RESUMO
OBJECTIVES: Nasal polyp (NP) and inverted papilloma (IP) are two common types of nasal masses. And their differentiation is essential for determining optimal surgical strategies and predicting outcomes. Thus, we aimed to develop several radiomic models to differentiate them based on computed tomography (CT)-extracted radiomic features. METHODS: A total of 296 patients with nasal polyps or papillomas were enrolled in our study. Radiomics features were extracted from non-contrast CT images. For feature selection, three methods including Boruta, random forest, and correlation coefficient were used. We choose three models, namely SVM, naive Bayes, and XGBoost, to perform binary classification on the selected features. And the data was validated with tenfold cross-validation. Then, the performance was assessed by receiver operator characteristic (ROC) curve and related parameters. RESULTS: In this study, the performance ability of the models was in the following order: XGBoost > SVM > Naive Bayes. And the XGBoost model showed excellent AUC performance at 0.922, 0.9078, 0.9184, and 0.9141 under four conditions (no feature selection, Boruta, random forest, and correlation coefficient). CONCLUSIONS: We demonstrated that CT-based radiomics plays a crucial role in distinguishing IP from NP. It can provide added diagnostic value by distinguishing benign nasal lesions and reducing the need for invasive diagnostic procedures and may play a vital role in guiding personalized treatment strategies and developing optimal therapies. CRITICAL RELEVANCE STATEMENT: Based on the extraction of radiomic features of tumor regions from non-contrast CT, optimized by radiomics to achieve non-invasive classification of IP and NP which provide support for respective therapy of IP and NP. KEY POINTS: ⢠CT images are commonly used to diagnose IP and NP. ⢠Radiomics excels in feature extraction and analysis. ⢠CT-based radiomics can be applied to distinguish IP from NP. ⢠Use multiple feature selection methods and classifier models. ⢠Derived from real clinical cases with abundant data.
RESUMO
Machine learning and pattern recognition have been widely investigated in order to look for the biomarkers of Alzheimer's disease (AD). However, most existing methods extract features by seed-based correlation, which not only requires prior information but also ignores the relationship between resting state functional magnetic resonance imaging (rs-fMRI) voxels. In this study, we proposed a deep learning classification framework with multivariate data-driven based feature extraction for automatic diagnosis of AD. Specifically, a three-level hierarchical partner matching independent components analysis (3LHPM-ICA) approach was proposed first in order to address the issues in spatial individual ICA, including the uncertainty of the numbers of components, the randomness of initial values, and the correspondence of ICs of multiple subjects, resulting in stable and reliable ICs which were applied as the intrinsic brain functional connectivity (FC) features. Second, Granger causality (GC) was utilized to infer directional interaction between the ICs that were identified by the 3LHPM-ICA method and extract the effective connectivity features. Finally, a deep learning classification framework was developed to distinguish AD from controls by fusing the functional and effective connectivities. A resting state fMRI dataset containing 34 AD patients and 34 normal controls (NCs) was applied to the multivariate deep learning platform, leading to a classification accuracy of 95.59%, with a sensitivity of 97.06% and a specificity of 94.12% with leave-one-out cross validation (LOOCV). The experimental results demonstrated that the measures of neural connectivities of ICA and GC followed by deep learning classification represented the most powerful methods of distinguishing AD clinical data from NCs, and these aberrant brain connectivities might serve as robust brain biomarkers for AD. This approach also allows for expansion of the methodology to classify other psychiatric disorders.