Short comment from my side: Using the DicomImage class for extracting frames from a multi-frame image without losing information is certainly not the way to go. This is because the main purpose of this class is rendering images for display.
There is no general answer to this question. If you have a DICOM image type that exists both in a single frame and a multiframe variety, then converting from multiframe to single frame requires an adaptation of all physical parameters related to the image acquisition. This is a fairly complex tasks and depends on the details of the modality.