Computational Complexity Reduction of Neural Networks of Brain Tumor Image Segmentation by Introducing Fermi–Dirac Correction Functions
Nowadays, deep learning methods with high structural complexity and flexibility inevitably lean on the computational capability of the hardware. A platform with high-performance GPUs and large amounts of memory could support neural networks having large numbers of layers and kernels. However, naivel...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2021-02-01
|
Series: | Entropy |
Subjects: | |
Online Access: | https://www.mdpi.com/1099-4300/23/2/223 |
id |
doaj-b9cd6dfaff324429abde579daeebbcfc |
---|---|
record_format |
Article |
spelling |
doaj-b9cd6dfaff324429abde579daeebbcfc2021-02-12T00:05:57ZengMDPI AGEntropy1099-43002021-02-012322322310.3390/e23020223Computational Complexity Reduction of Neural Networks of Brain Tumor Image Segmentation by Introducing Fermi–Dirac Correction FunctionsYen-Ling Tai0Shin-Jhe Huang1Chien-Chang Chen2Henry Horng-Shing Lu3Bio-Microsystems Integration Laboratory, Department of Biomedical Sciences and Engineering, National Central University, Taoyuan City 32001, TaiwanBio-Microsystems Integration Laboratory, Department of Biomedical Sciences and Engineering, National Central University, Taoyuan City 32001, TaiwanBio-Microsystems Integration Laboratory, Department of Biomedical Sciences and Engineering, National Central University, Taoyuan City 32001, TaiwanShing-Tung Yau Center, National Chiao Tung University, Hsinchu 30010, TaiwanNowadays, deep learning methods with high structural complexity and flexibility inevitably lean on the computational capability of the hardware. A platform with high-performance GPUs and large amounts of memory could support neural networks having large numbers of layers and kernels. However, naively pursuing high-cost 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 particle-like 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 U-net for the algorithmic validation, and the proposed Fermi–Dirac correction function exhibited comparable performance to other employed preprocessing methods. By comparing to the conventional z-score normalization function and the Gamma correction function, the proposed algorithm can save at least 38% of computational time cost under a low-cost 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.https://www.mdpi.com/1099-4300/23/2/223computational complexitydimensional fusion U-netFermi–Dirac distributionimage segmentation |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Yen-Ling Tai Shin-Jhe Huang Chien-Chang Chen Henry Horng-Shing Lu |
spellingShingle |
Yen-Ling Tai Shin-Jhe Huang Chien-Chang Chen Henry Horng-Shing Lu Computational Complexity Reduction of Neural Networks of Brain Tumor Image Segmentation by Introducing Fermi–Dirac Correction Functions Entropy computational complexity dimensional fusion U-net Fermi–Dirac distribution image segmentation |
author_facet |
Yen-Ling Tai Shin-Jhe Huang Chien-Chang Chen Henry Horng-Shing Lu |
author_sort |
Yen-Ling Tai |
title |
Computational Complexity Reduction of Neural Networks of Brain Tumor Image Segmentation by Introducing Fermi–Dirac Correction Functions |
title_short |
Computational Complexity Reduction of Neural Networks of Brain Tumor Image Segmentation by Introducing Fermi–Dirac Correction Functions |
title_full |
Computational Complexity Reduction of Neural Networks of Brain Tumor Image Segmentation by Introducing Fermi–Dirac Correction Functions |
title_fullStr |
Computational Complexity Reduction of Neural Networks of Brain Tumor Image Segmentation by Introducing Fermi–Dirac Correction Functions |
title_full_unstemmed |
Computational Complexity Reduction of Neural Networks of Brain Tumor Image Segmentation by Introducing Fermi–Dirac Correction Functions |
title_sort |
computational complexity reduction of neural networks of brain tumor image segmentation by introducing fermi–dirac correction functions |
publisher |
MDPI AG |
series |
Entropy |
issn |
1099-4300 |
publishDate |
2021-02-01 |
description |
Nowadays, deep learning methods with high structural complexity and flexibility inevitably lean on the computational capability of the hardware. A platform with high-performance GPUs and large amounts of memory could support neural networks having large numbers of layers and kernels. However, naively pursuing high-cost 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 particle-like 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 U-net for the algorithmic validation, and the proposed Fermi–Dirac correction function exhibited comparable performance to other employed preprocessing methods. By comparing to the conventional z-score normalization function and the Gamma correction function, the proposed algorithm can save at least 38% of computational time cost under a low-cost 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. |
topic |
computational complexity dimensional fusion U-net Fermi–Dirac distribution image segmentation |
url |
https://www.mdpi.com/1099-4300/23/2/223 |
work_keys_str_mv |
AT yenlingtai computationalcomplexityreductionofneuralnetworksofbraintumorimagesegmentationbyintroducingfermidiraccorrectionfunctions AT shinjhehuang computationalcomplexityreductionofneuralnetworksofbraintumorimagesegmentationbyintroducingfermidiraccorrectionfunctions AT chienchangchen computationalcomplexityreductionofneuralnetworksofbraintumorimagesegmentationbyintroducingfermidiraccorrectionfunctions AT henryhorngshinglu computationalcomplexityreductionofneuralnetworksofbraintumorimagesegmentationbyintroducingfermidiraccorrectionfunctions |
_version_ |
1724273867963236352 |