Perifissural pulmonary nodules (PFNs) are benign and are often located close to or attached to a lung fissure. They are also usually characterized by a triangle-like shape, said presenter Francesco Ciompi, PhD, of Radboud University Medical Center.
Despite their benign nature, it's common to observe PFNs with a fast growth rate similar to that of malignant lesions, Ciompi said. As a result, it's important during screening procedures to accurately classify a nodule as a PFN to avoid unnecessary biopsies.
The researchers sought to use computer vision and machine-learning algorithms to tackle this problem.
"We believe that machines can help radiologists speed up the reading procedure and quickly provide additional information useful to deciding the best kind of patient-tailored follow-up," Ciompi told AuntMinnie.com.
In testing, the group's approach was feasible and offered potential to be utilized as an automatic tool in computer-aided detection, according to the researchers.
