New Course Proposal - ENSC 474-4 Biomedical Signal and Image Processing

Calendar Information

Course Number: ENSC 474

Course Title: Biomedical Signal and Image Processing

Credit Hours: 4

Vector: 3-0-2 (lecture-tutorial-lab)

Course Description

Develops signal processing techniques of wide applicability, presented in the context of processing and analysis of biomedical images. Forms a sequel to the course ENSC 374-4, Introduction to Biomedical Imaging, which covers acquisition of medical images. The subsequent visualization, processing and analysis tools applied to multidimensional signals such as 2D/3D medical images are covered. Students will become proficient in several basic tools used in signal processing by looking at their multidimensional counterparts for image processing.

Prerequisites: ENSC 380-4 Linear Systems and either ENSC 327-4 Communication Systems or ENSC 328-1 Random Processes in Engineering.

Recommended: None

Corequisite: None

Rationale for Introduction of This Course

This course is prominent in the Biomedical Signals and Instrumentation concentration of the new Biomedical Engineering curriculum.  Much of biomedical engineering has to do with signals, typically physiological measurements or images.  The course introduces important techniques for reduction of measurement noise, enhancement and transformation of images, and detection of patterns within signals. 

The course forms a sequel to ENSC 374-4, Biomedical Image Acquisition, which covers acquisition of medical images. The subsequent visualization, processing and analysis tools applied to multidimensional signals such as 2D/3D medical images are covered. Students will become proficient in various basic tools used in signal processing by looking at their multidimensional counterparts for image processing.

The topics are often seen at the graduate level, but there does not seem to be a reason why a good set of signal processing tools cannot be made available to senior undergrad students.

Will this be a required or elective course in the curriculum; probable enrolment when offered?

This is an elective course in the Biomedical Signals and Instrumentation concentration of the BME curriculum.  Probable enrolment: 20

Scheduling and Registration Information

Indicate Semester and Year this course would be first offered and planned frequency of offering thereafter.

First offering to be Fall 2008.  Annually thereafter in the Fall semester.

Which of your present CFL faculty have the expertise to offer this course? Will the course be taught by sessional or limited term faculty?

Dr. Faisal Beg, Dr. Jie Liang.  The course will be taught by tenure-track faculty.

Are there any proposed student fees associated with this course other than tuition fees?

No.

Is this course considered a `duplicate' of any current or prior course under the University's duplicate course policy? Specify, as appropriate.

No.

Resource Implications

Note: Senate has approved (S.93-11) that no new course should be approved by Senate until funding has been committed for necessary library materials. Each new course proposal must be accompanied by a library report and, if appropriate, confirmation that funding arrangements have been addressed.

Provide details on how existing instructional resources will be redistributed to accommodate this new course. For instance, will another course be eliminated or will the frequency of offering of other courses be reduced; are there changes in pedagogical style or class sizes that allow for this additional course offering.

Drs. Beg and Liang are recently hired faculty members in the School of Engineering Science.  Biomedical image processing is Dr. Beg’s primary research area.  No redistribution of resources or elimination of courses will be required.

Does the course require specialized space or equipment not readily available in the department or university, and if so, how will these resources be provided?

Existing space and equipment is adequate for this course offering.

Does this course require computing resources (e.g. hardware, software, network wiring, use of computer laboratory space) and if so, describe how they will be provided.

The computing resources it requires exist in Engineering Science. 

Course Outline

Topics:

1. Representation and sampling of digital images. Overview of the components of an image processing system. Visualization and display of medical images.
2. Noise reduction and enhancement techniques for biomedical images. Histogram Equalization. Spatial Filter design - smoothing, order-statistics, sharpening, adaptive filters. Frequency domain filter design - ideal, Butterworth, Gaussian filters. Convolution and correlation of images. Applications to feature detection.
3. Interpolation and model-fitting for discrete multidimensional data. Linear synthesis model, B-splines, Interpolant functions, Cardinal Splines.
4. Rigid and Non-rigid transformations of medical images. Computing transformations based on point correspondences and grayscale image intensity matching.
5. Morphological Image processing - dilation, erosion, opening, closing operations applied to images for feature extraction and noise removal. Histogram Equalization techniques. Simple segmentation techniques based on morphology, histogram thresholding, Bayesian segmentation.
6. Representation and description of anatomical objects: medial-axis-based representation, Fourier Descriptors, Topological descriptors. 3D Representation using surface models.
7. Pattern recognition. In one-dimensional and two-dimensional signals; feature extraction; linear and nonlinear discriminant functions; training. Statistical Moments, Principal Components Analysis. Applied to distinguishing normal and diseased cells, and to classification of EEG and EMG signals for biocontrol.

Projects/Laboratory Work

Weekly assignments will consist of both a theoretical and a computational component. The computational component will require writing computer programs in Matlab/C++ to process medical images for enhancement, visualization and information extraction. Sample projects can be: (1) implementing various filters for noise removal from images (2) recovering transformations for correction of distortion or registration of medical images to a chosen template image for data pooling and averaging (3) segmentation of objects in medical images using morphological operators as an example. (The medical images used in the course will be from publically available databases; thus, no issues of patient privacy will arise.)

Grading

Grading will consist of Midterm (20%), Final (25%), Project (15%) and weekly assignments (40%).

Textbook

Possible choice is the textbook Handbook of Medical Imaging by Isaac Bankman. Another possible choice is Digital Image Processing (2nd Edition) by Gonzalez and Woods.