Eon uses computational linguistics to identify pulmonary nodule

28 October 2020 (Last Updated October 28th, 2020 14:33)

US-based health-tech company Eon has announced that it is leveraging data science platform to identify incidental pulmonary nodules on magnetic resonance (MR) and X-Ray radiology reports.

Eon uses computational linguistics to identify pulmonary nodule
Eon’s EPM platform has been updated to identify incidental pulmonary nodules on MR and X-Ray radiology reports. Credit: Minerva Studio / Shutterstock.

US-based health-tech company Eon has announced that it is leveraging data science platform to identify incidental pulmonary nodules (IPN) on magnetic resonance (MR) and X-Ray radiology reports.

The company’s Essential Patient Management (EPM) platform is an all-inclusive lung cancer screening and IPN identification management solution.

The platform uses computational linguistics to identify incidental pulmonary nodules on computed tomography (CT) reports with 98.95% accuracy and 97% accuracy on MR and X-Ray radiology reports.

Additionally, EPM draws out clinically relevant findings from radiology reports.

The expanded capability of the platform is expected to help facilities in capturing approximately 25% more incidental pulmonary nodules and equip them to spot lung cancer earlier.

Eon chief science officer Dr Erika Schneider said: “Any imaging that covers a lung field can identify an unexpected pulmonary finding such as an IPN. Hundreds of thousands of IPNs each year are identified on CT and MR exams, often of anatomy other than the chest.

“Suspicious or concerning areas of abnormal density on radiographs are also common. Unfortunately, these nodules and abnormal regions are frequently lost to follow-up or inappropriately followed.”

The use of computational linguistics enables healthcare providers to positively identify and track incidental pulmonary nodules with better accuracy compared to other forms of artificial intelligence (AI) such as natural language processing and computer-aided detection.

Schneider added: “Computational Linguistics is the gold standard for language understanding, in particular for lung nodule identification and characteristics extraction. By embedding evidence, the nodule characteristics focus providers’ attention on patients with a high probability of having lung cancer.

“The high accuracy and reproducibility of our model reduces false positives and does not require radiologists to use a structured report. This approach, along with the embedded risk prediction and automation, should enable providers to prioritise patients and improve their outcomes.”

Eon recently launched an AAA solution, called Actionable Findings module. The company plans to expand the developments for application in areas of pancreas, thyroid, and breast next year.