ACRIN CT colonography trial: does reader's preference for primary two-dimensional versus primary three-dimensional interpretation affect performance?