Clinical Validation
View or Download Our Validation Resources Below
Optellum’s technology has been independently validated by world-leading lung cancer experts interested in verifying our claims and quantifying benefits to their own patient populations. Below is a selection of publications from high-quality pulmonary and respiratory journals.
Journal: American Journal of Respiratory and Critical Care Medicine (also known as ATS Blue Journal)
What is it? This paper introduces Optellum’s LCP-CNN (Lung Cancer Prediction Convolutional Neural Network) and its development using US screening data. It evaluates the LCP-CNN performance on two independent clinical datasets (one US, one UK), showing better AUC (area under the curve*) and NRI (Net Reclassification Index) than conventional clinical risk models like Mayo and Brock. This study demonstrates that this deep learning algorithm can correctly reclassify IPNs (Indeterminate Pulmonary Nodules) into low- or high-risk categories in more than a third of cancers and benign nodules when compared with conventional risk models, potentially reducing the number of unnecessary invasive procedures and delays in diagnosis.
*refers to ROC (Receiver Operator Characteristic) curve
Massion PP, et al
[See Fig 1 below]
Journal: Thorax (BMJ)
What is it? This is a retrospective clinical evaluation of the LCP-CNN on three UK populations (Oxford, Leeds and Nottingham). As well as showing significantly better patient stratification than the Brock model that is used in UK guidelines (P<0.005 on AUC), this paper also shows that low-scoring nodules can be “ruled out” of further follow-up, potentially meaning that over 24% of nodules do not need to be followed up without missing any cancer nodules.
Baldwin DR, Gustafson J, Pickup L, et al
Presented at: ATS 2020
Reader Study
The LCP-CNN has been embedded in a clinical workflow to allow readers (both pulmonologists and radiologists) to evaluate lung nodules with and without the LCP-CNN using a fully crossed MRMC (multi-reader, multi-case) paradigm. This represents the next step in clinical validation, where not only is the LCP-CNN shown to be a better tool than conventional clinical risk models, as in the journal papers above, but that having a better tool actually enables clinicians to make significantly better decisions.
Pilot Study
A pilot study of 100 cases (50 malignant/50 benign) was performed. Initial reclassification analysis (5 readers) was presented at ATS 2020. The pilot study has since been completed by a full cohort of 12 readers from the US and the UK, and a full analysis of the data was compiled by its clinical lead at the University of Oxford. This work is now undergoing peer review for publication in a journal.
Full Study
A full study with 12 readers and 300 cases was performed in the summer of 2020, using respiratory medicine and radiology readers from across the United States. The results of this work are currently in preparation for publication.
Dotson TL, et al
[See Fig 2 below]
Fig 1:
Traditional clinical risk models such as the Mayo model give indeterminate scores to many incidentally-detected nodules, leaving many patients undergoing long follow-up and potentially delaying life-saving cancer treatment. In contrast, the LCP-CNN helps clarify the patient stratification, sending more patients correctly into low-risk and high-risk groups.
Independently validated on more than 16,000 indeterminate lung nodules 1, 2
- Baldwin DR, et al External validation of a convolutional neural network artificial intelligence tool to predict malignancy in pulmonary nodules Thorax 2020;75:306-312
- Massion P, et al Assessing the accuracy of a deep learning method to risk stratify indeterminate pulmonary nodules American Journal of Respiratory and Critical Care Medicine
*Up to, depending on the patient population
Fig 2:
Example of two pulmonary nodules with the same diameter for which the Optellum LCP (Lung Cancer Prediction) score helps to correctly reclassify. Based on diameter only, both nodules would have been classified in the intermediate-risk category. However, with the support of the Optellum LCP score, the nodule on the left is correctly reclassified in the high-risk category while the nodule on the right in the low-risk category.
Optellum LCP Score
INTERPRETATION
Optellum LCP score 9 = 84% risk of cancer
Optellum LCP score 1 = 0.2% risk of cancer