Chinese Optics, Volume. 15, Issue 6, 1339(2022)
Classification model based on fusion of multi-scale feature and channel feature for benign and malignant brain tumors
Figures & Tables(17)
Fig. 9. Evaluation diagram of classification results for three kinds of Nets
Table 1. Distribution of experimental datasets
View tableView in ArticleTable 1. Distribution of experimental datasets
数据集 肿瘤 类别 数据分布 总数 训练集 测试集 BraTS2017 数据集 HGG 840 210 1050 LGG 900 225 1125 BraTS2019 数据集 HGG 1035 260 1295 LGG 915 225 1140
Table 2. Evaluation of classification results on BraTS2017 before optimization
View tableView in ArticleTable 2. Evaluation of classification results on BraTS2017 before optimization
网络 ACC（%） SEN（%） SPE（%） PPV（%） NPV（%） ResNet 89.15±1.83 88.76±3.38 89.51±3.38 88.92±3.67 89.60±2.58 SENet 90.44±3.25 93.05±2.09 89.25±7.41 88.21±5.69 93.19±1.77 ResNeXt 90.34±1.14 89.23±3.36 92.53±2.18 91.85±1.96 90.22±2.57 MDCA-ResNeXt 93.19±0.35 93.05±1.67 93.33±1.69 92.91±1.54 93.54±1.36
Table 3. Evaluation of classification results on BraTS2019 before optimization
View tableView in ArticleTable 3. Evaluation of classification results on BraTS2019 before optimization
网络 ACC（%） SEN（%） SPE（%） PPV（%） NPV（%） ResNet 91.83±2.73 93.31±2.12 90.13±4.59 91.71±3.65 92.10±2.51 SENet 91.91±2.42 90.54±5.63 91.74±4.83 93.00±3.76 93.25±2.54 ResNeXt 93.57±1.50 94.23±2.02 92.80±2.99 93.85±2.31 93.35±2.15 MDCA-ResNeXt 94.10±1.40 94.38±1.67 93.78±2.33 94.76±2.12 93.60±1.59
Table 4. Evaluation of classification results on BraTS2017 after optimization
View tableView in ArticleTable 4. Evaluation of classification results on BraTS2017 after optimization
网络 ACC（%） SEN（%） SPE（%） PPV（%） NPV（%） Improved ResNet 96.87±1.49 96.76±0.71 96.98±2.90 96.84±2.97 96.98±0.64 Improved SENet 97.56±1.04 96.67±0.89 98.40±1.43 98.27±1.52 96.94±0.83 Improved ResNeXt 97.98±1.33 97.43±2.06 98.49±1.28 98.38±1.39 97.63±1.88 Improved MDCA-ResNeXt 98.11±0.41 97.43±0.26 98.76±0.91 98.66±0.97 97.63±0.22
Table 5. Evaluation of classification results on BraTS2019 after optimization
View tableView in ArticleTable 5. Evaluation of classification results on BraTS2019 after optimization
网络 ACC（%） SEN（%） SPE（%） PPV（%） NPV（%） Improved ResNet 97.03±1.95 97.31±2.02 96.62±3.04 97.20±2.55 96.91±2.23 Improved SENet 96.69±0.88 94.38±1.89 98.74±0.93 98.62±1.01 94.99±1.57 Improved ResNeXt 97.98±0.57 97.69±0.47 98.31±0.73 98.53±0.64 97.36±0.54 Improved MDCA-ResNeXt 98.72±0.31 98.62±0.64 98.85±0.51 99.00±0.44 98.41±0.73
Table 6. Comparison of classification results of advanced methods
View tableView in ArticleTable 6. Comparison of classification results of advanced methods
文献 方法 肿瘤分割 数据集 准确率(%) 文献[ 7 ]HCS+ Multi-SVNN 是 BraTs2014 93.00 文献[ 15 ]Inception V3+POS 是 BraTs2017 96.90 文献[ 16 ]VGG16+ELM 否 BraTs2017 96.90 文献[ 17 ]3D CNN+VGG19+FNN 是 BraTs2017 96.97 文献[ 8 ]FBSO 是 BraTs2018 93.85 文献[ 19 ]3D U-Net 否 BraTs2018 91.67 本文方法 Improved MDCA-ResNeXt 否 BraTs2017 98.11 否 BraTs2019 98.72
Tools
Get Citation
Copy Citation Text
Lin-qi JIANG, Chun-yu NING, Hai-tao YU. Classification model based on fusion of multi-scale feature and channel feature for benign and malignant brain tumors[J]. Chinese Optics, 2022, 15(6): 1339
Paper Information
Category: Original Article
Received: Apr. 12, 2022
Accepted: Aug. 24, 2022
Published Online: Feb. 9, 2023
The Author Email:
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20