English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Open

Presents the pdf analysis as premium feature

PDF Analysis

Insights

Presents the summarisation as premium feature

Summarisation

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

The work presented in this paper is to propose a reliable high-quality systemof Convolutional Neural Network (CNN) for brain tumor segmentation with a lowcomputation requirement. The system consists of a CNN for the main processingfor the segmentation, a pre-CNN block for data reduction and post-CNNrefinement block. The unique CNN consists of 7 convolution layers involvingonly 108 kernels and 20308 trainable parameters. It is custom-designed,following the proposed paradigm of ASCNN (application specific CNN), to performmono-modality and cross-modality feature extraction, tumor localization andpixel classification. Each layer fits the task assigned to it, by means of (i)appropriate normalization applied to its input data, (ii) correct convolutionmodes for the assigned task, and (iii) suitable nonlinear transformation tooptimize the convolution results. In this specific design context, the numberof kernels in each of the 7 layers is made to be just-sufficient for its task,instead of exponentially growing over the layers, to increase informationdensity and to reduce randomness in the processing. The proposed activationfunction Full-ReLU helps to halve the number of kernels in convolution layersof high-pass filtering without degrading processing quality. A large number ofexperiments with BRATS2018 dataset have been conducted to measure theprocessing quality and reproducibility of the proposed system. The resultsdemonstrate that the system reproduces reliably almost the same output to thesame input after retraining. The mean dice scores for enhancing tumor, wholetumor and tumor core are 77.2%, 89.2% and 76.3%, respectively. The simplestructure and reliable high processing quality of the proposed system willfacilitate its implementation and medical applications.

A Computation-Efficient CNN System for High-Quality Brain Tumor Segmentation