Application of RESNET50 Convolution Neural Network for the Extraction of Optical Parameters in Scattering Media

2024-04-25

arXiv.org

AI Summary

Accurate, low-sample inversion of optical parameters—reduced scattering coefficient (μₛ′) and absorption coefficient (μₐ)—in strongly scattering media (e.g., biological tissue) remains challenging. To address this, we propose a ResNet50-based multimodal end-to-end regression framework. This work is the first to adapt ResNet50 for joint modeling of scattered-light features encompassing multiplanar intensity, exit angles, and spatial position distributions; training leverages lightweight Monte Carlo simulation data. The method achieves <5% error in μₛ′ and <15% error in μₐ under limited-sample conditions—outperforming conventional inverse methods and early CNNs by 22% in accuracy over unimodal inputs. Furthermore, we identify and characterize the fundamental bottleneck in μₐ estimation and introduce a dedicated enhancement pathway to mitigate it. Our framework establishes a new paradigm for rapid, robust optical parameter quantification in biological tissue imaging.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Medical and Biological Imaging
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Estimation of the optical properties of scattering media such as tissue is important in diagnostics as well as in the development of techniques to image deeper. As light penetrates the sample scattering events occur that alter the propagation direction of the photons in a random manner leading degradation of image quality. The distribution of the scattered light does, however, give a measure of the optical properties such as the reduced scattering coefficient and the absorption coefficient. Unfortunately, inverting scattering patterns to recover the optical properties is not simple, especially in the regime where the light is partially randomized. Machine learning has been proposed by several authors as a means of recovering these properties from either the back scattered or the transmitted light. In the present paper, we train a general purpose convolutional neural network RESNET 50 with simulated data based on Monte Carlo simulations. We show that compared with previous work our approach gives comparable or better reconstruction accuracy with training on a much smaller dataset. Moreover, by training on multiple parameters such as the intensity distribution at multiple planes or the exit angle and spatial distribution one achieves improved performance compared to training on a single input such as the intensity distribution captured at the sample surface. While our approach gives good parameter reconstruction, we identify factors that limit the accuracy of the recovered properties, particularly the absorption coefficient. In the light of these limitations, we suggest how the present approach may be enhanced for even better performance.

Problem

• Estimating optical properties in scattering media using RESNET50
• Improving reconstruction accuracy with smaller training datasets
• Addressing limitations in recovering absorption coefficient accurately

Innovation

• RESNET50 CNN for optical parameter extraction
• Monte Carlo simulated data training
• Multi-parameter input enhances reconstruction accuracy