Artificial Intelligence facilitates breakthrough in diagnostics — ScienceDaily

Researchers at LMU and Charité hospital in Berlin have developed a method to classify difficult-to-diagnose nasal cavity tumors.

Although tumors in the nasal cavity and paranasal sinus are confined to a small area, they cover a very broad spectrum with many tumor types. They are difficult to diagnose as they often do not exhibit a particular pattern or appearance. This is especially true for so-called sinonasal undifferentiated carcinomas (SNUCs).

Now, from the Institute of Pathology at LMU, Dr. Philipp Jurmeister and Prof. Frederick Klauschen and Professor of Charité University Hospital. a decisive advance in diagnosis. The team developed an AI tool that reliably distinguishes tumors based on chemical DNA modifications and assigns SNUCs that previously available methods were unable to distinguish into four clearly distinct groups. This breakthrough could open new opportunities for targeted therapies.

Tumor-specific DNA modifications

Chemical modifications in DNA play a vital role in regulating gene activity. This includes DNA methylation, in which an extra methyl group is added to the DNA building blocks. In previous work, scientists had already shown that the methylation pattern of the genome is specific to different tumor types because it can be traced back to the tumor cell of origin.

“On this basis, we recorded DNA methylation patterns of approximately 400 tumors in the nasal cavity and paranasal sinus,” says Capper. Thanks to extensive international collaboration, the researchers were able to compile a large number of samples, although these tumors are rare and make up only about four percent of all malignant tumors in the nose and throat area.

Four tumor groups with different prognosis

For the analysis of DNA methylation data, the researchers developed an AI model that assigns tumors to different classes. “Machine learning methods are indispensable because of the large volumes of data they contain,” says Jurmeister. “To really recognize the patterns, we had to evaluate several thousand methylation positions in our study.” This revealed that SNUCs can be divided into four groups that also differ in further molecular properties.

Moreover, these results are clinically relevant as various groups have different prognoses. “For example, although tumors look very aggressive under the microscope, one group does surprisingly well,” says Klauschen. “However, another group has a poor prognosis.” Based on the molecular characteristics of the groups, researchers may also develop new targeted treatment approaches in the future.

Story Source:

materials provided by Ludwig-Maximilians-Universität München. Note: Content can be edited for style and length.

Leave a Reply

Your email address will not be published. Required fields are marked *