Stacked hybrid learning U-NET for segmentation of multiple articulators in speech MRI