InferRead CT Lung

Efficacy

1. Retrospective observational study (OCEBM 3). retrospective observational study focusing on the development and evaluation of an Intelligent Imaging Layout System (IILS) for lung nodule detection and classification using chest CT scans. High accuracy in detecting lung nodules. It had an Area Under the Curve (AUC) of 90.6% for distinguishing between benign and malignant nodules. Sensitivity (the ability to correctly identify malignant nodules) was 76.5%, and specificity (the ability to correctly identify benign nodules) was 89.1%. (The Lancet)
   ‍
2. Deep learning model vs radiologists on diagnosing pneumoconiosis from images from two centres. Assessed performance via AUC, accuracy, recall, and F1 score. The accuracy of the MLANet model for pneumoconiosis diagnosis on the internal test set, external validation set, and prospective test set reached 97.87%, 98.03%, and 95.40%, respectively, which was close to the level of qualified radiologists. Moreover, the model can effectively screen stage I pneumoconiosis with an accuracy of 97.16%, a recall of 98.25, a precision of 93.42%, and an F1 score of 95.59%, respectively. (Springer)
   ‍
3. Retrospective observational study (OCEBM 3), deep learning model based on 3D CNNs for predicting specific histological patterns (micropapillary or solid) in invasive lung adenocarcinoma (ILADC) CT scans - 617 patients separated 4:1 training:validation. Model 1 (predicting the micropapillary/solid (M/S) pattern in ILADC) achieved: AUC (Area Under the Curve) of 0.924 in the training cohort, 0.807 in the internal validation cohort, and 0.857 in the external validation cohort. Accuracy, precision, recall, and F1-scores were similarly high across all cohorts, indicating strong predictive performance. Model 2 (predicting the M/S pattern in tumors smaller than 2 cm) also performed well, with AUCs of 0.946 in the training cohort, 0.869 in internal validation, and 0.831 in external validation. (Springer)
   ‍
4. Retrospective cohort study (OCEBM 3). NC-CT and CTPA scans to detect PE, 178 cases split into 133 for training, 45 for testing. The AUC (Area Under the Curve) scores were: 0.857 in the testing set, 0.810 in the validation set, outperforming radiologists and the YEARS algorithm. The machine learning model achieved higher sensitivity (87.1%) and specificity (75.0%) compared to the radiologists and the YEARS algorithm, especially in identifying acute pulmonary thromboembolism (APE) from NC-CT scans. The AUC scores for the radiologists ranged from 0.504 to 0.527, while the YEARS algorithm had an AUC of 0.618. (AME groups)