Ability of Radiomics in Differentiation of Anaplastic Oligodendroglioma From Atypical Low-Grade Oligodendroglioma Using Machine-Learning Approach

Abstract

Objectives: To investigate the ability of radiomics features from MRI in differentiating anaplastic oligodendroglioma (AO) from atypical low-grade oligodendroglioma using machine-learning algorithms. Methods: A total number of 101 qualified patients (50 participants with AO and 51 with atypical low-grade oligodendroglioma) were enrolled in this retrospective, single-center study. Forty radiomics features of tumor images derived from six matrices were extracted from contrast-enhanced T1-weighted (T1C) images and fluid-attenuation inversion recovery (FLAIR) images. Three selection methods were performed to select the optimal features for classifiers, including distance correlation, least absolute shrinkage and selection operator (LASSO), and gradient boosting decision tree (GBDT). Then three machine-learning classifiers were adopted to generate discriminative models, including linear discriminant analysis, support vector machine, and random forest (RF). Receiver operating characteristic analysis was conducted to evaluate the discriminative performance of each model. Results: Nine predictive models were established based on radiomics features from T1C images and FLAIR images. All of the classifiers represented feasible ability in differentiation, with AUC more than 0.840 when combined with suitable selection method. For models based on T1C images, the combination of LASSO and RF classifier represented the highest AUC of 0.904 in the validation group. For models based on FLAIR images, the combination of GBDT and RF classifier showed the highest AUC of 0.861 in the validation group. Conclusion: Radiomics-based machine-learning approach could potentially serve as a feasible method in distinguishing AO from atypical low-grade oligodendroglioma.

Differentiation of Brain Abscess From Cystic Glioma Using Conventional MRI Based on Deep Transfer Learning Features and Hand-Crafted Radiomics Features

“…The RQS of our study was satisfactory at 15 points (41.7% of the ideal quality score), and the detailed result is listed in 7. The RQSs of the relevant work (16)(17)(18)(19)(20) were analyzed in our study, but only our study is open to science and data, only one study conducted a multivariable analysis with non-radiomics features (20), and only one study based on multicenter validation (16). Besides, no research has conducted a phantom study, collected images of individuals at additional time points, discussed biological correlates, conducted a prospective study, or reported the cost-effectiveness of the clinical application.…”

“…It has great potential for oncology practice, including differential diagnosis, prediction of pathological classification, lymph node metastasis, and survival (6)(7)(8)(9)(10). Radiomics has been applied to brain tumor diseases (11)(12)(13)(14)(15), especially in differentiating brain tumors (16)(17)(18)(19)(20)(21). For example, Qian et al investigated the ability of radiomic analysis to distinguish between isolated brain metastases and glioblastoma (16); Dong et al used the radiomic features derived from the areas of peripheral enhancing edema to differentiate glioblastoma from supratentorial single brain metastasis (17); Zhang et al investigated the feasibility of contrast-enhanced T1WI radiomics features extracted by machine-learning algorithms to distinguish between low-grade oligodendroglioma and atypical anaplastic oligodendroglioma (18); Chen et al applied radiomics analysis to distinguish between metastatic brain tumors and glioblastomas based on contrast-enhanced T1WI, and they validated the discriminative performance of this method (19); Artzi et al used radiomics-based machine learning to differentiate between brain metastasis subtypes and glioblastoma based on conventional postcontrast T1WI (20).…”

Employment of Artificial Intelligence Based on Routine Laboratory Results for the Early Diagnosis of Multiple Myeloma

“…Ni et al achieved a general accuracy of 84.48% when using radionics analysis based on the LASSO + GBDT method for the noninvasive diagnosis of microvascular invasion in hepatocellular carcinoma [34]. Zhang et al used LASSO + GBDT to examine the ability of radionics characteristics from MRI in differentiating anaplastic ol igoden drogl ioma (AO) f ro m a ty pi cal l ow-grade oligodendroglioma (35). The majority of researchers have performed quantitative analysis of multi-modality image data for diagnosis and prognosis by using artificial intelligence methods, but few of them have focused on the routine laboratory tests that easily obtained from clinic.…”

Detailed identification of epidermal growth factor receptor mutations in lung adenocarcinoma: Combining radiomics with machine learning

“…21,[37][38][39][40] In some other studies, radiomics and ML methods were combined to analyze the imaging phenotype of tumors with other cancers, which showed good results. [41][42][43][44][45] Here, we used 5 ML methods to identify the Medical Physics, 47 (8), August 2020 differences in imaging phenotypes in various states of EGFR, and determined the best-performing method. Among the above-mentioned studies of lung adenocarcinoma EGFR mutations, only Rios et al used data from different institutions, revealing an AUC of 0.75 and accuracy of 0.65.…”