Machine learning, fetal MRI ID patients requiring CSF diversion

In babies born with fetal ventriculomegaly, an enlargement of the cerebral ventricles in utero, determining when cerebrospinal fluid (CSF) diversion is required after birth can be difficult.

A new study, published online Feb. 1 in JAMA Pediatrics, found image analysis paired with machine learning of fetal MRI can predict a need for postnatal CSF diversion among patients with fetal ventriculomegaly. In fact, authors found the method was able to make this prediction with an accuracy rate of 82 percent.

“The ability to prospectively identify which patients with fetal ventriculomegaly will develop postnatal hydrocephalus and require subsequent CSF diversion would provide critical prognostic information to parents and aid in the clinical management in the postnatal period,” wrote corresponding author Jared M. Pisapia, MD, in the department of neurosurgery at the University of Pennsylvania, and colleagues.

The team performed the retrospective case-control study by analyzing information of 253 patients with fetal ventriculomegaly from 2008 through 2014. Data were taken from an institutional database.

A total of 50 patients with ventriculomegaly were analyzed—25 who required postnatal CSF diversion were selected and matched by gestational age with 25 patients with the same condition who did not require CSF diversion.

The model was applied to a sample of 24 consecutive patients with fetal ventriculomegaly who underwent evaluation at a different institution. Data were analyzed from 1998 through 2009.

A total of 74 patients were included from both groups with the main outcome measures including accuracy, sensitivity and specificity of the model to classify patients in need of postnatal CSF diversion correctly.

Results were as follows: 82 percent accuracy, 80 percent sensitivity and 84 percent specificity. The replicated cohort model predicted with 91 percent accuracy, 75 percent sensitivity and 95 percent specificity.

“Image analysis and machine learning can be applied to fetal MRI findings to predict the need for postnatal CSF diversion. The model provides prognostic information that may guide clinical management and select candidates for potential fetal surgical intervention,” wrote Pisapia et al.