Using Artificial Intelligence to Improve the Diagnosis of Colon Cancer

AI and infrared imaging help categorize tumors automatically and are quicker compared to earlier techniques.

In the field of therapy, enormous development over the past years has considerably increased the chances of cures for patients with colon cancer. But such new methods, like immunotherapies, need an accurate diagnosis so that they can be specifically customized to the individual.

At the Centre for Protein Diagnostics PRODI at Ruhr University Bochum, Germany, scientists are making use of artificial intelligence alongside infrared imaging to fully customize colon cancer therapy to individual patients.

The label-free and automatable technique could complement present pathological analyses. The team headed by Professor Klaus Gerwert reported their findings in the “European Journal of Cancer” in January 2023.

Deep Insights into Human Tissue Within One Hour

In recent years, the PRODI group has been developing a new digital imaging method. The label-free infrared (IR) imaging quantifies the proteomic and genomic composition of the reviewed tissue, which offers molecular information depending on the infrared spectra.

Using artificial intelligence, this data is exhibited as false-color images. To perform this, the scientists made use of image analysis techniques from the field of deep learning.

In collaboration with clinical partners, the PRODI team was able to display the fact that the use of deep neural networks makes it feasible to reliably identify the alleged microsatellite status, a therapeutically and prognostically applicable parameter, in colon cancer.

In this process, the tissue sample abides by a user-independent, standardized, automated process and allows a spatially resolved differential classification of the tumor in just one hour.

Indication of the Effectiveness of Therapies

As far as classical diagnostics is concerned, microsatellite status is identified either by complicated immunostaining of several proteins or by DNA analysis.

15 to 20 percent of colon cancer patients show microsatellite instability in the tumor tissue. This instability is a positive biomarker indicating that immunotherapy will be effective.

Andrea Tannapfel, Professor and Head, Institute of Pathology, Ruhr Universitat Bochum

Also, with the ever-increasing therapy options, the quick and straightforward determination of such biomarkers is turning out to be very important. Depending on IR microscopic data, neuronal networks were altered, optimized, and trained at PRODI to fix label-free diagnostics.

Contrary to immunostaining, this method does not need dyes and is considerably quicker compared to DNA analysis.

We were able to show that the accuracy of IR imaging for determining microsatellite status comes close to the most common method used in the clinic, immunostaining.

Stephanie Schörner, PhD Student, Ruhr Universitat Bochum

“Through constant further development and optimization of the method, we expect a further increase in accuracy,” added Dr Frederik Großerüschkamp.

It was possible to execute this project through a long-lasting and intensive collaboration between the Institute of Pathology at Ruhr University (Professor Andrea Tannapfel), the Clinic for Haematology and Oncology at the St. Josef Hospital, Clinical Centre of Ruhr University (Professor Anke Reinacher-Schick) and the Centre for Protein Diagnostics (Professor Klaus Gerwert).

The work of the Research Center for Protein Diagnostics (PRODI) was financially supported by the State of North Rhine-Westphalia, Ministry of Culture and Science (grant number: 111.08.03.05-133974).

The register study was financially supported by Roche Pharma AG. Parts of the project were funded by the Slide2Mol project through the Computational Life Science program of the Federal Ministry of Education and Research.

Journal Reference:

Gerwert, K., et al. (2023) Fast and label-free automated detection of microsatellite status in early colon cancer using artificial intelligence integrated infrared imaging. European Journal of Cancer. doi.org/10.1016/j.ejca.2022.12.026