WM-DOVA maps for accurate polyp highlighting in colonoscopy: Validation vs. saliency maps from physicians

By Jorge Bernal, F. Sánchez, G. Fernández-Esparrach, Debora Gil, C. R. Miguel, F. Vilariño. Comput. Medical Imaging Graph., 2015

2,058 citations recorded by the source corpus.

We introduce in this paper a novel polyp localization method for colonoscopy videos. Our method is based on a model of appearance for polyps which defines polyp boundaries in terms of valley information. We propose the integration of valley information in a robust way fostering complete, concave and continuous boundaries typically associated to polyps. This integration is done by using a window of radial sectors which accumulate valley information to create WM-DOVA (Window Median Depth of Valleys Accumulation) energy maps related with the likelihood of polyp presence. We perform a double validation of our maps, which include the introduction of two new databases, including the first, up to our knowledge, fully annotated database with clinical metadata associated. First we assess that the highest value corresponds with the location of the polyp in the image. Second, we show that WM-DOVA energy maps can be comparable with saliency maps obtained from physicians' fixations obtained via an eye-tracker. Finally, we prove that our method outperforms state-of-the-art computational saliency results. Our method shows good performance, particularly for small polyps which are reported to be the main sources of polyp miss-rate, which indicates the potential applicability of our method in clinical practice.

Subfields: Computer Vision and Pattern Recognition, Oncology

View Paper: https://www.semanticscholar.org/paper/0245171475005c171bd359fe9f461f5b6b7fb634

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Oncology

Citations in the corpus

CaraNet: context axial reverse attention network for segmentation of small medical objects (2021, methodology): “We choose five polyp datasets: ETIS [10], CVC-ClinicDB [11], CVC-ColonDB [12], EndoScene [13] and Kvasir [14].”
TransFuse: Fusing Transformers and CNNs for Medical Image Segmentation (2021, methodology): “Five public polyp datasets are chosen: Kvasir [9], CVC-ClinicDB [1], CVCColonDB [17], EndoScene [20] and ETIS [15].”
Automatic Polyp Segmentation via Multi-scale Subtraction Network (2021, methodology): “We evaluate the proposed model on ﬁve benchmark datasets: CVC-ColonDB [22], ETIS [21], Kvasir [11], CVC-T [24] and CVC-ClinicDB [2].”
Unified Focal loss: Generalising Dice and cross entropy-based losses to handle class imbalanced medical image segmentation (2021, methodology): “We use the CVC-ClinicDB dataset, which consists of 612 frames containing polyps with image resolution 288 × 384 pixels, generated from 23 video sequences from 13 different patients using standard colonoscopy interventions with white light (Bernal et al., 2015).”
Comparative Validation of Polyp Detection Methods in Video Colonoscopy: Results From the MICCAI 2015 Endoscopic Vision Challenge (2017, positive): “Deepening the analysis, we observe a decrease in the difference in performance observed in global analysis between hand-crafted methods (CVC-CLINIC) and CNN-based methods when videos with only polyp frames are analyzed.”
DS-TransUNet: Dual Swin Transformer U-Net for Medical Image Segmentation (2021, methodology): “Polyp Segmentation: For polyp segmentation task, we select ﬁve public polyp datasets including Kvasir [33], CVC-ColonDB (ColonDB) [34], EndoScene [35], ETIS [36], and CVC-ClinicDB (ClinicDB) [37].”
DoubleU-Net: A Deep Convolutional Neural Network for Medical Image Segmentation (2020, methodology): “The proposed architecture uses dice loss as the loss function and produces an improved output segmentation maps as compared to U-Net and ResUNet on the Kvasir-SEG [2] and CVC-ClinicDB [19] datasets.”
Segment Anything Model for Medical Image Segmentation: Current Applications and Future Directions (2024, background): “Zhou et al. [51] evaluate the performance of SAM with unprompted setting in segmenting polyps from colonoscopy images on five benchmark datasets [52, 53, 54, 55, 56].”
Kvasir-SEG: A Segmented Polyp Dataset (2019, methodology): “These are CVC-ColonDB [24], ASU-Mayo Clinic Colonoscopy Video c (cid:13) Database [3], ETIS-Larib Polyp DB [23], and CVC-Clinic DB [2].”
ResUNet++: An Advanced Architecture for Medical Image Segmentation (2019, methodology): “The second dataset we have used is the CVC-ClinicDB database [32], which is an open-access dataset of 612 images with a resolution of 384×288 from 31 colonoscopy sequences.”
Cross-level Feature Aggregation Network for Polyp Segmentation (2023, methodology): “C l i n i c D B [ 47 ] K v a s i r [ 49 ] m D i c e”
Learning with Limited Annotations: A Survey on Deep Semi-Supervised Learning for Medical Image Segmentation (2022, neutral): “[121] 2D Colonoscope Kvasir-SEG [134], CVC-Clinic DB [135] Introduce collaborative and adversarial learning of focused and dispersive representations Adversarial learning”
DCSAU-Net: A Deeper and More Compact Split-Attention U-Net for Medical Image Segmentation (2022, methodology): “3) Experimental analysis is conducted with four diﬀerent medical imagesegmentationdatasets, including2018Data Science Bowl [14], ISIC-2018Lesion BoundarySegmen-tation [ 15, 16], CVC-ClinicDB [17], and a multi-class segmentation dataset: SegPC-2021 [18].”
Medical Image Segmentation via Cascaded Attention Decoding (2023, methodology): “CVC-ClinicDB [1] contains 612 images, which are extracted from 31 colonoscopy videos.”
U-KAN Makes Strong Backbone for Medical Image Segmentation and Generation (2024, background): “The CVC-ClinicDB dataset [5], often abbreviated simply as "CVC," serves as a publicly accessible resource for polyp diagnosis within colonoscopy videos.”
UACANet: Uncertainty Augmented Context Attention for Polyp Segmentation (2021, background): “We validate our method with five famous polyp segmentation benchmarks, Kvasir, CVC-ClinicDB, ETIS, CVC-ColonDB and CVC-300 and achieve state-of-the-art performance among previous methods. ar X iv :2 10 7.”