Abstract:Gastrointestinal (GI) bleeding is a common problem and may lead to fatal consequences. The detection of bleeding is currently determined through conventional examination of wireless capsule endoscopy (WCE) images. Image-based bleeding detection is performed using only color pixels; therefore, image and color distortion affect the accuracy of RGB-based methods adversely. On the other hand, imaging modalities like multispectral features contain several unique characteristics of the target material, and therefore are less error-prone. In this regard, an on-chip bleeding detection sensor based on blood's optical properties is developed in this article. For creating this sensor, an array of 12 optical sensors, six in visible and six in near-infrared range, is tested with blood samples (BS) in various concentrations and non-blood samples (NBS), such as food pigments and natural foods (digestible and nondigestible). Various feature selection and machine learning approaches have been used to select the optimal bandwidth for the best detection accuracy. A capsule prototype is designed that uses 450 nm, 610 nm, and 810 nm of wavelengths. The prototype is tested in two in vitro experiments using two porcine intestines to validate the proof of concept. The results indicate that the designed system can distinguish occult and acute bleeding from a distance of 1 cm with angles of 45 and 90 degrees. An F1-score of 99% is achieved using the bagged decision tree algorithm. The proposed solution for detecting GI bleeding will eliminate the need to use a camera module and large data transmission, and thereby consumes 1.48 times less power than conventional WCE systems.

Bleeding Detection from Wireless Capsule Endoscopy Images Using Improved Euler Distance in CIELab

Computer-aided Bleeding Detection in WCE Video.

Bleeding Detection in Wireless Capsule Endoscopy Image Video Using Superpixel-Color Histogram and a Subspace KNN Classifier.

Bleeding Detection in Wireless Capsule Endoscopy Images Based on Color Invariants and Spatial Pyramids Using Support Vector Machines

Saliency-Based Bleeding Localization for Wireless Capsule Endoscopy Diagnosis

Design of the Color Vector Similarity Coefficient Classifier for Bleeding Detection in Wireless Capsule Endoscopy Images

Detection of Bleeding Image in Wireless Capsule Endoscope Based on Superpixel Segmentation

Detection of Bleeding in Wireless Capsule Endoscopy Images Using Range Ratio Color

GI Bleeding Detection in Wireless Capsule Endoscopy Images Based on Pattern Recognition and a MapReduce Framework

Research on a New Feature Detection Algorithm for Wireless Capsule Endoscope Bleeding Images Based on Super-pixel Segmentation.

A Deep Convolutional Neural Network for Bleeding Detection in Wireless Capsule Endoscopy Images

Bleeding detection in wireless capsule endoscopy based on MST clustering and SVM

[A Method for Bleeding Detection in Endoscopy Images Using SVM].

Blood detection in wireless capsule endoscope images based on salient superpixels

Gastrointestinal Bleeding Detection In Wireless Capsule Endoscopy Images Using Handcrafted And Cnn Features

Detection of lymphangiectasia disease from wireless capsule endoscopy images with adaptive threshold

Bleeding Detection in Wireless Capsule Endoscopy Based on Probabilistic Neural Network.

Detection of Protruding Lesion in Wireless Capsule Endoscopy Videos of Small Intestine.

Automated Detection of Bleeding in Capsule Endoscopy Using On-Chip Multispectral Imaging Sensors

Multi-Stage Attention-Unet For Wireless Capsule Endoscopy Image Bleeding Area Segmentation

DFCA-Net: Dual Feature Context Aggregation Network for Bleeding Areas Segmentation in Wireless Capsule Endoscopy Images