Wouldn't that be just great, instead of having your cavities filled with resin or a metal alloy, simply regenerate the enamel at the cavity site. Sounds simple enough because enamel (the 2 mm thick outer layer of a human tooth) is just inorganic hydroxylapatite and the cells that made it have disappeared long ago when construction was done. Haifeng Chen et al. @ Peking University has this simple recipe (Yin et al. 2009 DOI): a human molar is etched with phosphoric acid and then put in a solution of calcium phosphate , the chelating agent N-(2-hydroxyethyl)ethylene-diamine-N,N,N-triacetic acid and KI at 37°C for 8 days.
Lo and behold new hydroxyapatite layer forms, nucleated from the surface of the existing enamel, 2 to 4 micrometer in thickness in typical enamel hexagonal rodlike crystals. In fact, the initial morphology is amorphous, it then changes to spherical crystals (Ostwald ripening) and finally (if you wait long enough) the thermodynamically favored nanorods form. The researchers also report comparable nanomechanical properties.

In 2008, US-based team investigated the role played by a enamel matrix protein called amelogenin (Wang et al. 2008 DOI). This team thinks the protein assists in the kinetic formation of the nanorods from nucleating nanoparticles via self-assembly.

Earlier in 2005, the Japanese competition prepared a ready-to-use paste for enamel restoration based on phosphoric acid / hydrogen peroxide / fluorized-apatite powder, that when applied to tooth at the site of a lesion (early stage of cavity formation) showed immediate results (20 micrometers growth) within 20 minutes with the the apatite crystals dissolving and regrowing (Yamagishi et al. DOI). To demonstrate the durability of the new layer the site was exposed to 10,000 brushing actions (150 rev./min, load force 200 g, brushing amplitude 30 mm) and luckily the Japanese invented a machine to do that for them.