- •Contents
- •1.1. Introduction to the Eye
- •1.2. The Anatomy of the Human Visual System
- •1.3. Neurons
- •1.4. Synapses
- •1.5. Vision — Sensory Transduction
- •1.6. Retinal Processing
- •1.7. Visual Processing in the Brain
- •1.8. Biological Vision and Computer Vision Algorithms
- •References
- •2.1. Introduction to Computational Methods for Feature Detection
- •2.2. Preprocessing Methods for Retinal Images
- •2.2.1. Illumination Effect Reduction
- •2.2.1.1. Non-linear brightness transform
- •2.2.2. Image Normalization and Enhancement
- •2.2.2.1. Color channel transformations
- •2.2.2.3. Local adaptive contrast enhancement
- •2.2.2.4. Histogram transformations
- •2.3. Segmentation Methods for Retinal Anatomy Detection and Localization
- •2.3.1. A Boundary Detection Methods
- •2.3.1.1. First-order difference operators
- •2.3.1.2. Second-order boundary detection
- •2.3.1.3. Canny edge detection
- •2.3.2. Edge Linkage Methods for Boundary Detection
- •2.3.2.1. Local neighborhood gradient thresholding
- •2.3.2.2. Morphological operations for edge link enhancement
- •2.3.2.3. Hough transform for edge linking
- •2.3.3. Thresholding for Image Segmentation
- •2.3.3.1. Segmentation with a single threshold
- •2.3.3.2. Multi-level thresholding
- •2.3.3.3. Windowed thresholding
- •2.3.4. Region-Based Methods for Image Segmentation
- •2.3.4.1. Region growing
- •2.3.4.2. Watershed segmentation
- •2.4.1. Statistical Features
- •2.4.1.1. Geometric descriptors
- •2.4.1.2. Texture features
- •2.4.1.3. Invariant moments
- •2.4.2. Data Transformations
- •2.4.2.1. Fourier descriptors
- •2.4.2.2. Principal component analysis (PCA)
- •2.4.3. Multiscale Features
- •2.4.3.1. Wavelet transform
- •2.4.3.2. Scale-space methods for feature extraction
- •2.5. Summary
- •References
- •3.1.1. EBM Process
- •3.1.2. Evidence-Based Medical Issues
- •3.1.3. Value-Based Evidence
- •3.2.1. Economic Evaluation
- •3.2.2. Decision Analysis Method
- •3.2.3. Advantages of Decision Analysis
- •3.2.4. Perspective in Decision Analysis
- •3.2.5. Decision Tree in Decision Analysis
- •3.3. Use of Information Technologies for Diagnosis in Ophthalmology
- •3.3.1. Data Mining in Ophthalmology
- •3.3.2. Graphical User Interface
- •3.4. Role of Computational System in Curing Disease of an Eye
- •3.4.1. Computational Decision Support System: Diabetic Retinopathy
- •3.4.1.1. Wavelet-based neural network23
- •3.4.1.2. Content-based image retrieval
- •3.4.2. Computational Decision Support System: Cataracts
- •3.4.2.2. K nearest neighbors
- •3.4.2.3. GUI of the system
- •3.4.3. Computational Decision Support System: Glaucoma
- •3.4.3.1. Using fuzzy logic
- •3.4.4. Computational Decision Support System: Blepharitis, Rosacea, Sjögren, and Dry Eyes
- •3.4.4.1. Utility of bleb imaging with anterior segment OCT in clinical decision making
- •3.4.4.2. Computational decision support system: RD
- •3.4.4.3. Role of computational system
- •3.4.5. Computational Decision Support System: Amblyopia
- •3.4.5.1. Role of computational decision support system in amblyopia
- •3.5. Conclusion
- •References
- •4.1. Introduction to Oxygen in the Retina
- •4.1.1. Microelectrode Methods
- •4.1.2. Phosphorescence Dye Method
- •4.1.3. Spectrographic Method
- •4.1.6. HSI Method
- •4.2. Experiment One
- •4.2.1. Methods and Materials
- •4.2.1.1. Animals
- •4.2.1.2. Systemic oxygen saturation
- •4.2.1.3. Intraocular pressure
- •4.2.1.4. Fundus camera
- •4.2.1.5. Hyperspectral imaging
- •4.2.1.6. Extraction of spectral curves
- •4.2.1.7. Mapping relative oxygen saturation
- •4.2.1.8. Relative saturation indices (RSIs)
- •4.2.2. Results
- •4.2.2.1. Spectral signatures
- •4.2.2.2. Oxygen breathing
- •4.2.2.3. Intraocular pressure
- •4.2.2.4. Responses to oxygen breathing
- •4.2.2.5. Responses to high IOP
- •4.2.3. Discussion
- •4.2.3.1. Pure oxygen breathing experiment
- •4.2.3.2. IOP perturbation experiment
- •4.2.3.3. Hyperspectral imaging
- •4.3. Experiment Two
- •4.3.1. Methods and Materials
- •4.3.1.1. Animals, anesthesia, blood pressure, and IOP perturbation
- •4.3.1.3. Spectral determinant of percentage oxygen saturation
- •4.3.1.5. Preparation and calibration of red blood cell suspensions
- •4.3.2. Results
- •4.3.2.2. Oxygen saturation of the ONH
- •4.3.3. Discussion
- •4.3.4. Conclusions
- •4.4. Experiment Three
- •4.4.1. Methods and Materials
- •4.4.1.1. Compliance testing
- •4.4.1.2. Hyperspectral imaging
- •4.4.1.3. Selection of ONH structures
- •4.4.1.4. Statistical methods
- •4.4.2. Results
- •4.4.2.1. Compliance testing
- •4.4.2.2. Blood spectra from ONH structures
- •4.4.2.3. Oxygen saturation of ONH structures
- •4.4.2.4. Oxygen saturation maps
- •4.4.3. Discussion
- •4.5. Experiment Four
- •4.5.1. Methods and Materials
- •4.5.2. Results
- •4.5.3. Discussion
- •4.6. Experiment Five
- •4.6.1. Methods and Materials
- •4.6.1.3. Automatic control point detection
- •4.6.1.4. Fused image optimization
- •4.7. Conclusion
- •References
- •5.1. Introduction to Thermography
- •5.2. Data Acquisition
- •5.3. Methods
- •5.3.1. Snake and GVF
- •5.3.2. Target Tracing Function and Genetic Algorithm
- •5.3.3. Locating Cornea
- •5.4. Results
- •5.5. Discussion
- •5.6. Conclusion
- •References
- •6.1. Introduction to Glaucoma
- •6.1.1. Glaucoma Types
- •6.1.1.1. Primary open-angle glaucoma
- •6.1.1.2. Angle-closure glaucoma
- •6.1.2. Diagnosis of Glaucoma
- •6.2. Materials and Methods
- •6.2.1. c/d Ratio
- •6.2.2. Measuring the Area of Blood Vessels
- •6.2.3. Measuring the ISNT Ratio
- •6.3. Results
- •6.4. Discussion
- •6.5. Conclusion
- •References
- •7.1. Introduction to Temperature Distribution
- •7.3. Mathematical Model
- •7.3.1. The Human Eye
- •7.3.2. The Eye Tumor
- •7.3.3. Governing Equations
- •7.3.4. Boundary Conditions
- •7.4. Material Properties
- •7.5. Numerical Scheme
- •7.5.1. Integro-Differential Equations
- •7.6. Results
- •7.6.1. Numerical Model
- •7.6.2. Case 1
- •7.6.3. Case 2
- •7.6.4. Discussion
- •7.7. Parametric Optimization
- •7.7.1. Analysis of Variance
- •7.7.2. Taguchi Method
- •7.7.3. Discussion
- •7.8. Concluding Remarks
- •References
- •8.1. Introduction to IR Thermography
- •8.2. Infrared Thermography and the Measured OST
- •8.3. The Acquisition of OST
- •8.3.1. Manual Measures
- •8.3.2. Semi-Automated and Fully Automated
- •8.4. Applications to Ocular Studies
- •8.4.1. On Ocular Physiologies
- •8.4.2. On Ocular Diseases and Surgery
- •8.5. Discussion
- •References
- •9.1. Introduction
- •9.1.1. Preprocessing
- •9.1.1.1. Shade correction
- •9.1.1.2. Hough transform
- •9.1.1.3. Top-hat transform
- •9.1.2. Image Segmentation
- •9.1.2.1. The region approach
- •9.1.2.2. The gradient-based method
- •9.1.2.3. Edge detection
- •9.1.2.3.2. The second-order derivative methods
- •9.1.2.3.3. The optimal edge detector
- •9.2. Image Registration
- •9.4. Automated, Integrated Image Analysis Systems
- •9.5. Conclusion
- •References
- •10.1. Introduction to Diabetic Retinopathy
- •10.2. Data Acquisition
- •10.3. Feature Extraction
- •10.3.1. Blood Vessel Detection
- •10.3.2. Exudates Detection
- •10.3.3. Hemorrhages Detection
- •10.3.4. Contrast
- •10.4.1. Backpropagation Algorithm
- •10.5. Results
- •10.6. Discussion
- •10.7. Conclusion
- •References
- •11.1. Related Studies
- •11.2.1. Encryption
- •11.3. Compression Technique
- •11.3.1. Huffman Coding
- •11.4. Error Control Coding
- •11.4.1. Hamming Codes
- •11.4.2. BCH Codes
- •11.4.3. Convolutional Codes
- •11.4.4. RS Codes14
- •11.4.5. Turbo Codes14
- •11.5. Results
- •11.5.1. Using Turbo Codes for Transmission of Retinal Fundus Image
- •11.6. Discussion
- •11.7. Conclusion
- •References
- •12.1. Introduction to Laser-Thermokeratoplasty (LTKP)
- •12.2. Characteristics of LTKP
- •12.3. Pulsed Laser
- •12.4. Continuous-Wave Laser
- •12.5. Mathematical Model
- •12.5.1. Model Description
- •12.5.2. Governing Equations
- •12.5.3. Initial-Boundary Conditions
- •12.6. Numerical Scheme
- •12.6.1. Integro-Differential Equation
- •12.7. Results
- •12.7.1. Pulsed Laser
- •12.7.2. Continuous-Wave Laser
- •12.7.3. Thermal Damage Assessment
- •12.8. Discussion
- •12.9. Concluding Remarks
- •References
- •13.1. Introduction to Optical Eye Modeling
- •13.1.1. Ocular Measurements for Optical Eye Modeling
- •13.1.1.1. Curvature, dimension, thickness, or distance parameters of ocular elements
- •13.1.1.2. Three-dimensional (3D) corneal topography
- •13.1.1.3. Crystalline lens parameters
- •13.1.1.4. Refractive index
- •13.1.1.5. Wavefront aberration
- •13.1.2. Eye Modeling Using Contemporary Optical Design Software
- •13.1.3. Optical Optimization and Merit Function
- •13.2. Personalized and Population-Based Eye Modeling
- •13.2.1. Customized Eye Modeling
- •13.2.1.1. Optimization to the refractive error
- •13.2.1.2. Optimization to the wavefront measurement
- •13.2.1.3. Tolerance analysis
- •13.2.2. Population-Based Eye Modeling
- •13.2.2.1. Accommodative eye modeling
- •13.2.2.2. Ametropic eye modeling
- •13.2.2.3. Modeling with consideration of ocular growth and aging
- •13.2.2.4. Modeling for disease development
- •13.2.3. Validation of Eye Models
- •13.2.3.1. Point spread function and modulation transfer function
- •13.2.3.2. Letter chart simulation
- •13.2.3.3. Night/day vision simulation
- •13.3. Other Modeling Considerations
- •13.3.1. Stiles Crawford Effect (SCE)
- •13.3.1.2. Other retinal properties
- •13.3.1.4. Optical opacity
- •13.4. Examples of Ophthalmic Simulations
- •13.4.1. Simulation of Retinoscopy Measurements with Eye Models
- •13.4.2. Simulation of PR
- •13.5. Conclusion
- •References
- •14.1. Network Infrastructure
- •14.1.1. System Requirements
- •14.1.2. Network Architecture Design
- •14.1.4. GUI Design
- •14.1.5. Performance Evaluation of the Network
- •14.2. Image Analysis
- •14.2.1. Vascular Tree Segmentation
- •14.2.2. Quality Assessment
- •14.2.3. ON Detection
- •14.2.4. Macula Localization
- •14.2.5. Lesion Segmentation
- •14.2.7. Patient Demographics and Statistical Outcomes
- •14.2.8. Disease State Assessment
- •14.2.9. Image QA
- •Acknowledgments
- •References
- •Index
Ooi, E.H. and Ng, E.Y.K.
where nr and nz are the components of the outward unit normal vector on i in the r and z direction, respectively, K and E denote the complete elliptic integral of the first and second kind, respectively, as defined by Abramowitz and Stegun,30 and m, a, and b are, respectively, given as:
2b(r; ξ)
m(r, z; ξ, η) = , a(r, z; ξ, η) + b(r; ξ)
a(r, z; ξ, η) = ξ2 + r2 + (η − z)2, b(r; ξ) = 2rξ.
Note that in the axisymmetric formulation, the z-axis does not form part of the curve boundary i.
The boundary element method is implemented by discretizing the boundary i into small straight-line segments, also known as boundary elements. The domain integral and the time-dependent variable in Eq. (12.10) are treated using the dual-reciprocity boundary element method and the time stepping scheme, respectively. For a detailed derivation, one may refer to the work carried out by Ooi et al.31
12.7. Results
Investigations on the transient temperature distribution inside the cornea during LTKP are carried out for both pulsed and continuous-wave lasers. Only lasers with a Gaussian beam profile are considered. The values of laser parameters used in the present study are selected based on the values used in a typical clinical LTKP treatment. Although a minimum of eight laser spots are usually applied to the corneal surface (see Sec. 12.2), in the present study, only a single laser spot, assumed to be applied to the center of the corneal surface, is considered. This assumption is necessary in order to maintain the axisymmetric feature of the human eye, which is of great computational advantage. With the laser spot placed at the center of the corneal surface, an increase in the temperature near the center of the corneal surface is expected.
To capture the large thermal variation over the small heated area around the center of the corneal surface accurately, the level of boundary discretization used has to be sufficiently fine. Likewise, the number of interior points
360
Temperature Changes Inside the Human Eye During LTKP
Fig. 12.3. Interior points selected inside each region of the human eye.
used to implement the dual reciprocity method has to be sufficiently large. In the present study, the human eye model is discretized into 253 boundary elements, with 72 of them located on the boundary of the cornea. A total of 316 interior points are selected inside the human eye model, with 53 and 142 of them placed inside the cornea and anterior chamber, respectively. The interior points selected inside each region of the human eye model are illustrated in Fig. 12.3.
12.7.1. Pulsed Laser
Table 12.3 summarizes the values of laser parameters of a typical LTKP treatment using pulsed laser. These values are obtained from Manns et al.12 Energy per pulse describes the amount of laser energy that is delivered onto the corneal surface during each laser pulse. The laser absorption coefficient of the cornea is obtained by assuming that the cornea has optical properties similar to those in water. The value of peak irradiance, Eo, in Table 12.3 is obtained using the following expression,11
Eo = Epp ,
πw2tp
where Epp is energy per pulse and tp is pulsed duration.
361
Ooi, E.H. and Ng, E.Y.K.
Table 12.3. Typical laser parameters chosen for the pulsed laser.
Parameter |
Value |
|
|
Energy per pulse, Epp (mJ) |
30 |
Pulse duration, tp (µs) |
200 |
Pulse repetition rate (Hz) |
5 |
Number of pulse |
7 |
Wavelength (µm) |
2.1 |
Laser beam radius, w (mm) |
0.3 |
Laser absorption coefficient, µ (m−1) |
2000 |
Peak irradiance, Eo (Wm−2) |
5.31 × 108 |
Based on the values given in Table 12.3, the function ψ(t) in Eq. (12.3) that describes the period when the laser is on and off may be expressed as:
ψ(t) |
= |
|
0, |
if t / J |
, |
(12.12) |
|
|
1, |
if t J |
|||
|
|
|
|
|
|
|
where t is time taken (in seconds) and J is the time interval defined by:
6 |
|
|
|
|
|
J = |
{t : 0.2002 m ≤ t ≤ 0.2002 m + 0.0002}. |
(12.13) |
= |
0 |
|
m |
|
|
In carrying out the time-stepping scheme, the time step, t is chosen to be
|
= |
0.05 s, |
if t / J |
t |
|
0.0002 s, |
if t J . |
|
|
|
|
Figure 12.4 shows the transient temperature changes along the pupillary axis (r = 0) at various depths of the cornea during treatment of pulsed LTKP. The seven temperature peaks seen in Fig. 12.4 corresponds to the seven laser pulses that are applied to the corneal surface. Cooling is observed at intervals between laser pulses where heat that is absorbed inside the cornea is diffused into the environment via convection and radiation and into the other ocular regions inside the eye. Throughout the treatment of LTKP, the corneal surface experiences the largest increase in temperature. At the end of the seventh laser pulse, temperature as high as 111◦C is reached; albeit for only a
362
Temperature Changes Inside the Human Eye During LTKP
Fig. 12.4. Transient temperature changes along the pupillary axis at various depth inside the cornea during pulsed LTKP with a Gaussian beam profile.
short duration. At the corneal endothelium, i.e. at z = 588 µm, temperature at the seventh laser pulse is found to be approximately 53◦C. Temperature at the corneal stroma, which is represented by the curves between z = 100 µm and z = 500 µm, increases beyond the threshold for corneal shrinkages after the third laser pulse.
Figure 12.5 illustrates the temperature profile along the pupillary axis at various depths of the cornea during each laser pulse, i.e. for t = 0.0002, 0.2002, 0.4006, 0.6008, 0.801, 1.0012, and 1.2014 s. The vertical lines separate the z-axis into regions occupied by the epithelium, stroma, and endothelium, while the horizontal lines represent the lower threshold for corneal shrinkages (T = 55◦C) and corneal relaxation (T = 90◦C). A large part of the corneal stroma, where the corneal collagens are located, is found to have temperatures that are greater than the threshold for corneal shrinkages. The high temperature indicates corneal shrinkages. At the corneal epithelium, temperature increases beyond the relaxation threshold. However, this may not severely affect the overall shrinkage of the cornea, since the collagens that are responsible for the contraction and relaxation of the cornea are not found in the corneal epithelium. At the end of the seventh laser pulse, the
363
