A multi-stage deep learning network toward multi-classification of polyps in colorectal images
Accurate classification of colorectal polyps (CRPs) is critical for the early diagnosis and treatment of colorectal cancer (CRC). This paper presents an efficient deep learning method specifically developed to enhance the accuracy of CRPs classification, thereby assisting physicians in making inform...
Saved in:
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-04-01
|
| Series: | Alexandria Engineering Journal |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S111001682500136X |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832087836814213120 |
|---|---|
| author | Shilong Chang Kun Yang Yucheng Wang Yufeng Sun Chaoyi Qi Wenlong Fan Ying Zhang Shuang Liu Wenshan Gao Jie Meng Linyan Xue |
| author_facet | Shilong Chang Kun Yang Yucheng Wang Yufeng Sun Chaoyi Qi Wenlong Fan Ying Zhang Shuang Liu Wenshan Gao Jie Meng Linyan Xue |
| author_sort | Shilong Chang |
| collection | DOAJ |
| description | Accurate classification of colorectal polyps (CRPs) is critical for the early diagnosis and treatment of colorectal cancer (CRC). This paper presents an efficient deep learning method specifically developed to enhance the accuracy of CRPs classification, thereby assisting physicians in making informed decisions. Drawing inspiration from the sequential procedure of colonoscopy, where endoscopists first locate polyps and then proceed to detailed observations and diagnoses, we developed a novel multi-stage classification network. This network cascades several convolutional neural networks (CNNs) to mimic the gradual increase in diagnostic specificity seen in clinical settings. Furthermore, we introduced a novel attention module, the Cross-Stage Weighted Attention (CSWA), designed to amplify the effectiveness of multi-stage feature fusion by focusing on the most informative features across different stages. To train and validate our proposed network, we curated a dataset consisting of 2568 white light endoscopic images. Facing a significant class imbalance, particularly in the underrepresented categories of villous and serrated adenomas, we employed Generative Adversarial Network Augmentation (GAN-Aug) to synthesize additional images, thereby ensuring a more balanced dataset for training. An assessment by six endoscopists confirmed the high realism of polyp characteristics in the images generated by GAN-Aug. Subsequent quantitative evaluation of our CSWA-enhanced multi-stage classification network on this augmented dataset achieved an accuracy of 0.832 ± 0.006. In convolution, our approach not only demonstrates a significant improvement over existing methods by effectively emulating the step-by-step diagnostic process of endoscopists, but also promises to greatly enhance early detection and treatment strategies for CRC, ultimately improving patient outcomes. |
| format | Article |
| id | doaj-art-cfa206c8a01043e78650c6294241a241 |
| institution | Kabale University |
| issn | 1110-0168 |
| language | English |
| publishDate | 2025-04-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Alexandria Engineering Journal |
| spelling | doaj-art-cfa206c8a01043e78650c6294241a2412025-02-06T05:11:12ZengElsevierAlexandria Engineering Journal1110-01682025-04-01119189200A multi-stage deep learning network toward multi-classification of polyps in colorectal imagesShilong Chang0Kun Yang1Yucheng Wang2Yufeng Sun3Chaoyi Qi4Wenlong Fan5Ying Zhang6Shuang Liu7Wenshan Gao8Jie Meng9Linyan Xue10College of Quality and Technical Supervision, Hebei University, Baoding 071002, ChinaCollege of Quality and Technical Supervision, Hebei University, Baoding 071002, China; National & Local Joint Engineering Research Center of Metrology Instrument and System, Hebei University, Baoding 071002, China; Hebei Technology Innovation Center for Lightweight of New Energy Vehicle Power System, Hebei University, Baoding 071002, China; Corresponding authors at: College of Quality and Technical Supervision, Hebei University, Baoding 071002, China.College of Quality and Technical Supervision, Hebei University, Baoding 071002, ChinaCollege of Electronic Information Engineering, Hebei University, Baoding 071002, ChinaCollege of Electronic Information Engineering, Hebei University, Baoding 071002, ChinaCollege of Quality and Technical Supervision, Hebei University, Baoding 071002, ChinaCollege of Quality and Technical Supervision, Hebei University, Baoding 071002, ChinaCollege of Quality and Technical Supervision, Hebei University, Baoding 071002, China; National & Local Joint Engineering Research Center of Metrology Instrument and System, Hebei University, Baoding 071002, China; Hebei Technology Innovation Center for Lightweight of New Energy Vehicle Power System, Hebei University, Baoding 071002, ChinaDepartment of Orthopedics, Affiliated Hospital of Hebei University, Baoding 071000, ChinaDepartment of Gastroenterology, Affiliated Hospital of Hebei University, Baoding 071000, China; Corresponding author.College of Quality and Technical Supervision, Hebei University, Baoding 071002, China; National & Local Joint Engineering Research Center of Metrology Instrument and System, Hebei University, Baoding 071002, China; Hebei Technology Innovation Center for Lightweight of New Energy Vehicle Power System, Hebei University, Baoding 071002, China; Corresponding authors at: College of Quality and Technical Supervision, Hebei University, Baoding 071002, China.Accurate classification of colorectal polyps (CRPs) is critical for the early diagnosis and treatment of colorectal cancer (CRC). This paper presents an efficient deep learning method specifically developed to enhance the accuracy of CRPs classification, thereby assisting physicians in making informed decisions. Drawing inspiration from the sequential procedure of colonoscopy, where endoscopists first locate polyps and then proceed to detailed observations and diagnoses, we developed a novel multi-stage classification network. This network cascades several convolutional neural networks (CNNs) to mimic the gradual increase in diagnostic specificity seen in clinical settings. Furthermore, we introduced a novel attention module, the Cross-Stage Weighted Attention (CSWA), designed to amplify the effectiveness of multi-stage feature fusion by focusing on the most informative features across different stages. To train and validate our proposed network, we curated a dataset consisting of 2568 white light endoscopic images. Facing a significant class imbalance, particularly in the underrepresented categories of villous and serrated adenomas, we employed Generative Adversarial Network Augmentation (GAN-Aug) to synthesize additional images, thereby ensuring a more balanced dataset for training. An assessment by six endoscopists confirmed the high realism of polyp characteristics in the images generated by GAN-Aug. Subsequent quantitative evaluation of our CSWA-enhanced multi-stage classification network on this augmented dataset achieved an accuracy of 0.832 ± 0.006. In convolution, our approach not only demonstrates a significant improvement over existing methods by effectively emulating the step-by-step diagnostic process of endoscopists, but also promises to greatly enhance early detection and treatment strategies for CRC, ultimately improving patient outcomes.http://www.sciencedirect.com/science/article/pii/S111001682500136XColorectal polypConvolutional neural networkMulti-stage classificationGAN-based data augmentation |
| spellingShingle | Shilong Chang Kun Yang Yucheng Wang Yufeng Sun Chaoyi Qi Wenlong Fan Ying Zhang Shuang Liu Wenshan Gao Jie Meng Linyan Xue A multi-stage deep learning network toward multi-classification of polyps in colorectal images Alexandria Engineering Journal Colorectal polyp Convolutional neural network Multi-stage classification GAN-based data augmentation |
| title | A multi-stage deep learning network toward multi-classification of polyps in colorectal images |
| title_full | A multi-stage deep learning network toward multi-classification of polyps in colorectal images |
| title_fullStr | A multi-stage deep learning network toward multi-classification of polyps in colorectal images |
| title_full_unstemmed | A multi-stage deep learning network toward multi-classification of polyps in colorectal images |
| title_short | A multi-stage deep learning network toward multi-classification of polyps in colorectal images |
| title_sort | multi stage deep learning network toward multi classification of polyps in colorectal images |
| topic | Colorectal polyp Convolutional neural network Multi-stage classification GAN-based data augmentation |
| url | http://www.sciencedirect.com/science/article/pii/S111001682500136X |
| work_keys_str_mv | AT shilongchang amultistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT kunyang amultistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT yuchengwang amultistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT yufengsun amultistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT chaoyiqi amultistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT wenlongfan amultistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT yingzhang amultistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT shuangliu amultistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT wenshangao amultistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT jiemeng amultistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT linyanxue amultistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT shilongchang multistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT kunyang multistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT yuchengwang multistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT yufengsun multistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT chaoyiqi multistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT wenlongfan multistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT yingzhang multistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT shuangliu multistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT wenshangao multistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT jiemeng multistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages AT linyanxue multistagedeeplearningnetworktowardmulticlassificationofpolypsincolorectalimages |