Dear Hari Mohan Rai, My advice is not to search and limit ECG signals. Get some papers related to general signals and then orient your problem accordingly. Check MATLAB toolbox. Anyway, find below a link for an article. Check how much it is useful for you.

http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=4052239&abstractAccess=no&userType=inst

Afaq Ahmad is right, you have to refer to some other articles about wavelet and curvelet in sigal processing field.

http://apps.webofknowledge.com/summary.do?SID=X1kNwnhNDGaXohBLaUX&product=WOS&qid=1&search_mode=GeneralSearch

Hi Mr.Hari Mohan, did you find means to denoise ECG using curvelets?

How did you manage to bypass the 2D requirement of curvelets since ECGs are

1D only? Your input is highly appreciated since i am new to curvelet and shearlets.

I think the best way is to try some experiments. Take a clean ECG signal and normalize it. This signal is now your ground truth signal. Add SNR= 10-30 db additive white gaussian noise to the signal. Denoise this noisy signal using wavelets and curvlets. Once you do this compute some simple metrics to evaluate the similarity between your ground truth signal and your denoised signal. Metrics such as MSE, SNR etc. come in handy. Do this trial for a few ECG signals. The evaluate the average value of the metrics as average +- standard deviation. The method that yields the lowest value of MSE and the highest value of SNR is the one that performs the best. This is typically the procedure followed in any signal denoising paper today ! You can use the code below to compute the metrics. It also contains a demo of how signal denoising and performance evaluation typically works, with the example of a simple sine wave.

http://www.mathworks.com/matlabcentral/fileexchange/52342-evaluating-performance-of-denoising-algorithms-using-metrics---mse-mae-snr-psnr---cross-correlation