- Photonic Data Science
- Research results
- Second opinion computer: Smart algorithms detect details
Second opinion computer: Smart algorithms detect details
Artificial intelligence (AI) has long ago arrived in our everyday lives, and has become a valuable addition to the lives of many people: Whether it’s self-controlling drones, intelligent room lighting or smart voice assistants – AI permeates almost all areas. In addition to logistics, the automotive industry and the entertainment sector, the smart support from AI also finds its way into medicine, providing valuable information about cancer and infectious diseases. Leibniz IPHT researchers are working together with Biophotonics Diagnostics GmbH on the AI-supported analysis of Raman spectra and image data.
“The aim is to use computer-assisted algorithms to identify abnormalities in imaging diagnostics and in spectrally measured data within a very short time. This would allow physicians to confirm their suspicion of a disease or an infectious agent in the sense of a second opinion. It would also be possible to pay special attention to conspicuous areas of a tissue sample and clarify them further medically,“ explains PD Dr. Thomas Bocklitz, head of the Photonic Data Science Department at Leibniz IPHT. Together with his team, he researches the smart algorithms to achieve exactly this aim, and the benefits they can provide to medicine. The expert is convinced that AI-based methods will not replace physicians in the future, but will offer real added value: With their ability to quickly recognize patterns in pathological findings, computer-assisted methods support therapy decisions.
Raman spectroscopy, in particular, can usefully support diagnostics and analytics through biochemical and molecular characterization of samples. However, due to the complexity and volume of the data, its potential is not fully utilized in clinical applications. With its intelligent and AI-supported software solutions, Biophotonics Diagnostics GmbH demonstrates that analysis of even comprehensive spectroscopic data can succeed easily and without specialized knowledge.
The spin-off, which emerged from a cooperation of Leibniz IPHT with Friedrich Schiller University Jena and the University Hospital Jena, successfully combines the intensive research of the three institutions in a user-friendly product: The RAMANMETRIX software developed by the company accelerates fast and intuitive evaluation of Raman spectra.
Future ideal solution
In order to further advance application of Raman spectroscopy in the medical environment, uniform procedures, and measurement methods, such as standardized measurement conditions and setups, are also required for generation and evaluation of comparable measurement data. Thomas Bocklitz’s team has therefore written a manual to Raman spectral data analysis using AI. With it, the researchers would like to contribute to the international standardization of data collection, processing, and AI-supported evaluation.
„In the future, we want to increasingly use AI for inverse modeling of measurement processes. With the support of AI, conclusions about the sample and possible errors in measurement will be drawn from measurement data that has already been generated. This will allow us to improve the initial data and further optimize diagnoses,“ says the scientist, who also heads the network „AI for Diagnostics and Therapy“ as part of the Leibniz Center for Photonics in Infection Research (LPI) at Friedrich Schiller University Jena.
What is artificial intelligence?
AI stands for artificially generated intelligent behavior of machines by means of mathematical algorithms. Thinking, solving problems and learning are its characteristic features. One subdomain of AI is machine learning, whose algorithms are trained by data sets. Thanks to pattern recognition, machine learning can solve tasks better and better. Deep learning as a sub-discipline of machine learning is mostly modeled on the human brain. It is able to learn from large complex amounts of data using artificial neural networks, to recognize regularities intelligently and to draw logical conclusions.