Ultrasound scans are used routinely for important medical reasons, such as checking a baby’s heartbeat or identifying potential birth defects. They’re also used for trivial purposes, such as making a video of the fetus for grandparents. Whatever its use, though, the technology, which relies on high-frequency sound waves, has been widely assumed to be harmless. But a new study by neuroscientists at Yale University shows that prolonged exposure of prenatal mice to ultrasound can cause neural defects in the area of the brain responsible for complex functions, including those involved in memory, language processing, and consciousness in humans.

Researchers say that, while the study should not scare pregnant women away from sonography appointments, it provides an important reminder that ultrasound scans should be used only when medically necessary. The Yale scientists are now conducting long-term behavioral studies in mice and rhesus macaque monkeys, which represent a much closer model of human gestation and brain development. They hope these further studies will provide a better assessment of the procedure’s safety.

The mouse ultrasound study, led by Pasko Rakic, director of the Kavli Institute of Neuroscience at Yale, specifically found problems in the migration of neurons into the developing cerebral cortex. If misplaced neurons live, and fail to make the right connections, they can cause seizures, delayed language acquisition, or other behavioral problems. These neurons don’t cooperate properly with the neurons around them because they behave as though they’re somewhere else in the brain. Problems with neuron migration have been associated with fetal alcohol syndrome and birth defects that result from pregnant women taking cocaine.

In order to find out whether ultrasound affected the ability of cortical neurons to reach their proper destinations, Rakic’s group injected pregnant mice with a DNA-labelling molecule that becomes incorporated into the chromosomes of rapidly dividing cells. The injection was timed to coincide with the generation of cortical neurons. Over the next three days, when these neurons are known to migrate, the mice were exposed to multiple ultrasound sessions totalling from five to 420 minutes. Ten days after their birth, the mice were killed and the labelled neurons were located. When prenatal mice are exposed to ultrasound for 30 minutes or longer, the Yale researchers found, a small but significant number of neurons fail to reach the appropriate positions. Because the mice were killed after ten days, the researchers don’t know whether the anatomical abnormalities they observed would have led to behavioral differences or seizures.

Researchers have known that ultrasound beams can generate tissue-damaging heat. But Rakic’s study suggests a different possible mechanism by which ultrasound might disrupt tissues: shear stress on cellular walls. The vibrations might make it difficult for migrating neurons to cling to the guides, called glial shafts, that help them reach the proper destination.

Rakic and other neuroscientists caution that the mouse results do not directly apply to humans for several reasons. First, the mice had much longer exposures to ultrasound than is recommended for human prenatal exams, and the sound waves were aimed directly at their brains. Prenatal ultrasound scans may last around 30 minutes, but during that time the entire fetus, as well as the amniotic fluid and the mother’s cervix, are being examined. At a comparable stage of development, says Verne Caviness, chief of pediatric neurology at Massachusetts General Hospital, the human brain is about 1,000 times bigger than the mouse brain; by comparison to the mouse, the ultrasound beam scans only a small portion of the human brain at once. No area of the human brain is exposed to ultrasound for more than a minute during a standard prenatal exam, says Caviness.

But Rakic’s study is still very important, says Caviness, because it “reminds us that ultrasound, which has been used confidently and routinely, is not to be dealt with trivially.” Having state-of-the-art equipment that can accurately report the energy levels of the sound beams it generates is crucial, says Caviness.

John Newnham, head of the School of Women’s and Infants’ Health and professor of obstetrics and gynecology at the University of Western Australia, says that the greatest danger to women and unborn children from ultrasound comes from inadequately trained practitioners. “There needs to be great attention paid to the training, accreditation, and ongoing education of sonographers,” says Newnham.

Because migration problems in mice exposed to ultrasound were rare, Rakic’s study required 335 animals to demonstrate the effect. His study in rhesus monkeys, which is funded by the National Institutes of Health, could take many years to demonstrate the same effect – if it occurs in the primates. A rhesus macaque pregnancy lasts six and a half months. The animals will be kept alive for a few years and examined by behavioral specialists for, among other characteristics, their cognitive abilities and attachment to their mothers.

Rakic says the DNA labels on the monkey’s nerves will last years, allowing his group to locate cortical neurons. Because these studies require such labelling, they cannot be done in humans. But the monkeys provide a better model of human development than mice: their neurons migrate for 60 days, six times longer than mice’s. Human neurons may migrate from the first trimester through the 24th week.

“If these changes are seen in monkeys, that will be something to take seriously,” says Dale Purves, director of the Center for Cognitive Neuroscience at Duke University. Meanwhile, he says, pregnant women should not be alarmed, although they should avoid unnecessary ultrasound scans.