|
Description:In many areas of experimental science, the acquisition and computational analysis of 2D, 3D, or 4D images is a necessary step in the scientific investigation. This class will cover concepts and methods of scientific image processing and analysis, in the context of real-world imaging data, from ongoing research on campus. In this offering of the class (Autumn 2013), datasets and driving scientific problems come mainly from research in biology and physics. Class has four components: readings on the image processing methods and applications, lectures to cover concepts and mathematical background, guest lectures by researchers to learn about the scientific context of the imaging, and programming exercises with datasets shared by the researchers. |
Note: If you are taking this class to satisfy an elective requirement in the Computer Science department, you will need to propose (by roughly mid-quarter) and complete (by the end of the quarter) a programming project. The project proposal and submission will be done by arrangement with the instructor. The scope and difficulty of the project will have to be distinctly larger than that of the regular programming exercises.
Each week of the course is expected to cover the following material. The first half of the course covers mathematical fundamentals and the methodological underpinnings of the image analysis methods used by the scientists. The second half goes into detail on specific applications, including a presentation from a scientist to describe the research context of the image data acquisition, the analysis questions that are asked of the data, and the methods used to answer those questions.
Week: | Material |
---|---|
Week 1 (Oct 2) |
Basics: Representation and storage of values and arrays, structure of computer memory hierarchy, array manipulations (slicing, tiling, projections), histograms and histogram analysis, thresholding and quantization, image file formats |
Week 2 (Oct 9) |
Linear Filtering: Convolution (discrete and continuous), Blurring, derivatives, separability, Fourier transform and the FFT, continuous kernel design |
Week 3 (Oct 16) |
Guest Lecture by Yali Amit |
Week 4 (Oct 23) |
Image Features: Isocontours, edges, ridge lines and ridge surfaces, local extrema, extraction of features in the discrete and continuous image domains |
Week 5 (Oct 30) |
Scale: multi-scale processing (discrete in scale), blurring as diffusion, non-linear PDE-based filtering, scale-space (continuous in scale), image analysis in scale-space |
Week 6 (Nov 6) |
Presentation from a scientist, and discussion of the associated data analysis challenges |
Week 7 (Nov 13) |
Scientific presentation and data analysis: Patrick La Riviere (Medical Physics) |
Week 8 (Nov 20) |
Scientific presentation and data analysis: Ka Yee Lee (Chemistry), and David Biron (Biological Physics) |
Week 9 (Nov 27) |
Scientific presentation and data analysis |
Week 10 (Dec 4) |
Scientific presentation and data analysis: Kevin White (Genetics and Systems Biology) |