How to use a trained convolutional network when input has only 1 channel?

Hi experts,

I have used [XTrain,YTrain] = digitTrain4DArrayData; to obtain 5000 images with size 28*28*1. They are handwriting images with only 1 channel. I followed the Guide Manual to create and train a convolutional network successfully. And then, I want to use the trained net by using

Y=net(X);

where, X is a 28*28 image. Because it has only one channel, MATLAB reduces it from 28*28*1 to 28*28 automatically. So the input format becomes wrong! It returns

"Array indices must be positive integers or logical values."

I checked on MATLAB webpage. All about the introduction is how to train the network, they do not even show a single example to test using the trained network by their 28*28*1 images. So did anybody of you have tried inputing something into the network (trained with 28*28*1 images)? What did you get? Thanks.