Automated Interstitial Lung Abnormality Probability Prediction at CT: A Stepwise Machine Learning Approach in the Boston Lung Cancer Study.

Background It is increasingly recognized that interstitial lung abnormalities (ILAs) detected at CT have potential clinical implications, but automated identification of ILAs has not yet been fully established. Purpose To develop and test automated ILA probability prediction models using machine learning techniques on CT images. Materials and Methods This secondary analysis of a retrospective study included CT scans from patients in the Boston Lung Cancer Study collected between February 2004 and June 2017. Visual assessment of ILAs by two radiologists and a pulmonologist served as the ground truth. Automated ILA probability prediction models were developed that used a stepwise approach involving section inference and case inference models. The section inference model produced an ILA probability for each CT section, and the case inference model integrated these probabilities to generate the case-level ILA probability. For indeterminate sections and cases, both two- and three-label methods were evaluated. For the case inference model, we tested three machine learning classifiers (support vector machine [SVM], random forest [RF], and convolutional neural network [CNN]). Receiver operating characteristic analysis was performed to calculate the area under the receiver operating characteristic curve (AUC). Results A total of 1382 CT scans (mean patient age, 67 years ± 11 [SD]; 759 women) were included. Of the 1382 CT scans, 104 (8%) were assessed as having ILA, 492 (36%) as indeterminate for ILA, and 786 (57%) as without ILA according to ground-truth labeling. The cohort was divided into a training set (n = 96; ILA, n = 48), a validation set (n = 24; ILA, n = 12), and a test set (n = 1262; ILA, n = 44). Among the models evaluated (two- and three-label section inference models; two- and three-label SVM, RF, and CNN case inference models), the model using the three-label method in the section inference model and the two-label method and RF in the case inference model achieved the highest AUC, at 0.87. Conclusion The model demonstrated substantial performance in estimating ILA probability, indicating its potential utility in clinical settings. © RSNA, 2024 Supplemental material is available for this article. See also the editorial by Zagurovskaya in this issue.

Automatic inference model construction for computer-aided diagnosis of lung nodule: Explanation adequacy, inference accuracy, and experts' knowledge.

We aimed to describe the development of an inference model for computer-aided diagnosis of lung nodules that could provide valid reasoning for any inferences, thereby improving the interpretability and performance of the system. An automatic construction method was used that considered explanation adequacy and inference accuracy. In addition, we evaluated the usefulness of prior experts' (radiologists') knowledge while constructing the models. In total, 179 patients with lung nodules were included and divided into 79 and 100 cases for training and test data, respectively. F-measure and accuracy were used to assess explanation adequacy and inference accuracy, respectively. For F-measure, reasons were defined as proper subsets of Evidence that had a strong influence on the inference result. The inference models were automatically constructed using the Bayesian network and Markov chain Monte Carlo methods, selecting only those models that met the predefined criteria. During model constructions, we examined the effect of including radiologist's knowledge in the initial Bayesian network models. Performance of the best models in terms of F-measure, accuracy, and evaluation metric were as follows: 0.411, 72.0%, and 0.566, respectively, with prior knowledge, and 0.274, 65.0%, and 0.462, respectively, without prior knowledge. The best models with prior knowledge were then subjectively and independently evaluated by two radiologists using a 5-point scale, with 5, 3, and 1 representing beneficial, appropriate, and detrimental, respectively. The average scores by the two radiologists were 3.97 and 3.76 for the test data, indicating that the proposed computer-aided diagnosis system was acceptable to them. In conclusion, the proposed method incorporating radiologists' knowledge could help in eliminating radiologists' distrust of computer-aided diagnosis and improving its performance.

A study of computer-aided diagnosis for pulmonary nodule: comparison between classification accuracies using calculated image features and imaging findings annotated by radiologists.

PURPOSE: In our previous study, we developed a computer-aided diagnosis (CADx) system using imaging findings annotated by radiologists. The system, however, requires radiologists to input many imaging findings. In order to reduce such an interaction of radiologists, we further developed a CADx system using derived imaging findings based on calculated image features, in which the system only requires few user operations. The purpose of this study is to check whether calculated image features (CFT) or derived imaging findings (DFD) can represent information in imaging findings annotated by radiologists (AFD). METHODS: We calculate 2282 image features and derive 39 imaging findings by using information on a nodule position and its type (solid or ground-glass). These image features are categorized into shape features, texture features and imaging findings-specific features. Each imaging finding is derived based on each corresponding classifier using random forest. To check whether CFT or DFD can represent information in AFD, under an assumption that the accuracies of classifiers are the same if information included in input is the same, we constructed classifiers by using various types of information (CTT, DFD and AFD) and compared accuracies on an inferred diagnosis of a nodule. We employ SVM with RBF kernel as classifier to infer a diagnosis name. RESULTS: Accuracies of classifiers using DFD, CFT, AFD and CFT [Formula: see text] AFD were 0.613, 0.577, 0.773 and 0.790, respectively. Concordance rates between DFD and AFD of shape findings, texture findings and surrounding findings were 0.644, 0.871 and 0.768, respectively. CONCLUSIONS: The results suggest that CFT and AFD are similar information and CFT represent only a portion of AFD. Particularly, CFT did not contain shape information in AFD. In order to decrease an interaction of radiologists, a development of a method which overcomes these problems is necessary.