Abstract 1192P
Background
Detection and risk stratification of pulmonary nodules during a Lung Cancer Screening (LCS) examination is time-consuming, prone to false positives and missed detections. Leveraging Artificial Intelligence (AI) in such tasks has outperformed human readers in detection, and current risk models (Brock, MAYO) in risk prediction. Here, we present the results of an independent verification study to explore the software’s detection and characterization performance on an external dataset.
Methods
The model was trained on 10,872 patients (543 cancers) independently annotated from the NLST (National Lung Screening Trial) and LIDC (Lung Image Database Consortium) cohorts. 264 patients meeting the USPSTF criteria (age 50-80, Smoker) were collected from the EU (26%) and the USA (74%) for independent verification. The dataset comprised 88 cancer patients and 176 benign, with average size for all nodules of 6.2 ± 3.2 mm, and 17.5 ± 5.8 mm for cancerous nodules. To generate reference standards, two radiologists independently annotated each patient, with a third acting as an adjudicator to arrive at a consensus for location and nodule diagnosis (histopathology or ≥12 month stability).
Results
The verification AUC for risk prediction was 0.95, with a sensitivity of 93.2% and specificity of 87.5% at the Youden index. Cancer detection sensitivity was 91.2%, with an average of 0.44 false positive detections per scan. Performances were consistent across multiple technical and clinical parameters, including CT manufacturer, kernel hardness, kernel slice thickness, patient sex, data source, and nodule solidity (see the table). Table: 1192P
Subclass type | Subclass | AUC |
Manufacturer | SIEMENS Healthineers | 0.96 |
GE Healthcare | 0.94 | |
Canon/Toshiba | 0.95 | |
Kernel | Sharp | 0.98 |
Average | 0.94 | |
Soft | 0.93 | |
Slice thickness (mm) | 0.5-0.75 | 0.96 |
1-1.25 | 0.94 | |
1.25-1.5 | 0.96 | |
Sex | Female | 0.95 |
Male | 0.95 | |
Source | EU | 0.90 |
USA | 0.97 | |
Nodule solidity | Solid | 0.96 |
Part-solid | 0.93 |
Conclusions
This study demonstrates that the AI software is robust to external data with similar performances versus the NLST test set (AUC: 0.97) and among subclasses. More accurate screening driven by AI promises to be beneficial to LCS, reducing unneeded exams, costs, and radiologist time.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
Median Technologies.
Funding
Median Technologies.
Disclosure
All authors have declared no conflicts of interest.
Resources from the same session
212P - BRGSF-HIS mice as a predictive tool for safety assessment of biologics
Presenter: Kader Thiam
Session: Poster session 09
213P - Constructing a high-definition patient-digital twin (PDT) in treatment-naïve women with advanced cancer
Presenter: Leonardo Garma
Session: Poster session 09
215P - Detection of MUTYH for the prognosis and chemotherapy responsiveness of patients with non-small cell lung cancer
Presenter: Chi Wai Wong
Session: Poster session 09
216P - β-catenin is a potential prognostic biomarker in uterine sarcoma
Presenter: Ying Cai
Session: Poster session 09
218P - Exploiting a unique glycosaminoglycan for novel pan-cancer therapies and diagnostics
Presenter: Mette Agerbæk
Session: Poster session 09
219P - The landscape and prognostic impact of germline HLA-A subtypes in patients with advanced solid cancers
Presenter: Kyrillus Shohdy
Session: Poster session 09
220P - The role of fucosyltransferase 1 (FUT1) in CRC as a putative prognostic and predictive biomarker
Presenter: Lorenz Pammer
Session: Poster session 09
221P - ANGPTL4's role in cancer: A meta analysis and bioinformatics exploration
Presenter: Osama Younis
Session: Poster session 09
222P - Artificial intelligence (AI) based prognostication from baseline computed tomography (CT) scans in a phase III advanced non-small cell lung cancer (aNSCLC) trial
Presenter: Omar Khan
Session: Poster session 09