HeadCT-Reconstruction

This example shows the 3D head CT scan reconstruction using various hole-filling methods of pixel nearest-neighbour (PNN).

Firstly, the program reads a volume dataset, and removes a whole slice at position n. User can also choose to remove 2 or 3 slices. The slice removal can be removed at a specific sparsity s. Then, the hole-filling method is employed to reconstruct the missing region.

Pre-requisite

1. Microsoft Visual Studio
2. The Visualization Toolkit (VTK)
3. CMake

How to run this example?

1. Using CMake to create this project. The procedure is the same as any VTK projects.

2. Build the project using Visual Studio. A new "Debug" folder will be created.

3. Put the volume dataset file (headsq) in the "Debug" folder.

4. Open command prompt and navigate to the "Debug" folder.

5. Run HeadCTReconstruction *outputDatasetName(string) *sliceNo(int) *method(string) *parameter(int, float)

   e.g.:

HeadCTReconstruction headsq 7 mean 3 
HeadCTReconstruction headsq 7 butterfly-my 0 3 0

6. Then, input the number of continuous slice to remove.
7. Input the sparsity value.
8. Lastly, input the increment limit of sparsity value. If you want to remove 2 slices in every 7 slices spacing for 10 times: 2 -> 7 -> 10.
9. The result is displayed. The output axial slices are stored in the "figures" folder.

NOTE:

1. The dataset used in this example consists of 12-bits grey-scale pixel with little-endian arrangement.
2. The input dataset is preset to "fullHeadRaw" folder. See main().

For more information, please visit this conference proceeding or its ResearchGate counterpart.