Detecting patterns in waves

Question

I\'m trying to read a image from a electrocardiography and detect each one of the main waves in it (P wave, QRS complex and T wave). Now I can read the image and get a vector li

Answer 1

One approach that will very likely yield good results is curve fitting:

• Divide the continuous wave into intervals (probably it's best to have the interval borders about half way between the sharp peaks of the qrs complexes). Only consider a single interval at a time.

• Define a model function that can be used to approximate all possible variations of electrocardiographic curves. This is not as difficult as it seems first. The model function can be constructed as a sum of three functions with parameters for the origin (t_), amplitude (a_) and width (w_) of each wave.

     f_model(t) = a_p   *  f_p  ((t-t_p  )/w_p) + 
                  a_qrs *  f_qrs((t-t_qrs)/w_qrs) +
                  a_t   *  f_t  ((t-t_t  )/w_t)
  

  The functions f_p(t), f_qrs(t), f_t(t) are some simple function that can be uses to model each of the three waves.

• Use a fitting algorithm (e.g. the Levenberg-Marquardt-Algorithm http://en.wikipedia.org/wiki/Levenberg%E2%80%93Marquardt_algorithm) to determine the fitting parameters a_p, t_p, w_p, a_qrs, t_qrs, w_qrs, a_t, t_t, w_t for the dataset of each intervall.

  The parameters t_p, t_qrs and t_p are the ones you are interested in.