To main content

Does my child have cerebral palsy?

In a research laboratory in Trondheim, a physiotherapist is using a computer to identify signs of CP in new-born babies. Lars Adde of the Physiotherapy […]

In a research laboratory in Trondheim, a physiotherapist is using a computer to identify signs of CP in new-born babies.

CURVES AND LINES: The blue curves show Lars Adde’s key to fidgety movements, while the green lines show the neural network’s “evaluation”. In infants who are thought to be ill, the new measurement curve lies like a slack line without the peaks that characterize spontaneous activity in healthy babies.

CURVES AND LINES: The blue curves show Lars Adde’s key to fidgety movements, while the green lines show the neural network’s “evaluation”. In infants who are thought to be ill, the new measurement curve lies like a slack line without the peaks that characterize spontaneous activity in healthy babies.

Lars Adde of the Physiotherapy Section in St. Olav’s Hospital in Trondheim had an idea: he wanted to film and measure movements in new-born babies in order to be able to estimate the chances of them developing cerebral palsy (CP). Worried parents would no longer have to put their child through a long series of tests and then have to wait for nearly a year to find out whether or not it has the condition.

Today, the new research laboratory is an established fact, and the analyses so far are very promising. Video recordings and other measurements of movements have been made of 90 babies, some of them premature and others born at full term. Between ten and 20 of these babies have displayed an unusual pattern of motor development.

Measuring movements

An international methodology already exists based on evaluating movements in new-born babies as a way of estimating the risk of developing CP. Adde based his method on this system, but he wished to simplify the method, which is both time-consuming and requires a high level of specialist expertise.

Adde contacted Øyvind Stavdahl at SINTEF and asked him for technical support, and SINTEF has helped to build up the laboratory and analyse the data that has been collected.

The babies who are to be diagnosed are placed in a mobile bed and are fitted out with little sensors that are fastened to the arms, legs, head and chest with Velcro. The sensors are connected to a PC and video-camera, and they transmit 36 measurement signals 25 times a second. The data contain information that allows different aspects of the babies’ movements to be analyzed and studied.

Movement “key”

When infants are between two and four months old, they are in a phase of characteristic spontaneous movements known as fidgety. Although there are a wide variety of such movements, they obey a distinctive pattern which is a sign that the brain is functioning properly, while stereotyped movements and other disturbances of this pattern are a typical sign of CP.

Lars Adde studies all the recordings and draws up a “key” that tells us when the babies exhibit fidgety movements and when they are not. All the measurements and the key are fed into an artificial neural network which is trained to recognize the characteristic patterns of movement. When the neural network is subsequently fed with data from new babies it can evaluate whether they are displaying normal or deviant patterns of movement.

MEASUREMENTS: Lars Adde uses sensors fastened to the baby’s arms, legs, head and chest to transfer measurement data to a PC. Photo: Rune Petter Ness

MEASUREMENTS: Lars Adde uses sensors fastened to the baby’s arms, legs, head and chest to transfer measurement data to a PC.
Photo: Rune Petter Ness

Pioneer project

“You might say that we have taught the neural network to perform the same sort of evaluation as Adde does. We hope that in the future it will be possible to compare such technical diagnoses with those made by doctors and physiotherapists, so that we get an earlier, more reliable diagnosis of risk”, says Stavdahl.

The project is a good example of collaboration between a clinical milieu

(St. Olav’s Hospital) and a technology group (NTNU/SINTEF). Fresh competence in physiotherapy and modern methods of data analysis have been adopted. Financing the project has been difficult, but a high level of creativity on the part of Lars Adde has kept the wheels turning. Now we are hoping that the new method can be verified in a large-scale clinical research project.

By Åse Dragland

Contact

Øyvind Stavdahl