Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 Jun 26;10(13):e33655.
doi: 10.1016/j.heliyon.2024.e33655. eCollection 2024 Jul 15.

Efficient colorectal polyp segmentation using wavelet transformation and AdaptUNet: A hybrid U-Net

Devika Rajasekar  1 , Girish Theja  1 , Manas Ranjan Prusty  2 , Suchismita Chinara  3
Affiliations

Efficient colorectal polyp segmentation using wavelet transformation and AdaptUNet: A hybrid U-Net

Devika Rajasekar et al. Heliyon. .

Abstract

The prevalence of colorectal cancer, primarily emerging from polyps, underscores the importance of their early detection in colonoscopy images. Due to the inherent complexity and variability of polyp appearances, the task stands difficult despite recent advances in medical technology. To tackle these challenges, a deep learning model featuring a customized U-Net architecture, AdaptUNet is proposed. Attention mechanisms and skip connections facilitate the effective combination of low-level details and high-level contextual information for accurate polyp segmentation. Further, wavelet transformations are used to extract useful features overlooked in conventional image processing. The model achieves benchmark results with a Dice coefficient of 0.9104, an Intersection over Union (IoU) coefficient of 0.8368, and a Balanced Accuracy of 0.9880 on the CVC-300 dataset. Additionally, it shows exceptional performance on other datasets, including Kvasir-SEG and Etis-LaribDB. Training was performed using the Hyper Kvasir segmented images dataset, further evidencing the model's ability to handle diverse data inputs. The proposed method offers a comprehensive and efficient implementation for polyp detection without compromising performance, thus promising an improved precision and reduction in manual labour for colorectal polyp detection.

Keywords: Colorectal polyps; Data augmentation; Deep learning; Polyp segmentation; U-Net; Wavelet transformation.

PubMed Disclaimer

Conflict of interest statement

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Figures

Fig. 1
Fig. 1
Proposed workflow.
Fig. 2
Fig. 2
Visual comparison of original and augmented images and masks.
Fig. 3
Fig. 3
Comparison of images before and after wavelet transformation.
Fig. 4
Fig. 4
Proposed AdaptUNet architecture.
Fig. 5
Fig. 5
ROC Curves of the model training on (a) CVC-300 and (b) CVC-ColonDB dataset.
Fig. 6
Fig. 6
Visual evaluation results for CVC-300 Dataset.
Fig. 7
Fig. 7
Visual evaluation results for CVC-ColonDB dataset.
Fig. 8
Fig. 8
(a) ROC Curve of Kvasir dataset, (b) ROC Curve of ETIS Datasets.
Fig. 9
Fig. 9
Visual evaluation results on the Kvasir dataset.
Fig. 10
Fig. 10
Visual evaluation results on the ETIS Dataset.
See this image and copyright information in PMC

References

    1. Ahmed M. Colon cancer: a clinician's perspective in 2019. Gastroenterol. Res. Feb. 2020;13(1):1–10. doi: 10.14740/gr.v13i1.1239. - DOI - PMC - PubMed
    1. Godkhindi A.M., Gowda R.M. 2017 International Conference on Energy, Communication, Data Analytics and Soft Computing (ICECDS) Aug. 2017. Automated detection of polyps in CT colonography images using deep learning algorithms in colon cancer diagnosis; pp. 1722–1728. - DOI
    1. Petrova J., Hostalkova E. Edge detection in medical image using the Wavelet transform. Rep. Res. Dep. Comput. Control Eng. Czech Public. 2011;78(1):1–5.
    1. Chlap P., Min H., Vandenberg N., Dowling J., Holloway L., Haworth A. A review of medical image data augmentation techniques for deep learning applications. J. Med. Imaging Radiat. Oncol. 2021;65(5):545–563. doi: 10.1111/1754-9485.13261. - DOI - PubMed
    1. Tharwat M., Sakr N.A., El-Sappagh S., Soliman H., Kwak K.-S., Elmogy M. Colon cancer diagnosis based on machine learning and deep learning: modalities and analysis techniques. Sensors. Nov. 2022;22(23):9250. doi: 10.3390/s22239250. - DOI - PMC - PubMed

LinkOut - more resources