IDDF2018-ABS-0260 Deep learning for polyp segmentation

Liansheng Wang, Cong Xie, Yanxing HuClinical 2018 · DOI

Abstract

Colorectal cancer is a significant global health threat, often originating from adenomatous polyps. While endoscopic examination is the primary screening method, high polyp miss rates remain a critical limitation. This study investigates the application of deep learning techniques to automate polyp localization and segmentation, aiming to improve detection accuracy and clinical efficiency. Using a combined dataset of 912 images (CVC-ColonDB and CVC-ClinicDB), we evaluated three convolutional neural network architectures: Fully Convolutional Network (FCN), SegNet, and a modified ResNet. Results indicate that deep learning approaches, particularly the modified ResNet, achieve high accuracy (>90%) and Intersection over Union (IoU) scores (>50%), demonstrating significant potential for real-time clinical assistance.

Introduction

Screening for colorectal cancer through colonoscopy is effective for both early detection and prevention. However, the procedure's efficacy is hampered by human factors, leading to a notable miss rate for polyps. Traditional computer-aided detection methods relied on hand-crafted features, which are often computationally expensive and struggle with the high morphological variability of polyps. This research explores deep learning as a more robust, time-saving alternative for automated segmentation.

Methods

The study framed polyp detection as a semantic segmentation task, classifying pixels as either "polyp" or "background." The researchers utilized a combined dataset of 912 images from the CVC-ColonDB and CVC-ClinicDB benchmarks.

Examples of input colonoscopy images and their corresponding segmentation masks.

Three distinct architectures were compared:

  1. FCN8: A standard for semantic segmentation that preserves spatial information by avoiding fully connected layers.
  2. SegNet: An encoder-decoder architecture optimized for pixel-wise efficiency.
  3. Modified ResNet: A Residual Network adapted with skip connections to facilitate deeper feature extraction and improve boundary definition.

Performance was measured using Accuracy and Intersection over Union (IoU), which quantifies the overlap between predicted and ground-truth masks.

Results

Quantitative analysis revealed that while all models achieved high overall accuracy, the modified ResNet was superior in precise segmentation. The results on the test set are summarized below:

Architecture IoU (Polyp) IoU (Mean) Accuracy (Mean)
FCN8 46.99% 70.51% 94.32%
SegNet 53.80% 74.35% 95.18%
Modified ResNet 64.03% 80.03% 96.29%

Qualitative results (Figure 1) demonstrate that all three architectures successfully localized polyps. However, SegNet and the modified ResNet (rows b and c) produced significantly finer boundaries compared to FCN8.

Qualitative segmentation results for FCN, SegNet, and modified ResNet architectures across three sample cases (a, b, c).

Conclusion

Deep learning-based approaches offer a time-saving and effective solution for polyp segmentation in colonoscopy images compared to traditional hand-crafted methods. The evaluated architectures, particularly the modified ResNet, not only perform with high accuracy but also support near real-time processing. This suggests a strong potential for integrating these models into clinical workflows to assist endoscopists in reducing polyp miss rates.

Summary updated

Abstract

<h3>Background</h3> Colorectal cancer is one of the commonest malignancies which threaten the health of human. Most of the colorectal cancers arise from adenomatous polyps, screening for this cancer is effective not only for early detection but also for prevention. Endoscopic examination is an important way to reveal polyps. At present, the main limitation of colonoscopy is the high polyp miss rate. This study aims to improve the accuracy of detection and save doctors’ time with deep learning techniques. <h3>Methods</h3> The data we used is a combination of CVC-ColonDB and CVC-ClinicDB which contains 912 images with associated polyps. We treat the rest part of a colonoscopy image except polys as background. Convolutional networks are applied to polyp localization and segmentation. There are three different architectures which include FCN, SegNet and modified ResNet. We trained convolutional networks with these architectures respectively and evaluated these models based on Intersection over Union (IoU) and accuracy. <h3>Results</h3> The segmentation results on the test set are reported in table 1 and the accuracy of three architectures is greater than 90%. Results achieved in SegNet and modified ResNet are very encouraging, with a IoU higher than 50% for polyps. IDDF2018-ABS-0260 Figure 1 shows qualitative results of three architectures. It can be seen that all of them recognise the location of polyp successfully and rows 2 and 3 show finer boundaries. The results are reported on test set <h3>Conclusions</h3> Compare to traditional time-consuming hand-crafted segmentation methods, when considering polyp segmentation, approaches based on deep learning are time-saving and effective, showing good results in colonoscopy images. Given that three architectures we mentioned above not only performs well but also allows for nearly real-time processing, it has a great potential in polyp localization and segmentation.
BibTeX
@inproceedings{wang2018iddf,
  author = {Liansheng Wang and Cong Xie and Yanxing Hu},
  title = {{IDDF2018-ABS-0260 Deep learning for polyp segmentation}},
  booktitle = {Clinical},
  year = {2018},
  doi = {10.1136/gutjnl-2018-iddfabstracts.181},
}
Figure