Toshiba algorithm promises safer neurointerventions

Rotational angio option adds bone to 3D model

Toshiba's new algorithm for creating a volumetric model of bone and soft tissue from data obtained during rotational angiography promises faster and safer neurointerventions, according to a luminary and clinical developer of the technology. Volumetric reconstructions typically eliminate the need to take multiple oblique images, said neuroradiologist Kieran Murphy. This shortens procedure time, reducing x-ray dose to the patient and lessening the risk of an adverse reaction, such as stroke.

Murphy and colleagues at Johns Hopkins Medical Institutions routinely use the new software, called 3D fusion digital subtraction angiography (3D-FDSA), to diagnose patients with brain tumors or complex intracranial aneurysms, as well as to plan and evaluate interventions. The package is helpful in planning difficult interventions, particularly those involving access at the skull base, he said.

Toshiba engineers in Japan started working on the 3D-FDSA algorithm 18 months ago, consulting primarily with Murphy and fellow Hopkins neuroradiologist Dr. Philippe Gailloud on how to improve it.

"We have been working with a particularly astute group of engineers who have been incredibly responsive," said Murphy, director of interventional neuroradiology and an associate professor of radiology and neurosurgery at Johns Hopkins. "They have responded in a remarkably rapid fashion to our advice."

Toshiba has begun selling the novel algorithm, unveiled last month (SCAN 2/25/04), as an option on its Infinix i-series of vascular x-ray systems. The product is scheduled to begin shipping within six months, and early response has been encouraging, according to Don Volz, director of Toshiba's vascular x-ray business.

"My guess is that in the next few months, we will be getting few orders that do not at least plan on adding this package at a later date," he said.

Although the clinical capability has evolved sufficiently, computing performance lags. The hybrid models of bone and soft tissue require about seven minutes to reconstruct. Low-resolution views can be processed more quickly, however, according to Volz.

Data for the reconstructions are obtained during two rotations. Each 200º sweep requires about five seconds. The first, which is done without contrast, establishes the mask needed for digital subtraction. The second, a contrast-enhanced sweep, records the vasculature.

If the pathology is located away from bony structures, a 3D DSA model may satisfy the clinical needs of neuroradiologists. If not, a second model combining DSA with topographic information would be created.

"If it is a tumor or complex aneurysm of the skull base, then we want to know its relationship with the bone, and we will do both-DSA and bone fusion," Murphy said.