[Paper Review] A Capsule-Sized Multi-Wavelength Wireless Optical System for Edge-AI-Based Classification of Gastrointestinal Bleeding Flow Rate

Yunhao Bian, Dawei Wang+6 more|arXiv (Cornell University)|Jan 25, 2026

Gastrointestinal Bleeding Diagnosis and Treatment0 citations

TL;DR

A capsule-sized, multi-wavelength optical sensor with on-device edge-AI classifies GI bleeding flow rate and reduces energy use by performing inference on-device rather than streaming raw data.

ABSTRACT

Post-endoscopic gastrointestinal (GI) rebleeding frequently occurs within the first 72 hours after therapeutic hemostasis and remains a major cause of early morbidity and mortality. Existing non-invasive monitoring approaches primarily provide binary blood detection and lack quantitative assessment of bleeding severity or flow dynamic, limiting their ability to support timely clinical decision-making during this high-risk period. In this work, we developed a capsule-sized, multi-wavelength optical sensing wireless platform for order-of-magnitude-level classification of GI bleeding flow rate, leveraging transmission spectroscopy and low-power edge artificial intelligence. The system performs time-resolved, multi-spectral measurements and employs a lightweight two-dimensional convolutional neural network for on-device flow-rate classification, with physics-based validation confirming consistency with wavelength-dependent hemoglobin absorption behavior. In controlled in vitro experiments under simulated gastric conditions, the proposed approach achieved an overall classification accuracy of 98.75% across multiple bleeding flow-rate levels while robustly distinguishing diverse non-blood gastrointestinal interference. By performing embedded inference directly on the capsule electronics, the system reduced overall energy consumption by approximately 88% compared with continuous wireless transmission of raw data, making prolonged, battery-powered operation feasible. Extending capsule-based diagnostics beyond binary blood detection toward continuous, site-specific assessment of bleeding severity, this platform has the potential to support earlier identification of clinically significant rebleeding and inform timely re-intervention during post-endoscopic surveillance.

Motivation & Objective

• Develop a capsule-sized wireless optical platform for time-resolved, multi-spectral blood sensing in the GI tract.
• Enable order-of-magnitude-level classification of GI bleeding flow rate beyond binary detection.
• Implement lightweight on-device CNN inference to assess bleeding severity in real time.
• Validate physics-based consistency with wavelength-dependent hemoglobin absorption.
• Demonstrate energy efficiency improvements by reducing data transmission via embedded processing.

Proposed method

• Time-resolved, multi-spectral optical measurements are performed by a capsule-sized wireless sensor.
• A lightweight two-dimensional convolutional neural network is used for on-device flow-rate classification.
• Physics-based validation confirms consistency with hemoglobin absorption across wavelengths.
• Inference is executed on capsule electronics to minimize data transmission.
• In vitro experiments simulate gastric conditions to evaluate bleeding flow-rate classification across levels, including non-blood interference.

Experimental results

Research questions

• RQ1 Can a capsule-sized multi-wavelength optical system classify GI bleeding flow rate using edge-AI on-device inference?
• RQ2 Does on-device processing reduce energy consumption while maintaining high classification accuracy under simulated gastric conditions?
• RQ3 Is the bleeding flow-rate classification consistent with wavelength-dependent hemoglobin absorption physics?

Key findings

• Classification accuracy of 98.75% across multiple bleeding flow-rate levels in controlled in vitro experiments.
• The system robustly distinguishes diverse non-blood gastrointestinal interferences.
• Embedded inference reduces energy consumption by approximately 88% compared with continuous wireless transmission of raw data.
• The approach enables prolonged, battery-powered operation of capsule diagnostics.
• The platform moves beyond binary blood detection toward continuous, site-specific bleeding severity assessment.