The effect of image resolution on convolutional neural networks in breast ultrasound
Purpose: The objective of this research was to investigate the efficacy of various parameter combinations of Convolutional Neural Networks (CNNs) models, namely MobileNet and DenseNet121, and different input image resolutions (REZs) ranging from 64×64 to 512×512 pixels, for diagnosing breast cancer....
| Main Authors: | , , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2023-08-01
|
| Series: | Heliyon |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2405844023064617 |
| _version_ | 1797732857602375680 |
|---|---|
| author | Shuzhen Tang Chen Jing Yitao Jiang Keen Yang Zhibin Huang Huaiyu Wu Chen Cui Siyuan Shi Xiuqin Ye Hongtian Tian Di Song Jinfeng Xu Fajin Dong |
| author_facet | Shuzhen Tang Chen Jing Yitao Jiang Keen Yang Zhibin Huang Huaiyu Wu Chen Cui Siyuan Shi Xiuqin Ye Hongtian Tian Di Song Jinfeng Xu Fajin Dong |
| author_sort | Shuzhen Tang |
| collection | DOAJ |
| description | Purpose: The objective of this research was to investigate the efficacy of various parameter combinations of Convolutional Neural Networks (CNNs) models, namely MobileNet and DenseNet121, and different input image resolutions (REZs) ranging from 64×64 to 512×512 pixels, for diagnosing breast cancer. Materials and methods: During the period of June 2015 to November 2020, two hospitals were involved in the collection of two-dimensional ultrasound breast images for this retrospective multicenter study. The diagnostic performance of the computer models MobileNet and DenseNet 121 was compared at different resolutions. Results: The results showed that MobileNet had the best breast cancer diagnosis performance at 320×320pixel REZ and DenseNet121 had the best breast cancer diagnosis performance at 448×448pixel REZ. Conclusion: Our study reveals a significant correlation between image resolution and breast cancer diagnosis accuracy. Through the comparison of MobileNet and DenseNet121, it is highlighted that lightweight neural networks (LW-CNNs) can achieve model performance similar to or even slightly better than large neural networks models (HW-CNNs) in ultrasound images, and LW-CNNs' prediction time per image is lower. |
| first_indexed | 2024-03-12T12:20:35Z |
| format | Article |
| id | doaj.art-62e19865826f4497868e2783fb687e7a |
| institution | Directory Open Access Journal |
| issn | 2405-8440 |
| language | English |
| last_indexed | 2024-03-12T12:20:35Z |
| publishDate | 2023-08-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Heliyon |
| spelling | doaj.art-62e19865826f4497868e2783fb687e7a2023-08-30T05:54:05ZengElsevierHeliyon2405-84402023-08-0198e19253The effect of image resolution on convolutional neural networks in breast ultrasoundShuzhen Tang0Chen Jing1Yitao Jiang2Keen Yang3Zhibin Huang4Huaiyu Wu5Chen Cui6Siyuan Shi7Xiuqin Ye8Hongtian Tian9Di Song10Jinfeng Xu11Fajin Dong12Second Clinical College of Jinan University, Shenzhen 518020, Guangdong, ChinaSecond Clinical College of Jinan University, Shenzhen 518020, Guangdong, China; Shenzhen People's Hospital, Shenzhen 518020, Guangdong, ChinaResearch and Development Department, Illuminate, LLC, Shenzhen, Guangdong 518000, ChinaSecond Clinical College of Jinan University, Shenzhen 518020, Guangdong, China; Shenzhen People's Hospital, Shenzhen 518020, Guangdong, ChinaSecond Clinical College of Jinan University, Shenzhen 518020, Guangdong, China; Shenzhen People's Hospital, Shenzhen 518020, Guangdong, ChinaSecond Clinical College of Jinan University, Shenzhen 518020, Guangdong, China; Shenzhen People's Hospital, Shenzhen 518020, Guangdong, ChinaResearch and Development Department, Illuminate, LLC, Shenzhen, Guangdong 518000, ChinaResearch and Development Department, Illuminate, LLC, Shenzhen, Guangdong 518000, ChinaSecond Clinical College of Jinan University, Shenzhen 518020, Guangdong, China; Shenzhen People's Hospital, Shenzhen 518020, Guangdong, ChinaSecond Clinical College of Jinan University, Shenzhen 518020, Guangdong, China; Shenzhen People's Hospital, Shenzhen 518020, Guangdong, ChinaSecond Clinical College of Jinan University, Shenzhen 518020, Guangdong, China; Shenzhen People's Hospital, Shenzhen 518020, Guangdong, ChinaSecond Clinical College of Jinan University, Shenzhen 518020, Guangdong, China; Corresponding author.Second Clinical College of Jinan University, Shenzhen 518020, Guangdong, China; Corresponding author.Purpose: The objective of this research was to investigate the efficacy of various parameter combinations of Convolutional Neural Networks (CNNs) models, namely MobileNet and DenseNet121, and different input image resolutions (REZs) ranging from 64×64 to 512×512 pixels, for diagnosing breast cancer. Materials and methods: During the period of June 2015 to November 2020, two hospitals were involved in the collection of two-dimensional ultrasound breast images for this retrospective multicenter study. The diagnostic performance of the computer models MobileNet and DenseNet 121 was compared at different resolutions. Results: The results showed that MobileNet had the best breast cancer diagnosis performance at 320×320pixel REZ and DenseNet121 had the best breast cancer diagnosis performance at 448×448pixel REZ. Conclusion: Our study reveals a significant correlation between image resolution and breast cancer diagnosis accuracy. Through the comparison of MobileNet and DenseNet121, it is highlighted that lightweight neural networks (LW-CNNs) can achieve model performance similar to or even slightly better than large neural networks models (HW-CNNs) in ultrasound images, and LW-CNNs' prediction time per image is lower.http://www.sciencedirect.com/science/article/pii/S2405844023064617Artificial intelligenceConvolutional neural networksBreast cancerUltrasoundImage resolution |
| spellingShingle | Shuzhen Tang Chen Jing Yitao Jiang Keen Yang Zhibin Huang Huaiyu Wu Chen Cui Siyuan Shi Xiuqin Ye Hongtian Tian Di Song Jinfeng Xu Fajin Dong The effect of image resolution on convolutional neural networks in breast ultrasound Heliyon Artificial intelligence Convolutional neural networks Breast cancer Ultrasound Image resolution |
| title | The effect of image resolution on convolutional neural networks in breast ultrasound |
| title_full | The effect of image resolution on convolutional neural networks in breast ultrasound |
| title_fullStr | The effect of image resolution on convolutional neural networks in breast ultrasound |
| title_full_unstemmed | The effect of image resolution on convolutional neural networks in breast ultrasound |
| title_short | The effect of image resolution on convolutional neural networks in breast ultrasound |
| title_sort | effect of image resolution on convolutional neural networks in breast ultrasound |
| topic | Artificial intelligence Convolutional neural networks Breast cancer Ultrasound Image resolution |
| url | http://www.sciencedirect.com/science/article/pii/S2405844023064617 |
| work_keys_str_mv | AT shuzhentang theeffectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT chenjing theeffectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT yitaojiang theeffectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT keenyang theeffectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT zhibinhuang theeffectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT huaiyuwu theeffectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT chencui theeffectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT siyuanshi theeffectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT xiuqinye theeffectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT hongtiantian theeffectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT disong theeffectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT jinfengxu theeffectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT fajindong theeffectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT shuzhentang effectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT chenjing effectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT yitaojiang effectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT keenyang effectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT zhibinhuang effectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT huaiyuwu effectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT chencui effectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT siyuanshi effectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT xiuqinye effectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT hongtiantian effectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT disong effectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT jinfengxu effectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound AT fajindong effectofimageresolutiononconvolutionalneuralnetworksinbreastultrasound |