What makes this approach particularly interesting is that Dr Hoelscher wants to start treating stroke patients as soon as they are heaved into an ambulance, rather than waiting until they arrive in the emergency room. He plans to start a feasibility study before the end of the year in which some of San Diego's ambulances will be equipped with portable ultrasonic transducers. Andrei Alexandrov, a pioneer in therapeutic-ultrasound research who now directs the stroke centre at the University of Alabama, is pursuing a similar approach. He is designing an easy-to-use ultrasound helmet that an emergency-room nurse can attach before administering a vial of microbubbles. Both schemes are examples of a new idea in medicine, Which is to use tiny bubbles of gas not merely to highlight organs during ultrasonic scanning, as has been done for several years, but also as a form. of treatment. With clinical trials now getting under way, experts think it will take around five years for these new therapies to reach patients.

Microbubbles are not just any old bubbles. They contain a chemically stable gas, such as perfluoropropane, instead of air. This gas is encapsulated in a fatty shell rather like a very small balloon. Even the largest microbubbles being tested for medical use are only five microns across, less than the diameter of a red blood cell. More advanced bubbles are only a few hundred nanometres across and can move easily through the lining of a blood vessel. They may also, crucially, be able to cross the blood-brain barrier, a tightly sealed layer of cells that protects the brain from dangerous chemicals including many drugs. If you put such a drug in the surface layer of a microbubble, you might be able to smuggle it into the brain.

Having got into the brain (or anywhere else), a well-designed microbubble should also be able find a particular target. That is because the fatty layer can include molecules such as antibodies, which link up with proteins found on the surfaces of only one type of cell. A bubble with such an antibody in it would thus stick only to that type of cell. This sort of approach is being tested by Mark Borden and Paul Dayton, who work at another of the University of California's campuses, in Davis. They have demonstrated in rats that bubbles with an appropriate outer layer can be equipped with molecules that stick specifically to diseased cells. These molecules are initially hidden under a polymer layer to prevent the immune system from destroying the bubbles. When the bubble arrives at its target, however, it is blasted with ultrasound in a way that exposes the molecule and makes the bubble stick.

The two researchers can also use sound waves to steer bubbles towards a target, as if those bubbles were surfing a wave in the sea. Moreover, they can slow the bubbles down when they arrive where they are wanted. Once the bubbles have stuck good and fast to their targets, turning up the ultrasound still further will burst them, so that they release their payloads precisely where they can do most good. The result is smaller, better-aimed doses of drugs, which should mean fewer side-effects. In principle, such paytoads could be small-molecule drugs such as those used for cancer chemotherapy. They could be therapeutic prot