Optellum’s lung cancer prediction AI is clinically validated in the top-ranked US respiratory journal

Global leading doctors from Vanderbilt and Oxford have published independent clinical validation of the Optellum Lung Cancer Prediction AI, illustrating its potential to shorten time to diagnosis.

The paper, with commentary by Prof. Pierre Massion MD, was published in the American Journal of Respiratory and Critical Care Medicine (colloquially known as the ATS Blue Journal), ranked first among respiratory journals in the US for impact1.


Prof. Massion and the team from the Vanderbilt University Medical Center and Oxford University Hospitals, have independently validated the use of the Lung Cancer Prediction Convolutional Neural Network (LCP-CNN) to help physicians diagnose the malignancy of indeterminate pulmonary nodules (IPNs). These are small lesions frequently detected in Computed Tomography (CT) scans that may be caused by benign processes or lung tumours in their early and most easily treatable stages.

The Blue Journal paper evaluates the LCP-CNN performance on three datasets. First, an internal validation was performed on all indeterminate lung nodules from the US National Lung Cancer Screening (NLST) trial (15,693 nodule scans from 6547 patients, including 932 malignant nodules). Next, the model was independently externally validated on two clinical datasets of incidentally detected pulmonary nodules in regular clinical care: a prospectively collected dataset from Vanderbilt University Medical Center (nodule scans from 116 patients, including 64 malignant nodules) in the US and retrospectively collected from Oxford University Hospitals NHS Foundation Trust in the UK (463 nodule scans from 427 patients, including 63 malignant nodules). 

The study compared the discrimination performance of the LCP-CNN to the Mayo model; a conventional clinical risk model published in the literature and recommended in current lung nodule management guidelines. It also evaluate the impact on patient stratification through reclassification analysis using rule-in and rule-out thresholds.

AI as a rule-in test
(at ACCP threshold of >65%) AI
as a rule-out test
(at ACCP threshold of <5%)
Vanderbilt34%33%
Oxford30%58%

These findings indicate that, with the Optellum model, a third of cancerous nodules could be diagnosed and potentially treated earlier. Additionally up to 58% of benign nodules could be diagnosed earlier, avoiding distress and potentially harmful procedures for those patients.

Pierre Massion MD
(Joe Howell/Vanderbilt University)

“These results suggest the potential clinical utility of this deep learning algorithm to revise the probability of cancer among IPNs aiming to decrease invasive procedures and shorten time to diagnosis.”

Pierre Massion, MD, Cornelius Vanderbilt Chair in Medicine at Vanderbilt University, the study’s lead author.

Further reading: Read the paper in the American Journal of Respiratory and Critical Care Medicine (also known as ATS Blue Journal):

Related links:
Aunt Minnie news story
Vanderbilt news story
Becker’s Hospital Review news story

Footnotes

  1. Source: https://www.scimagojr.com/journalrank.php?category=2740
  2. American College of Chest Physician guidelines