Programming Language: OpenGL, C, C++, Titanium
Award:
Master Presentation: Download
OpenGL Source Code acquisition: Email toahyeek@gmail.com
Published Journal:
RESEARCH OVERVIEW - Research Framework
Award:
- Gold medal award of the Invention Exhibition of New Invention, Techniques and Products 2005,
- ITEX Bronze Medal award of the 16th International Invention Innovation Industrial Design & Technology Exhibition 2005 (ITEX 2005),
- “Saintis Cemerlang 2005” awarded by Ministry of Higher Education, 23 August 2005.
- Gold Medal award of the IPTA Expo 2005, PWTC, 2 October 2005.
Master Presentation: Download
OpenGL Source Code acquisition: Email toahyeek@gmail.com
Published Journal:
- N.Selvanathan, S. Y. Tan, S.Nagappan and M.Sankupellay. “The fiber-fluid model of the human heart”. Journal of Science & Technology in the Tropics, Vol 1. No. 1, Jun 2005.
- S. Y. Tan, Selvanathan Narainasamy, Somasundaram Nagappan. “NON-INVASIVE METHOD FOR PATIENT-SPECIFIC VIRTUAL HEART BASED ON FIBER-FLUID MODEL”. Journal of
RESEARCH OVERVIEW - Research Framework - Virtual Human Heart allows:
- Heart Surgery Planning
- Training of Cardiologist
- Drugs Testing
- Detailed Diagnostics
- Education of Cardiologist
Research framework for virtual heart model undertaken at University Malaya [1]:
[1] N. Selvanathan, Sieow Yeek Tan, Somasundaram Nagappan, Mangalam Sankupellay. (2005). The Fiber-Fluid Model Of The Human Heart. J. Science and Technology in the Tropics, Vol 1.
RESEARCH SCOPE
- 3D Rule-based cardiac fiber reconstruction mechanism.
- Immersed Boundary Method for fiber-fluid model.
- Titanium for distributed and high performance computing.
- OpenGL technologies for Visualization.
INTRODUCTION - Rule-Based Cardiac Fiber Reconstruction
- Novel technique : 3D Rule-based fiber orientation reconstruction.
- A single fiber consists of a set of fiber points.
- The fibers are constructed from a set of points (material points from epicardium) by registering them into a fiber.
- The selection mechanism of a registered material point will influence the constructed fiber sheet architecture.
- The mechanism is controlled tightly by a set of pre-defined rules that has been related carefully based on the formation of the fiber orientation.
- The rules take into consideration of:
- Fiber orientation angle
- Direction of fiber path
- Overlapping between two or more fibres
- Duplication in registered fiber points
INTRODUCTION - Immersed Boundary Method
- Proposed and developed by Charles Peskin and David McQueen.
- General numerical methodto efficiently simulate systems that are represented by elastic fibers immersed in an incompressible fluid [1].
- Fibers are considered as elastic material.
- Blood can be treated as a Newtonian fluid.
[1] Peskin, C. S. and McQueen, D. M., 1995. A general method for the computer simulation of biological systems interacting with fluids. Symposia of the Society for Experimental Biology, Cambridge, UK, Vol. 49 pp. 265-276.
INTRODUCTION - Visualization
- An important concept to display the simulated results on the computer screen.
- Translate numerical results into complex 3D objects.
- Interactive computer graphic is required during the model design and result analysis [1]:
- 3D objects can be animated, rotated, turned on and off.
- Magnify into any region of interest.
- [1] McQueen, D.M and Peskin, C. S. A Three-Dimensional Computer Model of the Human Heart for Studying Cardiac Fluid Dynamics. ACM Siggraph, 2000, Vol. 34 No1.
- Description of Fig (a) and (b):
- Displays the fluid flow and temperature distribution in a reactor.
- Fluid flow is represented as vector data (ribbons) and temperature as scalar data (colors).
- Displays fiber direction (yellow).
Methodology Outline
Acquisition of MR Images
- A set of high-resolution 2D MR images were obtained from a volunteer (27 years old healthy man) on breath-hold LV at the starting of systolic stage.
- Volumetric images are scanned based on T1-weighted Magnetization Prepared Rapid Acquisition Gradient Echo (MPRAGE) images with the dimension 256 x 256 x 16.
- Materialise Software “Mimics” was used to help in obtaining the images from MOD disk.
- Some selected MR images are shown below:
3D Left Ventricle Contour Model
- Segmentation:
- Image pre-processing to reduce image noise.
- Filtered a series tested of threshold values.
- Edge detection algorithm to identify and extract the ROI (LV).
- Segmentation algorithm using Active Shape Model (ASM).
- 3D contour model:
- Stacking all the contours (Fig. a).
- Transform all the 2D contour points into a 3D Cartesian coordinates.
- Scaled and translated the model to the centre of 64 x 64 x 64 3D experimental space (Fig. b).
Vertical Interpolation
- VI key points set is a set of key points identified throughout all the contours, noted as (K1).
First key point is identified in contour 1 (P1). - Evaluate the minimum distance between P1 and the key point on the contour 2 (P2).
- The same procedure is repeated till contour n.
- The VI points for set K1 is obtained by interpolating all the key points from P1 to Pn.
- The figures show the vertical interpolation process is being executed and the screenshots were captured from different points of view.
- Blue color points represent the vertical interpolated points while the contour points are in red color.
Horizontal Interpolation
- Key points are identified from vertical interpolated points.
- A maximum set of horizontal key points for each gap is fixed based on the maximum vertical interpolated points available in that gap.
- The point’s index in a horizontal key points set are arranged into a ‘ring’ manner.
- Figure below show the horizontal interpolation process was executing which captured from our visualization program from different points of view.
** Detail interpolation demo can be downloaded from here and view vie IE. (Note: Drag the downloaded GIF file to your IE or Open with your IE.) Download
Rule Based Cardiac Fiber Reconstruction
Rule Based Cardiac Fiber Reconstruction
- Rule and algorithm schematic. Please refer to the presentation, thesis or published paper for understand the fiber construction algorithm.
- Figure above show the required data representation in forming a single heart fiber in 3D environment and the model.
- Figure (a) and Figure (b) is the schematic to show how the fiber angle was formed in 3D space.
- Fiber path and fiber direction schematic. (Third figure).
- The author has faced difficulties to get the complex set of mathematic equations to be published on HTML web page. Please refer to the Master Presentation or Theis for the detail algorithm in constrcuting the heart fiber direction.
Refinement of Fiber Model
- Fiber-Point interpolation:
- Every pair of consecutive fiber points in each fiber list has varying gaps between each other.
- Some of the large gaps may cause inaccuracy during cardiac simulation.
- The gaps for each pair of consecutive points is needed to be limited to a reasonable distance.
- Applied to generate additional points in order to refine the fiber model (a).
- Fiber-Based Least Mean Square analysis:
- A fiber set will be removed if the fibers are too close to each other (b).
- When gap exists between two adjacent fibers is huge, a new fiber set need to be constructed (c).
- Fiber-Based Least Mean Square (FB-LMS) analysis on all pairs of adjacent constructed fiber sets.
- The LMS value of two adjacent fibers set is used to identify whether the two fibers are too far or too close from each other.
** Detail fiber construction demo can be downloaded from here and view vie IE. (Note: Drag the downloaded GIF file to your IE or Open with your IE.) Download
Adding Aortic Heart Valve
- A set of points from the constructed fiber model is obtained to act as the reference points.
- Adding the aortic valve to the fiber model aim for obtaining more realistic cardiac simulation results.
- Concentrated to build the aortic valve model.
- Obtained from the first point in each of the fiber set from model Ĥ.
- Generate the centre set of reference points from the original valve reference points.
- Combine the centre set of reference points to the valve reference points on the right hand side to form the mitral valve’s reference points.
- Combine the centre set of reference points to the valve reference points on the left hand side to form the aortic valve reference points.
- Divide the aortic valve reference points to three segments. Each segment is 120o sector.
Fluid Marker Generation
- Fluid marker is a set of fibers which they do not contribute any forces during cardiac simulation.
- Their movement will be influent by their surrounding fluid.
- Used to analysis the blood movement inside the LV chamber.
- Conform in terms of physical location within the constructed fiber model.
- Fluid markers are generated at multiple horizontal layers, each layer consists of multiple rings.
- Analyze the fluid dynamics at different locations in the LV chamber.
Visualization - Environment and Programs- Developed a series of visualization programs.
- Equipped with interactive computer graphics features .
- Toggle appearance of fiber in our heart model in real time.
- Interactively change our point of view.
- Rearrange data sets in frame by frame manner to produce animation effect.
- Allow us stop at a certain frame that we are interested.
- ØVisualize the simulate processes of fiber reconstruction mechanism.
- Verify the constructed fiber results in real time.
- Significant advantage: allows observation and modification of rules and simulation parameters setting.
- OpenGL Utility Toolkit (GLUT) 3.6. and C programming language.
- Compiled to run in Irix 6.5.13, SGI workstation, 2.4 MHZ Processor speed, 2GB of memory space.
Rule-Based Cardiac Fiber Reconstruction
- ØReconstructed Cardiac Fiber model shown without the model points.
- 1,200 fibers (yellow path) which contain 584,589 registered fiber points.
- ØFiber model is constructed by 1,038 fibers with the total of 371,658 fiber points after the refinement processes.
- ØModel is captured from different point of view.
- ØFiber model with some of the fibers turn off in order to view inside of the heart chamber.
- a.a) The constructed mitral valve fiber model shown in different point of views.
- b) b.Mitral valve fiber model location relative to aortic valve fiber model.
- c.c) The location of the valve fiber model (white) relative to the whole LV fiber model (green).
- d.d) First, second and third segment of aortic valve fibers together with the full aortic valve fiber model by combining each of the aortic segment.
- ØImmersed Boundary Method applied, simulate 1024 time step consumed approximately 11 CPU hours.
- ØFigures show the dynamic of LV cardiac muscle for each 0.05ms start from the beginning of ejection stage. Arranged from left to right for first and second row.
- The mechanics of blood flow in LV for each 0.05ms start from the beginning of ejection stage. Arranged from left to right for first and second row.
Aortic Valve- Aortic valve fiber dynamic start from early ejection for LV chamber. From (a) to (j), each frame represents the mechanics of simulated fibers muscle at different physical time. Start from 0.05ms for (a) and increase 0.05ms for the next. (j) shows the fiber dynamic at 0.5ms.
Fiber Model And Aortic Valve- The entire LV fiber model, including the valves fibers (white) and cardiac muscle fiber (green).
- a.a) The mechanics of blood flow nearest to the LV wall together with the cardiac fiber dynamic at the time 0.5ms.
- b.b) 2nd and 4th layer of ring fluid markers from each of the horizontal layer at 0.5ms respectively.
- c.c) The innermost ring fluid markers from each of the horizontal layers at 0.5ms.
- d.d) Several selected horizontal layers (outermost ring) of fluid markers which located near the apex of LV chamber.
- Observed that there are not much movement of blood at the middle of the LV chamber while the blood flow near the LV apex and the LV wall has a high expulsion.
- Cardiac Material Properties.
- Myocardial Activation.
- Cardiac Cell Model.
- High performance computing on:
- ØFiber-fluid model.
- ØRule-based fiber reconstruction mechanism.
- Fibrous-sheet structure.
- Visualization system.
Simulation Results Materials for Download:
(**Note: For GIF files, you can open it in IE. For .MOV file, please download quicktime player for view it.)
Fiber_Construction_Movie
Fiber_Construction_Movie
- The heart fiber construction process for the total number of 550 fibers. Download
- Others completed fiber construction movie (1700 fibers constructed). Download
- The fluid marker located near the apex of
- The fiuid marker located near the base of
- The simulation of blood flow in
- Same as FullMarker.mov simulation but with different view of perceptions are captured. Download
- Same as FullMarker.mov simulation but movie captured while horizontal view perceptions is changing. Download
- Fluid marker simulation with only the inner most rings for all horizontal layers are turn on. Download
- Fluid marker simulation with only the ring 3 and ring 4 for all horizontal layers are turn on. Download
- Another fluid marker simulation with Ring 3 and Ring 4 for all the horizontal layers are turn on. The 8 horizontal layers of ring 2 are also turn on. Download
- Fluid marker simulation with all outermost rings for all horizontal layers turn on. Download
- Simulation of the heart fiber contraction with valve attached. Download
- Simulation of the tricuspid valve. Download
- Simulation of the valve including the dummy aortic valve is turn on. Download
- Contain images show the cubic Hermite vertical and horizontal interpolation process and results. Download
No comments:
Post a Comment