You need to use the OrthographicCamera for that. It’s nice that THREE supports a similar api to the in-built canvas one, but it’s purpose is rather for 3D and is less efficient for memory when drawing/constructing frequently in it’s default implemention, if i’m going to be a little nitpicky. 
Using the default canvas might rather fit your needs maybe, since it also allows better antialiasing and thickness support. Or is there a specific reason using THREE for this task?