Projects per year
Abstract
Nowadays, deep learning methods with high structural complexity and flexibility inevitably lean on the computational capability of the hardware. A platform with highperformance GPUs and large amounts of memory could support neural networks having large numbers of layers and kernels. However, naively pursuing highcost hardware would probably drag the technical development of deep learning methods. In the article, we thus establish a new preprocessing method to reduce the computational complexity of the neural networks. Inspired by the band theory of solids in physics, we map the image space into a noninteraction physical system isomorphically and then treat image voxels as particlelike clusters. Then, we reconstruct the Fermi–Dirac distribution to be a correction function for the normalization of the voxel intensity and as a filter of insignificant cluster components. The filtered clusters at the circumstance can delineate the morphological heterogeneity of the image voxels. We used the BraTS 2019 datasets and the dimensional fusion Unet for the algorithmic validation, and the proposed Fermi–Dirac correction function exhibited comparable performance to other employed preprocessing methods. By comparing to the conventional zscore normalization function and the Gamma correction function, the proposed algorithm can save at least 38% of computational time cost under a lowcost hardware architecture. Even though the correction function of global histogram equalization has the lowest computational time among the employed correction functions, the proposed Fermi–Dirac correction function exhibits better capabilities of image augmentation and segmentation.
Original language  English 

Article number  223 
Pages (fromto)  114 
Number of pages  14 
Journal  Entropy 
Volume  23 
Issue number  2 
DOIs  
State  Published  Feb 2021 
Keywords
 Computational complexity
 Dimensional fusion Unet
 Fermi–Dirac distribution
 Image segmentation
Fingerprint
Dive into the research topics of 'Computational complexity reduction of neural networks of brain tumor image segmentation by introducing fermi–dirac correction functions'. Together they form a unique fingerprint.Projects
 1 Finished

Research on Automatic Identification and Localization of Medical Imagery Morphology and Its 3Dimensional Image Reconstruction(2/3)
1/08/20 → 31/07/21
Project: Research