A CADe System for Gliomas in Brain MRI using Convolutional Neural\n Networks

2018/06/20 by Banerjee, Subhasis, Sushmita Mitra, Mitra, Sushmita +5 · 6 citations
Computer Science · Neuroscience · #Advanced Neural Network Applications #Artificial intelligence #Brain Tumor Detection and Classification #Computer Vision and Pattern Recognition (cs.CV) #Computer science #Convolutional neural network #FOS: Computer and information sciences #False positive paradox #Initialization #Medical Image Segmentation Techniques #Neural Networks and Applications #Pattern recognition (psychology) #Segmentation #cs.CV

paper · pdf · doi:10.48550/arxiv.1806.07589

published in arXiv (Cornell University) (Cornell University) · The paper consists of 11 Pages, 6 Figures, 7 Tables, 56 References

arxiv created 2018/06/20 · openalex publication_date 2018/06/20 · arxiv updated 2018/06/21 · openalex created_date 2022/09/30 · openalex updated_date 2026/08/08

Abstract

Inspired by the success of Convolutional Neural Networks (CNN), we develop a\nnovel Computer Aided Detection (CADe) system using CNN for Glioblastoma\nMultiforme (GBM) detection and segmentation from multi channel MRI data. A\ntwo-stage approach first identifies the presence of GBM. This is followed by a\nGBM localization in each "abnormal" MR slice. As part of the CADe system, two\nCNN architectures viz. Classification CNN (C-CNN) and Detection CNN (D-CNN) are\nemployed. The CADe system considers MRI data consisting of four sequences\n(T₁, T_1c, T₂, and T_2FLAIR) as input, and automatically\ngenerates the bounding boxes encompassing the tumor regions in each slice which\nis deemed abnormal. Experimental results demonstrate that the proposed CADe\nsystem, when used as a preliminary step before segmentation, can allow improved\ndelineation of tumor region while reducing false positives arising in normal\nareas of the brain. The GrowCut method, employed for tumor segmentation,\ntypically requires a foreground and background seed region for initialization.\nHere the algorithm is initialized with seeds automatically generated from the\noutput of the proposed CADe system, thereby resulting in improved performance\nas compared to that using random seeds.\n

Citations

• ADADELTA: An Adaptive Learning Rate Method
• The Multimodal Brain Tumor Image Segmentation Benchmark (BRATS)
• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Gradient-based learning applied to document recognition
• N4ITK: Improved N3 Bias Correction
• Automatic Detection of Cerebral Microbleeds From MR Images via 3D Convolutional Neural Networks

Related

• Brain MRI study for glioma segmentation using convolutional neural networks and original post-processing techniques with low computational demand
• Brain Tumor Detection using Convolutional Neural Networks with Skip Connections
• Brain tumor detection using artificial convolutional neural networks
• Detection and Classification of Brain tumors Using Deep Convolutional Neural Networks
• MRI-based brain tumor detection using convolutional deep learning methods and chosen machine learning techniques