3D temperature mapping inside living tissue

Researchers from Ca’ Foscari University of Venice and the Universidad Autónoma de Madrid have found a way to map temperature in three dimensions within biological tissue, using invisible light and artificial intelligence. Described in the journal Nature Communications, their technique could transform how we monitor temperature inside the human body, potentially improving early disease detection and treatment monitoring without the need for costly or invasive imaging technologies.

The method relies on luminescent nanothermometers — ultrasmall particles made of silver sulfide (Ag₂S) that glow in the near-infrared when stimulated by light. The colour and intensity of that glow depend on both the temperature of the particle and the amount of biological tissue the light has to pass through.

To decode these subtle spectral shifts, the team trained a dual-layer neural network on hundreds of hyperspectral images collected under different conditions. The result is a model that can reconstruct accurate, three-dimensional thermal maps of tissue, even under biologically complex scenarios.

“We’re turning optical distortions, usually considered a problem, into a source of information,” said Riccardo Marin, an associate professor at Ca’ Foscari and co-lead author on the study. “With this method, we can detect both how hot a tissue is and how deep it lies beneath the surface.”

Proof-of-concept experiments demonstrated the system’s ability to detect temperature gradients in both artificial tissue phantoms and real biological samples. The researchers also succeeded in mapping blood vessels in a living animal; this is believed to be the first time that remote, high-resolution 3D thermal imaging has been achieved using light alone.

While conventional techniques like fMRI or PET scans require costly equipment and specialised training, this new optical method is designed to be portable, safer and significantly less expensive, potentially enabling diagnostics even outside the hospital setting. Beyond temperature sensing, the same principles could be adapted to measure other vital parameters such as oxygen concentration and pH, by tailoring the optical properties of the nanoparticles.

“We believe this is just the beginning,” said Erving Ximendes, an assistant professor and Ramón y Cajal Fellow at the Universidad Autónoma de Madrid. “Machine learning offers a powerful tool for navigating the complexity of real biological systems — far beyond what traditional models can achieve.”

Image caption: Student Liyan Ming, first author of the study, setting up a fluorescence imaging experiment. Image courtesy Riccardo Marin under CC BY-NC-SA