AI can identify chemicals from photos with up to 98% accuracy

Take a gander at this: Artificial intelligence is demonstrating remarkable capabilities in chemical analysis, with researchers now showing it can accurately predict chemical compositions from photos. This NASA-supported innovation represents a significant advancement in analytical chemistry, potentially democratizing access to chemical analysis technology by drastically reducing costs and equipment requirements. The breakthrough could transform everything from laboratory work to space exploration by replacing expensive specialized instruments with something as accessible as a smartphone camera.

The big picture: Researchers have developed an AI tool that can identify chemical compositions from photographs with 98.7% accuracy for salt type and 92.2% accuracy for concentration levels, according to a new study published in Digital Discovery.

• Florida State University professor Dr. Oliver Steinbock and colleagues created an AI system that can detect patterns in dried chemical deposits that are typically perceived as random or meaningless.
• The technology could eventually replace expensive analytical equipment with simple smartphone applications, making chemical analysis more accessible for fieldwork and remote locations.

How it works: The researchers used a robotic drop imager called RODI to create and photograph high-quality images of dried salt solutions for AI training data.

• The system generated over 23,400 real-world images capturing the visual patterns of seven different inorganic salts at five different concentration levels.
• For each image, 47 different pattern characteristics were analyzed to create comprehensive training data for the AI algorithms.

Technical approach: The researchers employed multiple AI algorithms in their system to achieve high accuracy in chemical identification and characterization.

• They used Random Forest and XGBoost algorithms to detect the type of salt present in the sample.
• A multi-layer perceptron (MLP) deep learning algorithm was then used to further analyze the samples and predict concentration levels.

In plain English: The system works like a highly trained eye that can recognize subtle patterns in dried salt deposits—patterns that humans can’t easily detect. This allows it to identify both what chemicals are present and in what amounts by simply looking at a photograph.

Why this matters: Chemical analysis is crucial across numerous industries including pharmaceuticals, biotechnology, environmental monitoring, and space exploration, but traditional methods often require expensive equipment and specialized facilities.

• The technology could dramatically lower barriers to chemical analysis in resource-limited settings or remote locations where transporting bulky equipment is impractical.
• It offers potential for “democratizing traditionally expensive analytical measurements” that currently rely on specialized instruments like mass spectrometers.

Potential applications: The researchers envision this technology extending beyond laboratory settings into practical field applications.

• The team suggests their approach could eventually be implemented using only “a phone camera and an app,” making sophisticated chemical analysis accessible anywhere.
• This could prove particularly valuable for space exploration, where payload weight and equipment limitations make traditional chemical analysis challenging.