Median Technologies · EURONEXT:ALMDT
Median Technologies — Technical Presentation (2022)
10 slides from Median Technologies's technical presentation, 2022
Slide 1: Development of an AI/ML Tech Based Pulmonary Nodule Malignancy Prediction Model
Development of an AI/ML Tech Based Pulmonary Nodule Malignancy Prediction Model: Application to Indeterminate Pulmonary Nodules (IPN)* and Early-Stage Cancers. RSNA 2022 Charles VOYTON, iBiopsy® Pierre BAUDOT, Van-Khoa LE, Danny FRANCIS, Elias MUNOZ, Yan LIU, Benoit HUET And the entire iBiopsy® team
Slide 2: Disclosures
Disclosures All Authors are employed by: Median Technologies, France 2 www.mediantechnologies.com @ RSNA 2022
Slide 3: Lung Cancer's Global Impact
Lung Cancer’s Global Impact The Need for Comprehensive Screening • Mortality: Lung cancer is the leading cancer killer worldwide, accounting for 18% of all cancer deaths – A total of 1,8M deaths in 2020 (2.4M in 2030) – Low 5-year overall survival rate 18% • Incidence: 2nd most commonly diagnosed ca
Slide 4: Indeterminate Pulmonary Nodules (IPN)
Indeterminate Pulmonary Nodules (IPN) An Unmet Medical Need and Hurdle to LCS Implementation › IPNs are nodules 6-15mm are that require follow up to prove benign status of the nodule according to risk management guidelines. › In the NLST, more than 96% of all positive scans represented a false posit
Slide 5: Lung Screening Dataset – NLST Trial
Lung Screening Dataset – NLST Trial Key features: large dataset, high quality ground truth, early stages NLST Dataset et CT-Arm Selection for Download 26,722 • CT-Arm Patients • Cancer: All Screening Detected Cancer NLST sub-set • Benign: all available with nodules and all imaging time points › Larg
Slide 6: CADx Classifier Model
CADx Classifier Model Key features: ensembling of very different deep Convolutional Neural Networks This method is proprietary and filed under European patent application N° 22315156.4 of July 15, 2022, as “METHOD AND SYSTEM FOR COMPUTER AIDED DIAGNOSIS USING ENSEMBLED 2D AND 3D NEURAL NETWORKS BASE
Slide 7: LCS Nodule Characterization Results (CADx)
LCS Nodule Characterization Results (CADx) Automatic lung nodule characterization on all nodule sizes › AUC-ROC on test = 0.988, Operating Point (OP) : Sensitivity = 90.0%, Specificity = 97.8% • (5-folds AUC: 0.988+/-0.007 stddev) Receiver Operating Characteristic Test Distribution of Malignancy pre
Slide 8: LCS Nodule Characterization Results (CADx)
LCS Nodule Characterization Results (CADx) A closer look at IPNs and stage 1 cancer Distribution of nodules’ diameter Distribution of cancer stages Stage I Cancers IPNs 6-15mm Number of lesions Number of lesions 6 15 6 15 Nodule diameter (mm) Cancer stage Cohort Benign Cancer Total AUC Sensitivity S
Slide 9: Conclusions and Takeaway Message
Conclusions and Takeaway Message AI, IPNs, and Lung Cancer Screening › The model demonstrates robust performance for the characterization of pulmonary nodules detected by radiologists. ▪ AUC 0.988+/-0.007, 95.3% sensitivity at a specificity of 95.6% › When focused on indeterminate nodules performanc
Slide 10: RSNA 2022
AXSNA 2022 + aE - + aii oF + z a = + + “8 Meeting: November 27 to December 1 Technical Exhibits: November 27 - 30 i“ Empowering Patients and Partners in Care Pe