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

Yifan Wang,Linlin Bo,Chao Yang,Zijian Zhang,Zekun Jiang,Pu Huang,Zhixiong Liu,Xiao Tan,Dengwang Li,Gang Yu,Quan Cheng,Tingyin Chen

doi:10.3389/fmed.2021.748144

Abstract

Objectives: To develop and validate the model for distinguishing brain abscess from cystic glioma by combining deep transfer learning (DTL) features and hand-crafted radiomics (HCR) features in conventional T1-weighted imaging (T1WI) and T2-weighted imaging (T2WI).Methods: This single-center retrospective analysis involved 188 patients with pathologically proven brain abscess (102) or cystic glioma (86). One thousand DTL and 105 HCR features were extracted from the T1WI and T2WI of the patients. Three feature selection methods and four classifiers, such as k-nearest neighbors (KNN), random forest classifier (RFC), logistic regression (LR), and support vector machine (SVM), for distinguishing brain abscess from cystic glioma were compared. The best feature combination and classifier were chosen according to the quantitative metrics including area under the curve (AUC), Youden Index, and accuracy.Results: In most cases, deep learning-based radiomics (DLR) features, i.e., DTL features combined with HCR features, contributed to a higher accuracy than HCR and DTL features alone for distinguishing brain abscesses from cystic gliomas. The AUC values of the model established, based on the DLR features in T2WI, were 0.86 (95% CI: 0.81, 0.91) in the training cohort and 0.85 (95% CI: 0.75, 0.95) in the test cohort, respectively.Conclusions: The model established with the DLR features can distinguish brain abscess from cystic glioma efficiently, providing a useful, inexpensive, convenient, and non-invasive method for differential diagnosis. This is the first time that conventional MRI radiomics is applied to identify these diseases. Also, the combination of HCR and DTL features can lead to get impressive performance.

Highlights

Brain glioma is the most common intracranial brain tumor that is extremely difficult to treat
The gold standard for distinguishing between brain abscess and cystic glioma was confirmed by pathologists through pathological examination
To extract the deep transfer learning (DTL) features, the tumor patch images were input to the pretrained convolutional neural networks (CNN) to extract 1,000 features from each MR image modality, and the extracted features were outputs from the last fully connected layer of VGG-19 and ResNet-50

Summary

Introduction

Brain glioma is the most common intracranial brain tumor that is extremely difficult to treat. Though the treatment and prognosis of these two diseases are different, accurate and timely differential diagnosis is crucial. CT and MR images lack specificity for cystic glioma and brain abscess, especially when the medical history and clinical manifestations of the diseases cannot provide a differential diagnosis for timely treatment measures. To accurately distinguish the two diseases, previous studies have proposed advanced MR images diagnosis techniques [2, 4], such as susceptibility-weighted imaging and apparent diffusion coefficients (ADC). These diagnosis techniques cannot obtain high accuracy, and they rely on the experience of radiologists [5]. The use of the most rudimentary imaging modalities of T1-weighted imaging (T1WI) and T2-weighted imaging (T2WI) for a training model with a large sample size contributes to more universality and fewer errors

Objectives

Methods

Results

Conclusion