Nowadays the most common approach for this kind of segmentation tasks is to train a unet network for this. You can create training data sets using SlicerAIGT extension’s “Single Slice Segmentation” module (it is specifically developed for ultrasound image sequence segmentation). I think there are a number of examples for ultrasound segmentation networks in the extension’s repository.
Do you track your ultrasound probe (e.g., using an optical or electromagnetic tracker)? If not yet, then you can use Plus toolkit for that. It integrates nicely with 3D Slicer.