A computation-efficient CNN system for high-quality brain tumor segmentation. (April 2022)

• Record Type: Journal Article
• Authors: Sun, Yanming; Wang, Chunyan
• Is Part Of: Biomedical signal processing and control. Volume 74(2022)
• Publication Date: 2022-04
• Subjects: Application-specific convolutional neural network (ASCNN) -- Activation function Full-ReLU -- Brain tumor segmentation -- Machine learning -- Mono-modality and cross-modality feature extraction; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2021.103475
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: Paradigm of Application-Specific CNN enabling simple CNN of high-quality processing. Custom-designed CNN for high-quality brain tumor segmentation. Full-ReLU preventing information loss and minimizing number of kernels. High-performance reliable CNN system of 108 kernels in 7 layers, 20308 parameters. Design for reproducibility and reliability of CNN systems. Abstract: In this paper, a reliable computation-efficient system of Convolutional Neural Network (CNN) is proposed for brain tumor segmentation. It consists of a segmentation-CNN, a pre-CNN block for data reduction and a refinement block. The unique CNN is custom-designed, following the proposed paradigm of ASCNN (Application Specific CNN), to perform mono-modality and cross-modality feature extractions, tumor localization and pixel classification. It features modality-wise normalization to improve the input data quality, depthwise convolution, combined with instance normalization, for the mono-modality feature extraction, bilinear upsampling for dimension expansion without introducing randomness, and weighted data addition for signal modulation. The proposed activation function Full-ReLU helps to halve the number of kernels in convolution layers of high-pass filtering without degrading processing quality. In this specific design context, the CNN is structured to have 7 convolution layers, requiring only 108 kernels and 20, 308 trainable parameters in total. The number of kernels in each layer is made just-sufficient for its task, instead of exponentially growing over the layers, with a view to a higher information density in data channels and lower randomness in network training. Extensive experiments with BRATS2018 dataset have been conducted to confirm the high-level processing quality and reproducibility of the system. The mean-dice-scores for enhancing-tumor, whole-tumor and tumor-core are 77.2%, 89.2% and 76.3%, respectively. Testing each patient case requires only 29.07G Flops, a tiny fraction of what found in literature. The simple structure and reliable high processing quality of the proposed system will facilitate its implementation and medical applications.