How to statistically detect "trend outliers"?

Hi all,

I have several data collected by experiments that show the dependency between two variables, let's say x and y. I have one trend for each experiment because the x is continously varied during the experiment. In the following link you can see a plot with all the experiments collected.

<http://s27.postimg.org/x61xxcs0j/lines.jpg>

As you can see there are more spiky trends or trends with a peak placed in a different location than others. These may be outliers and in case I have to identify them and remove the full trend.

Is there any statically way to remove line outliers that I can implement in matlab? Also a paper with the methodology may be fine.

Thank you

best regards

Pietro