Automatic Meningioma Segmentation and Grading Prediction: A Hybrid Deep-Learning Method.

Chaoyue Chen,Xin Shu,Lei Zhang,Jianguo Xu,Wei Huang,Jianfeng Xu,Yang Zhang,Yisong Cheng,Yuen Teng,Ting Zhang,Yu Hua

doi:10.3390/jpm11080786

Abstract

The purpose of this study was to determine whether a deep-learning-based assessment system could facilitate preoperative grading of meningioma. This was a retrospective study conducted at two institutions covering 643 patients. The system, designed with a cascade network structure, was developed using deep-learning technology for automatic tumor detection, visual assessment, and grading prediction. Specifically, a modified U-Net convolutional neural network was first established to segment tumor images. Subsequently, the segmentations were introduced into rendering algorithms for spatial reconstruction and another DenseNet convolutional neural network for grading prediction. The trained models were integrated as a system, and the robustness was tested based on its performance on an external dataset from the second institution involving different magnetic resonance imaging platforms. The results showed that the segment model represented a noteworthy performance with dice coefficients of 0.920 ± 0.009 in the validation group. With accurate segmented tumor images, the rendering model delicately reconstructed the tumor body and clearly displayed the important intracranial vessels. The DenseNet model also achieved high accuracy with an area under the curve of 0.918 ± 0.006 and accuracy of 0.901 ± 0.039 when classifying tumors into low-grade and high-grade meningiomas. Moreover, the system exhibited good performance on the external validation dataset.

Highlights

Meningioma is recognized as one of the most common intracranial neoplasms, accounting for 30% of primary intracranial lesions, with an annual incidence of 5/100,000 [1,2].Based on their biological behaviors, the lesions can be categorized as either low-grade lesions, which are correlated with better survival prognosis, or highgrade lesions, which are correlated with aggressive behaviors along with increased recurrence risk following treatment [2,3,4]
Study, we we proposed proposed aa meningioma meningioma assessment assessmentsystem systembased basedon on routine routine MR
Stable, as the were were trained with the largest sample size sosize far and theand system ture stable, asmodels the models trained with the largest sample so far the was tested on an external dataset

Summary

Introduction

Meningioma is recognized as one of the most common intracranial neoplasms, accounting for 30% of primary intracranial lesions, with an annual incidence of 5/100,000 [1,2]. Based on their biological behaviors, the lesions can be categorized as either low-grade lesions (grade I meningioma), which are correlated with better survival prognosis, or highgrade lesions (grade II and grade III meningiomas), which are correlated with aggressive behaviors along with increased recurrence risk following treatment [2,3,4]. The image patterns of tumors on MRIs are usually described as solitary round tissues that are in close contact with the dura mater, with apparent/homogeneous enhancement after gadolinium injection. The consensual radiological criteria that can be applied to distinguish meningiomas of different grades accurately are not yet clear [5]

Objectives

Methods

Results

Conclusion