It is pointed out,that with a real study example,the key regions may be inreliable if only according to the absolute values of difference of the mean values to pick out the key regions. That may result in false regions and miss the important ones. The methods of solving this problem are presented.