Lung cancer is the deadliest cancer in the world and about 75% of those are diagnosed with it will die within five years of discovery. Most patients are only diagnosed after the disease has advanced to the serious Stage III or IV level. However, if tumors are small and confined to the lung, the survival rate increases substantially. The need for early detection is clear and AI is powering the development of systems that can detect ever-smaller lung tumors.
Optellum, a lung health company aiming to redefine early diagnosis and treatment of lung disease, has received FDA 510(k) clearance for Virtual Nodule Clinic, a revolutionary AI-powered clinical decision support system that helps pulmonologists and radiologists uncover early stages of lung cancer. This is the FDA’s first application of AI decision support for early lung cancer diagnosis.
The world’s first FDA-cleared imaging AI/” Radiomics”-based digital biomarker for lung cancer, Virtual Nodule Clinic helps pulmonologists identify and track at-risk patients with suspicious lung nodules, then make optimal clinical management decisions for them. The software provides a clinically validated Lung Cancer Prediction (LCP), which is computed from full patterns of 3D pixels in standard CT image scans.
Use of the Virtual Nodule Clinic dramatically improves diagnostic accuracy and clinical decision-making. In the clinical study underpinning the FDA clearance, all readers in the study demonstrated a statistically significant improvement in diagnosing lung nodules when using the system, which consistently outperformed conventional risk prediction models. In an independent validation study, the AI correctly reclassified indeterminate nodules into high- and low-risk categories in more than a third of cancers and benign nodules, demonstrating the potential to speed up lung cancer diagnosis and reduce invasive biopsies and surgeries on patients without lung cancer.
Today’s AI systems utilize deep learning systems analyze real-world tumor examples and evaluate which constitute a problematic one, rather than looking for pre-defined features. The AI system is given a large data set of lung CT scan images, some with cancer and some without, and the machine learns what a lung cancer nodule looks in extraordinary detail. These systems can help radiologist overcome the limits of human vision, which can miss tiny malignant lesions that signify cancerous growth. Currently, up to 35% of lung nodules go unnoticed during initial screenings and AI systems can both discover hard to find lung spots as well as reduce the burden of overwhelming caseloads on busy specialists.
These developments promise to make lung-cancer screenings more precise and accessible to all, with AI should be leading the way for all.
FDA clearance 510k, pulmonologists, radiology, Lung Cancer Prediction, lung cancer treatment, artificial intelligence, AI, machine learning, lung cancer diagnosis, Virtual Nodule Clinic.