Endoscopic tissue time machine

Descripción del proyecto

Un endoscopio para predecir la evolución del carcinoma de cabeza y cuello

El carcinoma de cabeza y cuello es un grupo neoplasias malignas que se desarrollan en las membranas mucosas de la boca, la garganta, la nariz, las glándulas salivales y la laringe. Como en la mayoría de los tumores, el diagnóstico temprano es fundamental para el tratamiento. Sin embargo, las técnicas endoscópicas actuales no permiten predecir con fiabilidad qué lesiones precancerosas evolucionarán hacia un cáncer, lo que dificulta la intervención clínica. Para abordar esta limitación, el proyecto EPIC, financiado por el Consejo Europeo de Investigación, tiene por objeto desarrollar una sonda de fibra óptica miniaturizada que registre simultáneamente información molecular, estructural y vascular del tejido en tiempo real. La plataforma integra un modelo de aprendizaje profundo para predecir la evolución de las lesiones precancerosas a cáncer. La tecnología se validará primero en modelos animales y, a continuación, en un estudio longitudinal en humanos.

Objetivo

Head and neck cancers (HNC), which include malignancies of the oral cavity, tonsils, nasopharynx, and larynx, are life-threatening, particularly when diagnosed in advanced stages. Effective management requires early detection and continuous monitoring to enable earliest possible intervention of high-risk precancerous lesions. Unfortunately, current endoscopic techniques are limited in their ability to provide accurate predictions of disease trajectories.
The overarching goal of EPIC is to develop a new endoscopic technology to predict HNC tissue state trajectory with the fidelity needed to intervene in the present – an “endoscopic tissue time machine”. We will create a miniaturized 2.0 mm forward-viewing fibre-optic probe compatible with head and neck (H&N) endoscopes that integrates Raman spectroscopy (for molecular state), optical coherence tomography (for morphological state), and real-time optical vascular imaging (for microvascular state). This ground-breaking label-free platform technology will facilitate a synergistic integration of clinically relevant multi-modal endoscopic data, bridging different scales and modalities while putting endoscopic data in a clinical and biological context. We will develop an explainable deep learning model (an endoscopic time machine) to predict transition points in the progression from normal tissue to dysplasia and ultimately to cancer in an animal model. To address the unmet clinical need, we will conduct a longitudinal human observational study utilizing the endoscopic tissue time machine for real-time in vivo prediction of progression risk from dysplasia to cancer.
This project will integrate scientific and technological advances to offer new insights into the microenvironment of H&N carcinogenesis, enhance understanding of clinical diversity, and identify novel transition biomarkers for improved prognostics and timely interventions.