2022 World Conference on Lung Cancer (ePosters)-EP13.01-011. Combining Automated Malignancy Risk Estimation with Lung Nodule Detection May Reduce Physician Effort and Increase Diagnostic Accuracy

EP13.01-011. Combining Automated Malignancy Risk Estimation with Lung Nodule Detection May Reduce Physician Effort and Increase Diagnostic Accuracy

Back to course

Pdf Summary

A study conducted by RevealDx and contextﬂow GmbH explores the potential benefits of combining automated lung nodule detection with computer-aided diagnostic tools to assess malignancy risk. The aim of the study was to analyze the performance of these tools when combined in a small test dataset. Lung nodules, frequently found on chest CT scans, can be indicators of early lung cancer diagnosis but require significant physician time and healthcare cost for detection, diagnosis, and management. The study suggests that the combination of automated detection with accelerated review of low-risk nodules could save physician time and lead to early cancer diagnosis and reduced follow-up procedures for benign nodules.<br /><br />The study used a random sample of CT scans from the U.S. NLST study, with 15 patients having benign nodules and 17 including malignant nodules. The scans were analyzed using the nodule detection functionality of contextﬂow SEARCH Lung CT to identify nodules, and then the RevealAI-Lung tool was used to assess malignancy risk. Diagnostic performance and comparisons to Lung-RADS guidelines were reported.<br /><br />Previous research has shown the potential benefits of using SEARCH Lung CT for automated retrieval of lung nodules, reducing reading time by 31%. RevealAI-Lung has demonstrated increased diagnostic sensitivity and specificity across populations. The results of this study found that combining the two tools detected all manually selected index nodules and labeled a significant number of benign nodules as low-risk. If used in conjunction with Lung-RADS, earlier diagnosis could be achieved for some patients with malignant nodules and lower-risk classification for patients with benign nodules.<br /><br />Overall, the study suggests that combining automated nodule detection with computer-aided diagnostic tools could reduce physician effort, increase diagnostic accuracy, and result in early cancer diagnosis and reduced follow-up procedures. However, it should be noted that the authors have employment and shareholder interest in RevealDx or contextﬂow, and the research was funded by these entities.

Asset Subtitle

Michael E Calhoun

Meta Tag

Speaker Michael E Calhoun

Topic Pulmonology, Radiology, and Staging

Keywords

automated lung nodule detection

computer-aided diagnostic tools

malignancy risk assessment

chest CT scans

early lung cancer diagnosis

physician time

healthcare cost

benign nodules

malignant nodules

diagnostic performance