A number of landmark papers on iPS cells came out this summer. A combination of three to four transcription factors deliverd by retroviruses could change somatic cells into induced pleuripotent stem cells.  Cells could be taken out from patients suffering from diseases like  ALS, Parkinson's, Down's syndrome etc and induced to form stem cells.  These stem cells could further be reprogrammed to differentiate into specific cell types. These cells are not ready to be used in the regenerative medicine at present due to retroviruses used to make iPS cells.

A recent study published in Nature shows that somatic cells could be directly reprogrammed to go from one type to another without going through  the dedifferentiation-iPS step. Zhou et al show that by expressing three  developmental factors (different from the ones used for iPS generation)  in the pancreas of the mice, they could directly program exocrine cells into switching to a insulin producing beta cell state. Although induced beta cells did not form islets, they produced similar level of insulin as endogenous beta cells.

The reprogramming genes (Ngn3, Pdx1 and Mafa) were delivered through adenoviruses in vivo thru injection. The same injection in the muscles in vivo or fibroblasts in vitro did not cause them to acquire a insulin producing state, indicating that transdifferentiation or transdetermination would have to done on a tissue dependent basis i.e. it would be easier to reprogram closely related cells than distant ones and it would be require more factors to reprogram a muscle cells to acquire an islet cell state.

In principal it would be possible to get abundant tissues like skin, blood cells or adipocyte from a patient and reprogram them into desired cell type or form iPS cells for regenerative medicine.

I can't post the image here due to Nature's funky copyright and attribution policies, but the induced beta cells that appear after reprogramming which can be seen in the as single cells or small clusters is quite amazing.