There are an astonishingly few pre-requisites listed for this course. More on this later.
In Sajda's Signal Processing and Modeling (SPAM) class, you learn all about the mathematical concepts used for processing and modeling biosignal data. In a nutshell, Sajda goes over the basic theories and methodologies he uses largely in the context of his own biomedical research. He is approachable and knowledgeable and is a leader in the field of brain signal imaging and analysis. He holds a patent for a brain cap with electrodes and uses it in his research to sense what people are thinking– it is a fascinating field. Columbia is lucky to have him.
But about the course: BMEN E4420 is a graduate-level course with many PhD students largely from the EE and BME departments, many of whom have advanced statistical knowledge (beyond our pitiful W1211) and taken advanced signals classes (Signals and Systems and DSP). With this as a required class for the few biomedical signals/imaging undergrads in the BME department, we were constantly encountered new content that built upon completely foreign topics. One would attend class with absolutely zero knowledge on the topics covered and leave class with zero net gain. This would happen regularly. Often, something would be covered in this class and a week later would be covered in Signals and Systems (ELEN E3801). This made Signals slightly easier, but SPAM would continue to be a crapshoot. I later discovered that this class was originally offered in the spring, AFTER we were to have had taken Signals and Systems. It is unclear why we had the fortune of taking in the fall this year.
This course departs from the standard regimen of lecture classes with a meeting once a week, a different topic covered during each lecture. A typical class would have 3 hours allotted but usually ended 30 minutes early with a short break (10-15 min) in the middle of class. He assigns readings for each week consisting of textbook chapters that are long and not helpful for the homeworks but handy for the midterm. This material is complicated. Luckily he posts the slides before class and they are very helpful for when you finally sit down with your study group to do the assignment (no study group? find one, you're going to need one).
Homeworks are a bulk of the grading for this course and often involve using MATLAB to recreate the plots and models he has in his slides, with Sajda providing data sets and me and my partners dissecting the slides for anything pertinent. He tends to provide equations or basic MATLAB pseudocode and makes the assignment getting that code to work. I'm pretty sure each assignment took us easily 6+ hours of blank staring and trial and error. It's a good idea to have at least someone go to office hours, so that at least you'll know where the assignment is headed. Otherwise, you can spend hours in a state of confusion. This was the class that made me realize I had to give up at least one day in the weekend purely for schoolwork. It is a blessing that the homework was graded quite leniently.
So from taking this class with basically zero background knowledge, you will quickly realize that by the end of this course you will not have learned much aside from the names of a few mathematical concepts and models. Sajda is a pleasant professor and certainly knows his material well, but the shortcomings with this course come not from him but from the Biomedical Engineering Department not adequately preparing undergrads with the knowledge required for a class like this one. As a result, this course becomes a game of trial-and-error, praying and dead reckoning. I don't think any of the undergrads I took this class will finish with any real idea of what is going on.
Good luck and Godspeed to any future BME Imaging majors.