# Estimating the coverage in 3d reconstructions of the colon from colonoscopy videos

arXiv: https://arxiv.org/abs/2210.10459 · HTML: https://mlsanity.com/paper/2210.10459/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 9 of its 27 references that are in the map.
- First posted on arXiv 19 Oct 2022.

## Builds on (most cited first, 9 of 9)

- [PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation](https://mlsanity.com/paper/1612.00593/) (2016)
- [3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation](https://mlsanity.com/paper/1606.06650/) (2016)
- [Direct Sparse Odometry](https://mlsanity.com/paper/1607.02565/) (2016)
- [PCN: Point Completion Network](https://mlsanity.com/paper/1808.00671/) (2018)
- [Shape Completion using 3D-Encoder-Predictor CNNs and Shape Synthesis](https://mlsanity.com/paper/1612.00101/) (2016)
- [Point Cloud Completion by Skip-attention Network with Hierarchical Folding](https://mlsanity.com/paper/2005.03871/) (2020)
- [High-Resolution Shape Completion Using Deep Neural Networks for Global Structure and Local Geometry Inference](https://mlsanity.com/paper/1709.07599/) (2017)
- [Detecting Deficient Coverage in Colonoscopies](https://mlsanity.com/paper/2001.08589/) (2020)
- [C3Fusion: Consistent Contrastive Colon Fusion, Towards Deep SLAM in Colonoscopy](https://mlsanity.com/paper/2206.01961/) (2022)

## Abstract (arXiv)

Colonoscopy is the most common procedure for early detection and removal of polyps, a critical component of colorectal cancer prevention. Insufficient visual coverage of the colon surface during the procedure often results in missed polyps. To mitigate this issue, reconstructing the 3D surfaces of the colon in order to visualize the missing regions has been proposed. However, robustly estimating the local and global coverage from such a reconstruction has not been thoroughly investigated until now. In this work, we present a new method to estimate the coverage from a reconstructed colon pointcloud. Our method splits a reconstructed colon into segments and estimates the coverage of each segment by estimating the area of the missing surfaces. We achieve a mean absolute coverage error of 3-6% on colon segments generated from synthetic colonoscopy data and real colonography CT scans. In addition, we show good qualitative results on colon segments reconstructed from real colonoscopy videos.