Raman Spectroscopy-Based Deep Learning Method for Early Detection of Pan-Cancers A Prospective, Single-Arm, Multicentre Observational Study

What this Trial Is About

Researchers are exploring the use of Raman spectroscopy combined with deep learning to develop an early screening method for various types of cancer. This observational study includes healthy individuals, patients at risk, and those already diagnosed with cancers such as colorectal, gastric, hepatic, pancreatic, and esophageal cancers. The study aims to evaluate the deep learning model's accuracy in identifying cancer-specific features in blood samples and to interpret the model's diagnostic decisions. Blood samples are collected from participants during routine clinical blood tests and medical evaluations. The samples undergo preprocessing steps like alignment resampling, baseline removal, and normalization before being analyzed by different deep learning models. These models classify participants as healthy or having specific cancers or precancerous conditions. The data is divided into training, validation, and testing sets to evaluate model performance, and visualization techniques help explain the model's decision-making. Participants provide blood samples which are analyzed to distinguish cancer cases from healthy controls. Researchers measure the model's ability to accurately classify these cases using Raman spectral data. The study tracks the model construction and interpretable analysis phases, expected to be completed within a few months after data collection. The study includes healthy volunteers and patients with various cancer or precancerous diagnoses, with ongoing follow-up to confirm cancer types.

Raman Spectroscopy-Based Deep Learning Model for Early Pan-Cancer Early Diagnosis