI Brain TumorSegmentation and Classification of MRI Brain Tumor
Segmentation and Classification of MRI Brain Tumor
 
International Refereed Journal of Engineering and Science (IRJES)
International Refereed Journal of Engineering and Science (IRJES)International Refereed Journal of Engineering and Science (IRJES)
International Refereed Journal of Engineering and Science (IRJES)
 
A statistical path loss model
A statistical path loss modelA statistical path loss model
A statistical path loss model
 
A statistical path loss model
A statistical path loss modelA statistical path loss model
A statistical path loss model
 
Madhavi tippani
Madhavi tippaniMadhavi tippani
Madhavi tippani
 
Brain Tumor Detection and Classification using Adaptive Boosting
Brain Tumor Detection and Classification using Adaptive BoostingBrain Tumor Detection and Classification using Adaptive Boosting
Brain Tumor Detection and Classification using Adaptive Boosting
 
IRJET- Texture Analysis and Fracture Identification of Bones X-Ray Images...
IRJET-  	  Texture Analysis and Fracture Identification of Bones X-Ray Images...IRJET-  	  Texture Analysis and Fracture Identification of Bones X-Ray Images...
IRJET- Texture Analysis and Fracture Identification of Bones X-Ray Images...
 
IRJET - Heart Health Classification and Prediction using Machine Learning
IRJET -  	  Heart Health Classification and Prediction using Machine LearningIRJET -  	  Heart Health Classification and Prediction using Machine Learning
IRJET - Heart Health Classification and Prediction using Machine Learning
 
IRJET- Image Fusion using Lifting Wavelet Transform with Neural Networks for ...
IRJET- Image Fusion using Lifting Wavelet Transform with Neural Networks for ...IRJET- Image Fusion using Lifting Wavelet Transform with Neural Networks for ...
IRJET- Image Fusion using Lifting Wavelet Transform with Neural Networks for ...
 
Algorithm performance comparison for earthquake signal recognition on smartph...
Algorithm performance comparison for earthquake signal recognition on smartph...Algorithm performance comparison for earthquake signal recognition on smartph...
Algorithm performance comparison for earthquake signal recognition on smartph...
 
Hybrid Transforms Based Watermarking, Encryption and Compression of Bio-Medic...
Hybrid Transforms Based Watermarking, Encryption and Compression of Bio-Medic...Hybrid Transforms Based Watermarking, Encryption and Compression of Bio-Medic...
Hybrid Transforms Based Watermarking, Encryption and Compression of Bio-Medic...
 
Pulse Rate Monitoring Using Image Processing
Pulse Rate Monitoring Using Image ProcessingPulse Rate Monitoring Using Image Processing
Pulse Rate Monitoring Using Image Processing
 
Automatic Segmentation of Brachial Artery based on Fuzzy C-Means Pixel Clust...
Automatic Segmentation of Brachial Artery based on Fuzzy  C-Means Pixel Clust...Automatic Segmentation of Brachial Artery based on Fuzzy  C-Means Pixel Clust...
Automatic Segmentation of Brachial Artery based on Fuzzy C-Means Pixel Clust...
 
BATCH 1.pptx
BATCH 1.pptxBATCH 1.pptx
BATCH 1.pptx
 

Recently uploaded

Beyond Boundaries: Leveraging No-Code Solutions for Industry Innovation
Beyond Boundaries: Leveraging No-Code Solutions for Industry InnovationBeyond Boundaries: Leveraging No-Code Solutions for Industry Innovation
Beyond Boundaries: Leveraging No-Code Solutions for Industry InnovationSafe Software
 
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...shyamraj55
 
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptxMaking_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptxnull - The Open Security Community
 
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersEnhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersThousandEyes
 
Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Mattias Andersson
 
Unlocking the Potential of the Cloud for IBM Power Systems
Unlocking the Potential of the Cloud for IBM Power SystemsUnlocking the Potential of the Cloud for IBM Power Systems
Unlocking the Potential of the Cloud for IBM Power SystemsPrecisely
 
"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
"Federated learning: out of reach no matter how close",Oleksandr Lapshyn"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
"Federated learning: out of reach no matter how close",Oleksandr LapshynFwdays
 
Bluetooth Controlled Car with Arduino.pdf
Bluetooth Controlled Car with Arduino.pdfBluetooth Controlled Car with Arduino.pdf
Bluetooth Controlled Car with Arduino.pdfngoud9212
 
Pigging Solutions Piggable Sweeping Elbows
Pigging Solutions Piggable Sweeping ElbowsPigging Solutions Piggable Sweeping Elbows
Pigging Solutions Piggable Sweeping ElbowsPigging Solutions
 
Designing IA for AI - Information Architecture Conference 2024
Designing IA for AI - Information Architecture Conference 2024Designing IA for AI - Information Architecture Conference 2024
Designing IA for AI - Information Architecture Conference 2024Enterprise Knowledge
 
AI as an Interface for Commercial Buildings
AI as an Interface for Commercial BuildingsAI as an Interface for Commercial Buildings
AI as an Interface for Commercial BuildingsMemoori
 
Science&tech:THE INFORMATION AGE STS.pdf
Science&tech:THE INFORMATION AGE STS.pdfScience&tech:THE INFORMATION AGE STS.pdf
Science&tech:THE INFORMATION AGE STS.pdfjimielynbastida
 
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmaticsKotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmaticsAndrey Dotsenko
 
Build your next Gen AI Breakthrough - April 2024
Build your next Gen AI Breakthrough - April 2024Build your next Gen AI Breakthrough - April 2024
Build your next Gen AI Breakthrough - April 2024Neo4j
 
Swan(sea) Song – personal research during my six years at Swansea ... and bey...
Swan(sea) Song – personal research during my six years at Swansea ... and bey...Swan(sea) Song – personal research during my six years at Swansea ... and bey...
Swan(sea) Song – personal research during my six years at Swansea ... and bey...Alan Dix
 
SIEMENS: RAPUNZEL – A Tale About Knowledge Graph
SIEMENS: RAPUNZEL – A Tale About Knowledge GraphSIEMENS: RAPUNZEL – A Tale About Knowledge Graph
SIEMENS: RAPUNZEL – A Tale About Knowledge GraphNeo4j
 
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024BookNet Canada
 
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmaticsKotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmaticscarlostorres15106
 

Recently uploaded (20)

Beyond Boundaries: Leveraging No-Code Solutions for Industry Innovation
Beyond Boundaries: Leveraging No-Code Solutions for Industry InnovationBeyond Boundaries: Leveraging No-Code Solutions for Industry Innovation
Beyond Boundaries: Leveraging No-Code Solutions for Industry Innovation
 
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
 
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptxMaking_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
 
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersEnhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
 
Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?
 
Unlocking the Potential of the Cloud for IBM Power Systems
Unlocking the Potential of the Cloud for IBM Power SystemsUnlocking the Potential of the Cloud for IBM Power Systems
Unlocking the Potential of the Cloud for IBM Power Systems
 
"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
"Federated learning: out of reach no matter how close",Oleksandr Lapshyn"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
 
Bluetooth Controlled Car with Arduino.pdf
Bluetooth Controlled Car with Arduino.pdfBluetooth Controlled Car with Arduino.pdf
Bluetooth Controlled Car with Arduino.pdf
 
Pigging Solutions Piggable Sweeping Elbows
Pigging Solutions Piggable Sweeping ElbowsPigging Solutions Piggable Sweeping Elbows
Pigging Solutions Piggable Sweeping Elbows
 
The transition to renewables in India.pdf
The transition to renewables in India.pdfThe transition to renewables in India.pdf
The transition to renewables in India.pdf
 
DMCC Future of Trade Web3 - Special Edition
DMCC Future of Trade Web3 - Special EditionDMCC Future of Trade Web3 - Special Edition
DMCC Future of Trade Web3 - Special Edition
 
Designing IA for AI - Information Architecture Conference 2024
Designing IA for AI - Information Architecture Conference 2024Designing IA for AI - Information Architecture Conference 2024
Designing IA for AI - Information Architecture Conference 2024
 
AI as an Interface for Commercial Buildings
AI as an Interface for Commercial BuildingsAI as an Interface for Commercial Buildings
AI as an Interface for Commercial Buildings
 
Science&tech:THE INFORMATION AGE STS.pdf
Science&tech:THE INFORMATION AGE STS.pdfScience&tech:THE INFORMATION AGE STS.pdf
Science&tech:THE INFORMATION AGE STS.pdf
 
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmaticsKotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
 
Build your next Gen AI Breakthrough - April 2024
Build your next Gen AI Breakthrough - April 2024Build your next Gen AI Breakthrough - April 2024
Build your next Gen AI Breakthrough - April 2024
 
Swan(sea) Song – personal research during my six years at Swansea ... and bey...
Swan(sea) Song – personal research during my six years at Swansea ... and bey...Swan(sea) Song – personal research during my six years at Swansea ... and bey...
Swan(sea) Song – personal research during my six years at Swansea ... and bey...
 
SIEMENS: RAPUNZEL – A Tale About Knowledge Graph
SIEMENS: RAPUNZEL – A Tale About Knowledge GraphSIEMENS: RAPUNZEL – A Tale About Knowledge Graph
SIEMENS: RAPUNZEL – A Tale About Knowledge Graph
 
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
 
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmaticsKotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
 

PhD Qualifying Exam Slides

  • 1. A PATIENT POSITIONING SYSTEM FOR MASKLESS HEAD AND NECK RADIOTHERAPY Electrical Engineering Department, UT Dallas, TX 4/23/2015 1 Presented by Olalekan Ogunmolu. April 23, 2015
  • 2. Olalekan Ogunmolu: Bio Snippet o Semester of PhD Enrolment: Fall 2014 GPA: 4.00/4.00 o Advisor: Nicholas Gans, Assistant Professor o MSc (in Engineering) in Control Systems 2011 - 2012 The University of Sheffield, England, UK MS Thesis: “Autonomous Navigation of a Rotorcraft Unmanned Aerial Vehicle Using Machine Vision”, The University of Sheffield, South Yorkshire, England, Sept. 2012. Advisor: Tony J. Dodd, Professor of Autonomous Systems Engineering, ACSE Dept., UoS. o Research Thrusts o Vision-based Control: Features tracking, Active Appearance Models, Neural Networks o Control Systems and Automation: Classical Control, Non-linear Systems, Switched Systems o Publication o Olalekan Ogunmolu et. al, “A Real-Time Soft Robotic Patient Positioning System for Maskless Head and Neck Cancer Radiotherapy”, IEEE Conference on Automation and Systems Engineering [Submitted], Gothenburg, Sweden, August 2015Electrical Engineering Department, UT Dallas, TX 4/23/2015 2
  • 3. Overview of Presentation  A first-stage systematic examination of a novel method at automating patient positioning systems during cancer head and neck cancer radiotherapy (RT)  Visual-servoing control using a radio-transparent soft robot on the flexion/extension cranial motion of a patient during mask-less radiotherapy  Proof-of-concept validation of one degree of freedom patient’s motion control with designed soft robot system Electrical Engineering Department, UT Dallas, TX 4/23/2015 3
  • 4. Background and Motivation  Head and Neck (H&N) Cancers contribute nearly 3% to all cancer developments in the United States [Jemal, A.]  Very dangerous to benign tissues/organs nearby malign cancerous formations in H&N region  Treatment often involves intensity modulated radiotherapy (IMRT)  Clinical studies have shown that small perturbations cause high sensitivity to IMRT treatment dose [Xing., L]  State-of-the-art robotic positioning treatments assume immobilized patient motion on a 6D couch which is unrealistic Electrical Engineering Department, UT Dallas, TX 4/23/2015 4
  • 5. Recent Research Results  Frameless and Maskless Cranial SRS, Cervino et al  Anthropomorphic head phantoms employed in checking the accuracy of a 3D surface imaging system (AlignRT System)  Compared results from an infra-red optical tracking system with the AlignRT software system  For different couch angles, the difference between phantom positions recorded by the two systems were within 1mm displacement and 1° rotation Electrical Engineering Department, UT Dallas, TX 4/23/2015 5
  • 6. Recent Research Approaches  Recent Research Thrusts – 6D Motion Frameless SRS Approach, Wiersma et al Electrical Engineering Department, UT Dallas, TX 4/23/2015 6
  • 7. Typical Frame-based Radiotherapy Electrical Engineering Department, UT Dallas, TX 4/23/2015 7
  • 8. Research Overview  Proof-of-concept study and experiment demonstrating a 1-DOF flexion/extension control of patient cranial motion during H&N Cancer RT  Testbed is a Mannequin head lying in a supine position on an inflatable air bladder (IAB)  Soft-robot consists of the IAM, two two-port SMC Pnematics Co. proportional valves, and silicone tubes for conveying air from a pressurized air canister  A Kinect RGB-D camera is employed for head motion sensing and feedback to a classical control network implemented on an NI myRIO hardware  Work in partnership with the Drs. Xuejun Gu and Steve Jiang of the Radiation Oncology Department of UT Southwestern, Dallas, TX, USA Electrical Engineering Department, UT Dallas, TX 4/23/2015 8
  • 9. Recent Research Results  Average max. volunteer head motion in the head mold during the 20 minutes interval in any direction was 0.7mm (range 0.4 – 1.1mm)  Patient motion due to couch motion was less than 0.2mm  Drawbacks  Minimal immobilization provided by head mold results in poor positioning  Relies heavily on patient cooperation to achieve immobilization  Inter-fractional motions often ignored during pre-treatments Electrical Engineering Department, UT Dallas, TX 4/23/2015 9
  • 10. Research Aims and Objectives  Aims  Accurate and automatic patient positioning system (pre- treatment)  In-treatment automatic and accurate patient positioning with patient drift compensation  Objectives  Surface-image control of the cranial flexion/extension motion of a patient during simulated H&N RT (pre-treatment)  Use of radio-transparent soft robot system for positioning/manipulation tasks Electrical Engineering Department, UT Dallas, TX 4/23/2015 10
  • 11. Current System Set-up Electrical Engineering Department, UT Dallas, TX 4/23/2015 11 Kinect RGBD Camera Sensor Hair packed with medic neck pillow to reduce effect of infrared wavelengths scattering and minimize dropped pixels Inflatable Air Bladder (IAB) Mannequin Head NI myRIO microcontroller Current Source (Air Flow) Regulator Circuit Inlet/Outlet Silicone Tubes Torso Ball Joint Simulator Inlet Proportional Flow Control Valve Data Processing System 24V DC Power Supply
  • 12. Vision Sensing System  Kinect RGB-D Camera employed for position-based visual servoing: Better depth image and alignment; Skeleton tracking  Real-time Human Pose Recognition in Parts from Single Depth Images. Jamie Shotton, et. al, CVPR 2011, Best Paper Award  Real-time 3D face-tracking based on active appearance model constrained by depth data. Nikolai Smolyanski et. al, Image and Vision Computing, 2014, MS SDK v1.5.2  Online Resources: OpenNI, ofxKinect  OpenNI Process  Generates 640 × 480 image at 30 fps with depth resolution of 40 centimeters Electrical Engineering Department, UT Dallas, TX 4/23/2015 12
  • 13. Kinect Depth Imaging  Works with low light levels; Color and texture invariant  Subtraction of background simplified  Easy synthesis of realistic depth images of objects  Cheap computational cost of building a large training dataset  3 trees, 20 deep, 300k training images per tree, 2000 training example pixels per image, 2000 candidate features θ, and 50 candidate thresholds τ Electrical Engineering Department, UT Dallas, TX 4/23/2015 13
  • 14. Depth Constrained 3D Face Tracking  MS Kinect SDK v1.5.2  Uses depth data from the Kinect RGB-D Commodity camera to enable 3D tracking using an active appearance method  image difference patterns corresponding to changes in each model parameter are learned and used to modify a model estimate  Resolves the monocular tracking problem by fitting an energy term into the 2D+3D AAM fitting that minimizes a distance between 3D face model vertices and depth data coming from a RGBD camera. Electrical Engineering Department, UT Dallas, TX 4/23/2015 14
  • 15. Depth Constrained 3D Face Tracking  Process Flow in Face Tracking System  Find a face rectangle in a video frame using a face detector  Use a neural network to find five points inside the face area – eye centers, mouth corners, and tip of the nose  Precompute scale of tracked face from the five points un- projected to 3D camera space and scale 3D camera space appropriately.  Initialize next frame’s 2D face shape based on the correspondences found by a robust local feature matching between that frame and the previous frame.  Results  With the depth constrained 2D+3D AAM fitting, we found good position-estimation results on a human subject when object is at a distance of 1 to 2.5m from the Kinect System Electrical Engineering Department, UT Dallas, TX 4/23/2015 15
  • 16. Face Tracking Results  Generalization errors and hence incorrect position estimation errors with respect to the mannequin head due to inconsistency in depth data  An ongoing investigation  So we mostly relied on the OpenNI depth map centroids which we computed for our position estimation Electrical Engineering Department, UT Dallas, TX 4/23/2015 16 Tracking of a Human Face Tracking of a Mannequin Head
  • 17. Modeling of Soft Robot System  Modeling procedure: Overview  Collect Data Set, 𝒁 𝑵 , of input signal, 𝑢 𝑡 , and output measurement, 𝑦(𝑡), respectively where  𝒁 𝑵 = {𝑢 1 , 𝑦 1 , ⋯ , 𝑢 𝑁 , 𝑦 𝑁 }, 1 ≤ 𝑡 ≤ 𝑁 (1)  Obtain a continuous-time parametric model structure similar to a one- step ahead predictor  𝑦 𝑛 = −𝑎 𝑛−1 𝑦 𝑛−1 − ⋯ − −𝑎0 𝑦 + 𝑏 𝑚 𝑢 𝑚 + 𝑏 𝑚−1 𝑢 𝑚−1 + ⋯ + 𝑏1 𝑢 + 𝑏0 𝑢 (2)  Introduce vectors 𝜽 = 𝑎 𝑛−1, ⋯ , 𝑎0 𝑏 𝑚−1, ⋯ , 𝑏0 and 𝝓 𝒕 = −𝑦 𝑛−1 ⋯ − 𝑦 − 𝑦 𝑢 𝑚 ⋯ 𝑢 (3) Since the model output depends on past data (2), we call the estimated value 𝒚 𝑡 𝜃 . Therefore, 𝒚 𝑡 𝜃 = 𝝓 𝑻 𝑡 𝜽 .  Identification Goal: identify the best model, χ, in the set guided by the frequency distribution analysis  Choice of Excitation Input Signal: Sawtooth Waveform  Integral and differential of a sawtooth waveform preserves the sawtooth waveform with only phase and amplitude shifts Electrical Engineering Department, UT Dallas, TX 4/23/2015 17
  • 18. Modeling of Soft Robot System  Spectrum contains both even and odd harmonics of the fundamental frequency i.e. it contains all integer harmonics  𝑥 𝑠𝑡ℎ 𝑡 = 𝐴 2 − 𝐴 𝜋 𝑘=1 8800 sin 2𝜋𝑘𝑓𝑡 𝑘 (4) where A is the amplitude, A = 180mA and f is the signal frequency, 𝑓 = 10𝐻𝑧 Fig 1. Excitation Signal and Corresponding Head Position (Kinect Measurement) Electrical Engineering Department, UT Dallas, TX 4/23/2015 18
  • 19. Data Pre-Processing  Data Pre-processing  We removed means and linear trends in collected data to minimize disturbances above interest to desired system dynamics, mitigate measurement outliers and non-continuous records in collected data  Means removal  𝒖 𝒂𝒗𝒆(𝒕) = 𝒖 𝑡 − 𝒖(𝑡) ; 𝒚 𝒂𝒗𝒆 𝑡 = 𝒚 𝑡 − 𝒚(𝑡); (5) where 𝒖 𝒂𝒗𝒆(𝒕) and 𝒚 𝒂𝒗𝒆(𝒕) are the respective averages of the input and output signals; 𝒖(𝒕) = 1 𝑁 𝑡=1 𝑛 𝑢 𝑡 𝑎𝑛𝑑 𝒚 (𝒕) = 1 𝑁 𝑡=1 𝑛 𝑦 𝑡 ;  n is the discrete data length and N is the total data length  Data Pre-processing  Data detrending  𝒖 𝒅 = 𝒖 𝒂𝒗𝒆 − 𝑨𝜽 𝒖, 𝒚 𝒅 = 𝒚 𝒂𝒗𝒆 − 𝑨𝜽 𝒚, (6)  𝜽 𝒖 𝑎𝑛𝑑 𝜽 𝒚 are solutions to the least square fit equations  𝑨 𝑻 𝑨 𝜽 𝑢 = 𝑨 𝑻 𝒖, 𝑨 𝑻 𝑨 𝜽 𝒚 = 𝑨 𝑇 𝒚 Electrical Engineering Department, UT Dallas, TX 4/23/2015 19
  • 20. Data Pre-Processing  and  𝑨 𝑻 = 1 1 ⋯ 1 1 1 𝑁 1 𝑁 ⋯ 1 𝑁 1 𝑁 (7) Electrical Engineering Department, UT Dallas, TX 4/23/2015 20  Fig 2. Data Preprocessing: Means Removal
  • 21. Cross-Correlation Analysis  The cross-correlation function provides an estimate of the system impulse response and is defined as:  𝜓 𝑡 = 𝑡=𝜏+1 𝑁 𝑢 𝑡 −𝜏 − 𝑢 𝑦 𝑡 −𝜏 −𝑦(𝑡) √ 𝑡=1 𝑁 𝑢 𝑡 − 𝑢 2 √ 𝑡=1 𝑁 𝑦 𝑡 −𝑦(𝑡) 2, 𝜏 = 0, ±1, Λ, ± 𝑁 − 1 (8)  The cross-correlation function (CCF) is given by the convolution of the system impulse response and the process auto-correlation function (Wiener-Hopf equation)  𝜓 𝑢𝑦(𝜏)=∫ ℎ 𝜈 𝔼 [𝑢 𝑡 𝑢(𝑡 + 𝜏 − 𝜈)d𝜈 =∫ ℎ 𝜈 𝜓 𝑢𝑢(𝜏 − 𝜈)d𝜈  The cross correlation function between the output and test input is proportional to the system impulse response when the input is white noise  𝒖 𝑡 = 𝒖 𝒘 𝑡 1 𝑭(𝑧−1) where 𝒖 𝒘 𝑡 is a zero mean white input sequence and 𝑭 𝑧−1 is an autoregressive model filter thus defined: 𝑭 𝑧−1 = 1 + 𝜆1 𝑧−1 + Λ𝜆 𝜆 𝑧−𝜆 Electrical Engineering Department, UT Dallas, TX 4/23/2015 21
  • 22. Cross-Correlation Analysis  We estimate the parameters 𝜆𝑗 𝑗 = 1, 2, Λ, 𝜆 by fitting an AR model to 𝑢(𝑡) using a least squares algorithm.  The estimates of the filter, 𝑭 𝒛−𝟏 , are used to filter the input and output signals as follows 𝑢 𝑤 𝑡 = 𝑭 𝒛−𝟏 𝑢 (𝑡) 𝑦 𝑤 𝑡 = 𝑭 𝒛−𝟏 𝑦 (𝑡) (9)  By estimating the filter 𝑭 𝒛−𝟏 𝑤𝑖𝑡ℎ 𝜆 = 20, the estimate of the cross- correlation function was generated. The auto-correlation function tells of the quality of the filter 𝑭 𝒛−𝟏 and is defined as  𝜙11 𝜏 = 𝑡=𝜏+1 𝑁 𝒖 𝒘 𝑡 − 𝑢 𝑤 𝑡 [ 𝑢 𝑤 𝑡−𝜏 − 𝑢 𝑤] 𝑡=1 𝑁 [ 𝑢 𝑤 𝑡 − 𝑢 𝑤]2 , 𝜏 = 0, ±1, Λ, ± 𝑁 − 1 (10) Electrical Engineering Department, UT Dallas, TX 4/23/2015 22
  • 23. Correlation Analysis Electrical Engineering Department, UT Dallas, TX 4/23/2015 23 Fig 3. ACF of detrended excitation signal
  • 24. Correlation Analysis Electrical Engineering Department, UT Dallas, TX 4/23/2015 24 Fig. 4. CCF between pre-whitened input and output
  • 25. Correlation of Residuals  To determine the model structure of the system, we used the original detrended data  We chose a linear, second-order grey-box model set whose quality is measurable by a mean-square error (MSE)  Our choice, garnered from erstwhile analysis, ensures cost of model is not too high in solving for 𝜽 𝑁;  a high − order complex model is more difficult to use for simulation and control design. If it is not marginally better than a simpler model, it may not be worth the higher price [Llung, §16.8] Electrical Engineering Department, UT Dallas, TX 4/23/2015 25
  • 26. Sub-model Selection & Model Validation 4/23/2015Electrical Engineering Department, UT Dallas, TX 26  A control system will perform well with an optimal linear sub- model, tolerate disturbances and nonlinearities.  We pick the linear frequency range: 0.00232 rad/sec to 6.85 rad/sec to represented the model of the soft robot system  We carried out the canonical correlation to gain insight into the desired model characteristics  The correlation is measured between the measured head position, 𝒚(𝑡), and estimated position by the auto-regressive model we chose, i.e. the residuals  𝝐 𝑡, 𝜽 𝑁 = 𝒚 𝑡 − 𝒚 𝑡 𝜽 𝑁  The prediction errors (model error model) in the obtained model are computed as a frequency response from the input to the residuals
  • 27. Model Validation 4/23/2015Electrical Engineering Department, UT Dallas, TX 27  Fig. 6. Correlation analysis of residuals between output, 𝒚(𝒕), and its estimate, 𝒚(𝒕) Fig.5.Bodeplotofinputandoutput Fig.6.Correlationfunctionofresiduals Fig. 7. Estimates of standard variation of model from validation data set
  • 28. Model Validation 4/23/2015Electrical Engineering Department, UT Dallas, TX 28  The confidence interval compares the estimate with the estimated standard deviation from the validation dataset  A 99% confidence region (yellow bands) encloses the model response informing us we have a reliable model [Llung (1999), §16.6]  Comparing Model Structure  Evaluation of different model structures and comparing quality of offered models  Best fit:  a second-order process model with delay and a RHP zero  𝐺 𝑠 = −0.006(𝑠 −1.7137) (𝑠+0.01)(𝑠+0.1028) 𝑒−2𝑠 (11)  The model has an 87.35% fit to original data with a mean square error of 0.05498𝑚𝑚2 and a final prediction error of 1.672.  The open-loop step response of the identified transfer function is shown in Fig. 8.
  • 29. Soft Robot Step Response Electrical Engineering Department, UT Dallas, TX 4/23/2015 29 Fig. 8. Open Loop Step Response of Identified System
  • 30. Control Analysis  We see that the system is non-minimum phase with very slow transient response.  We require a controller that will increase the response time, guarantee closed-loop stability whilst balancing robustness and controller aggressiveness.  Approximating the delay with the second-order Pade function, 𝐻 𝑠 = 𝑠2 − 3𝑠 + 3 𝑠2 + 3𝑠 + 3  we introduce a PI controller, 𝐺𝑐 𝑠 = 3.79 + 0.0344 s  nested within a PID controller, 𝐺 𝑝𝑖𝑑 = 3.4993 + 0.054765 𝑠 + 55.8988𝑠, as in Fig. 9. Electrical Engineering Department, UT Dallas, TX 4/23/2015 30
  • 31. Close Loop Diagram and Step Responses 4/23/2015Electrical Engineering Department, UT Dallas, TX 31 Fig 10. Closed Loop Step Response of Simulated System Fig. 9. Block Diagram of Control Network Fig. 11. Experimental Results: Constant Set-point
  • 32. Experimental Results: Video  Video 1. Experimental Results: Set-points Tracking Electrical Engineering Department, UT Dallas, TX 4/23/2015 32
  • 33. Conclusions and Future Work  Deviations from desired positions during H&N Cancer RT cause dose variations and degenerate treatment efficacy  We have presented and demonstrated accurate control of cranial flexion/extension motion of a patient during maskless H&N RT  Our soft robot system can accurately track desired trajectory within 14 seconds after start-up with the aid of a PID/PI feedforward network  Future efforts include  Extending results to deformable motions of the upper torso, and H&N.  Improved bladder control: Adaptive Control, Gain Scheduling e.t.c.  Incorporation of multi-bladders to accommodate multi-axis positioning  Benefits  Comprehensive and accurate control of the patient’s position  Elimination of anatomical deformations as a result of positioning errors Electrical Engineering Department, UT Dallas, TX 4/23/2015 33
  • 34. References  Cervino, L. I., et al. Frame-less and mask-less cranial stereotactic radiosurgery: a feasibility study. 2010, Physics In Medicine And Biology 55(7): 1863-1873.  Jemal A, Siegel R, Xu J, Ward E. Cancer statistics, 2010. CA: A Cancer Journal for Clinicians2010; 60(5):277–300.  L. Llung, System Identification Theory for the User, 2nd Edition, Upper Saddle River, NJ, USA. Prentice Hall, 1999.  Xing, L. Dosimetric effects of patient displacement and collimator and gantry angle misalignment on intensity modulated radiation therapy. Radiotherapy & Oncology, 2000. 56(1): p. 97 - 108 Electrical Engineering Department, UT Dallas, TX 4/23/2015 34

Editor's Notes

  1. ~the head ring locks the patient’s head into a precise 3D position suppressing both voluntary and involuntary physiological motions ~head ring may not necessarily fit into patients’ head with prior intracranial surgery or other anatomical inhibitions
  2. A 3mm deviation of couch position in anterior-posterior position resulted in 38% decrease in minimum target dose or 41% in spinal cord dose [2]
  3. Pixels in a depth image indicate calibrated depth in the scene rather than a measure of intensity or color. Depth cameras offer several advantages over traditional intensity sensors, working in low light levels, giving a calibrated scale estimate, color and texture invariant, resolving silhouettes ambiguities in pose, simplifying background subtraction; straightforward to synthesize realistic depth images of people and cheap computational cost of building a large training dataset Set Parameters in real-time human-body parts tracking: 3 trees, 20 deep, 300k training images per tree, 2000 training example pixels per image, 2000 candidate features θ, and 50 candidate thresholds τ
  4. Face tracking algorithms belong in two broad categories: the first class consists of feature-based tracking algorithms which track local points from frame to frame to compute head pose and facial expressions based on locations of the points. Global regularization constraints are used to guarantee facial alignments. Local features matching make these algorithms less prone to generalization, illumination and occlusion problems. Cons: errors in features tracking lead to jittery and inaccurate results. Second method involves algorithms that use appearance-based generative face models such as active appearance methods and 3D morphable models. Generally more accurate and produces better global alignment since they compute over the input data space. But they do have the difficulty of generalizing to unseen faces and may suffer from illumination changes and occlusions. Both classes use a generative linear 3D facial alignment by either fitting a projected 3D model to video camera input or combine projected model fitting with 3D fitting to depth camera input. 2D + 3D AAM, introduced by Zhou et al, uses a projected linear 3D tracking parameters (head pose, expressions) but still suffers from illuminations, occlusions and generalization. AAM: image difference patterns corresponding to changes in each model parameter are learned and used to modify a model estimate. AAMs use a training set of examples; AAMs learn what are valid shape, and intensity variations from their training set. Cootes et al Paper Generate models based on a model of shape variation and a model of the appearance variations in a shape-normalized frame. A training set of labelled images with points where key landmark points are marked on each example object. Smolyanski’s Paper Initializes 3D face model by computing realistic face shapes from a set of input RGBD frames. This improves tracking accuracy since the 3D model and its projection are closer to the input RGBD data. Uses a linear 3D morphable model as a face shape model. It is computed by applying pca to a set of 3D human faces collected with the help of a high-resolution stereo color rig. At first, the face tracker uses an average face as the 3D face model. Once enough tracking data are collected, a personalized face model is computed. Uses 2D facial landmarks with corresponding depth frames for face shape computation. The shape modeling algorithm deforms the 3D model so that to all the collected data. After the computation of the personalized 3D face shape model, the face tracker accuracy is improved by using the 3D face model in the tracking runtime. Resolves the monocular tracking problem by fitting a depth energy term into the 2D+3D AAM fitting that minimizes the Euclidean distance between 3D face model vertices and depth data coming from a RGBD camera. The term is formulated similar to the energy function used in Iterative Closest Point (ICP) algorithm. This addition reduces the solution space and finds better 3D alignments. The extended energy function provides better accuracy in 3D space which is generally the issue with other 2D+3D AAM. Captures 500 neural faces (no expressions) in 3D with the stereo capture rig and annotate them. Used PCA to build a statistical linear 3D face shape model. A 3D artist created a realistic set of facial expression deformations for this model. The 2D AAM model was trained from face images with various expressions. The AAM was trained by using PCA for a set of annotated 2D images. The training process was less expensive than full 3D AAM training for real-world applications that require tracking many facial types. 3D model has explicit face shape and facial animation parts to simplify creation of avatar animation systems. The 3D mean face topology was created by a 3D artist and includes 1347 vertices. Mean face vertices and shape basis were computed by the PCA process; the shape basis consists of 110 shape deformations. The animation basis consists of 17 animations. The resulting animation basis is transformed to form an orthonormal basis together with the shape basis.
  5. In the event of the unavailability of depth data, user’s head is assumed to match the mean face size. Also align a 2D AAM shape to the five feature points.
  6. In the event of the unavailability of depth data, user’s head is assumed to match the mean face size. Also align a 2D AAM shape to the five feature points.
  7. With the regulated air compressor providing a standard pressure of 30 psi, we applied a persistently exciting input current in the form of a sawtooth waveform to deliver operational currents to the coil in the inlet PVQ valve. Airflow out of the outlet valve was kept constant by opening it to it mid-position thus varying the head pose through an open-loop inflation and deflation process. of the IAB.
  8. With the regulated air compressor providing a standard pressure of 30 psi, we applied a persistently exciting input current in the form of a sawtooth waveform to deliver operational currents to the coil in the inlet PVQ valve. Airflow out of the outlet valve was kept constant by opening it to it mid-position thus varying the head pose through an open-loop inflation and deflation process. of the IAB.
  9. For the identification of linear systems, there are three basic facts that govern the choices: The asymptotic properties of the estimate (bias and variance) depend only on the input spectrum – not the actual waveform of the input 2. The input must have limited amplitude: u \leq u(t) \leq \bar{u} 3 Periodic inputs mat=y have certain advantages The covariance matrix is typically inversely proportional to the input power. We want this to have as much input power as possible. In practice, the actual input limitation concerns amplitude constraints 𝑢 and 𝑢 . A good signal waveform is one that has a small crest factor. The lower bound =d for C_r is 1 which is achieved for binary, symmetric signals: u(t) = \pm \bar(u). A binary input will not however have validation against non-linearity. To achieve a large information matrix, we should spend the input power at frequencies where M(omega) is large, i.e., where the Bode plot is sensitive to parameter variations. Put differently, if a parameter is of special interest, then vary it and check where the Bode plot moves, and put the input power there. In many cases, this may give sufficient guidance for good input design.
  10. For the identification of linear systems, there are three basic facts that govern the choices: The asymptotic properties of the estimate (bias and variance) depend only on the input spectrum – not the actual waveform of the input 2. The input must have limited amplitude: u \leq u(t) \leq \bar{u} 3 Periodic inputs mat=y have certain advantages The covariance matrix is typically inversely proportional to the input power. We want this to have as much input power as possible. In practice, the actual input limitation concerns amplitude constraints 𝑢 and 𝑢 . A good signal waveform is one that has a small crest factor. The lower bound =d for C_r is 1 which is achieved for binary, symmetric signals: u(t) = \pm \bar(u). A binary input will not however have validation against non-linearity's. To achieve a large information matrix, we should spend the input power at frequencies where M(omega) is large, i.e., where the Bode plot is sensitive to parameter variations. Put differently, if a parameter is of special interest, then vary it and check where the Bode plot moves, and put the input power there. In many cases, this may give sufficient guidance for good inout design. Matlab, “detrend” removes the best straight-line fit from the vector x and returns in y. If x uis a matrix, detrend removes the trend from each column
  11. Cross-correlation is similar to convolution in that it it a nofolding (time reversal), In Matlab: Conv(x,fliplr(y))
  12. The correlation time musdt be finite. Second, the input signal must have a finite bandwidth and retain an autocorrelation function that is approximately a delta function. It should have the ff definitions: It is binary and assumes values of plus or minus one with equal probability It may (but not necessarily) change sign only every \delta t seconds It is periodic with period T = N\deltat, N>>1 The autocorrelation function will look like a delta function in every peiod T. But will have side lobes because of the finite signal length. The side lobe binary amplitude distribution is the binary probability distribution , and for large N is nearly normal with \sigma = 1/sqrt(N). The side lobes introduce very large errors in the cross correlation result except N is very large. The autocorrelation function is always symmetric with respect to its center, even for assymetric pulses Auto-correlation function used to differentiate the presence of a like-signal e.g. a zero or one it’s the correlation of a signal with itself The maximum value of the auto-correlartion function occurs at zero lag, i.e. when the signals are perfectly matched The transfer function of the system can be obtained by taking the Laplace transform of the impulse response function Use of autocorrelation to detect the presence of a periodic signal corrupted by noise
  13. The correlation time musdt be finite. Second, the input signal must have a finite bandwidth and retain an autocorrelation function that is approximately a delta function. It should have the ff definitions: It is binary and assumes values of plus or minus one with equal probability It may (but not necessarily) change sign only every \delta t seconds It is periodic with period T = N\deltat, N>>1 The autocorrelation function will look like a delta function in every peiod T. But will have side lobes because of the finite signal length. The side lobe binary amplitude distribution is the binary probability distribution , and for large N is nearly normal with \sigma = 1/sqrt(N). The side lobes introduce very large errors in the cross correlation result except N is very large.
  14. MSE – 12.26, Lennart Llung Akaike’s Final Prediction Error – Provides a measure of model quality by simulating the situation where the model is tested on a different data set. After computing several different models, one can compare them using this criterion. The most accurate model will hjave the lowest FPE. Akaike's Final Prediction Error (FPE) is defined by the following equation: FPE = 𝑉 ( 1+ 𝑑 𝑁 1− 𝑑 𝑁 ) where V is the loss function, d is the number of estimated parameters and N is the number of values in the estimation data set. where V is the loss function, d is the number of estimated parameters, and N is the number of values in the estimation data set. The Matlab toolbox assumes that the final prediction error is asymptotic for d<<N and uses the following approximation to compute FPE: FPE = 𝑉 (1+ 2𝑑 𝑁 ) where the loss function V is defined by V = det ├ 1 𝑁 𝑚𝑖𝑛 𝑚𝑎𝑥 𝜖 𝑡, 𝜃 𝑁 𝜖 𝑡, 𝜃 𝑁 𝑇