To diagnose a brain tumor, it is necessary to perform tests using image techniques such as magnetic resonance or computerized axial tomography, but once the neoplasm is detected, a biopsy is performed to analyze the tissue and this entails risks for the patient, especially when the tumor is located in certain areas of the brain.
Therefore, a team of experts in cancer medicine of the Charité University Hospital – Universitätsmedizin Berlin has decided to look for safer diagnostic alternatives and has developed an innovative Artificial Intelligence Model (AI) capable of identifying tumors through the analysis of genetic materialeven from the cerebrospinal fluid. The results of the tests have been published in the magazine Nature Cancer And they show that it is a fast and reliable tool, which allows to avoid invasive procedures.
The tumor footprint that allows you to determine its type
Each tumor has its own features: it grows at different speed, responds to different treatments and presents particularities in its metabolism. Although the organ of origin is the same, there are multiple types of tumors with different molecular characteristics. Grouping them correctly is key to administering directed treatments and precisely adjusting chemotherapy.
The new model is based on the Analysis of the “epigenetic profile” of the tumorthat is, the chemical modifications that regulate which genes are activated or silenced. These marks work as molecular switches and leave a unique pattern – a kind of fingerprint – that allows identifying the type of tumor. “In tumor cells, epigenetic information is altered in a characteristic way. Based on their profiles, we can differentiate tumors and classify them,” explains Dr. Philipp Euskirchen, a scientist at the Berlin headquarters of the German Cancer Consortium and the Charité Neuropathology Institute, which has directed the study in some cases, it is enough to analyze a sample of the liquid Cephalorraquido to obtain this information, without the need to operate.
The model, called crossn, is based on a simple but very effective neuronal network. After being trained with a large number of tumor samples, its precision was evaluated in more than 5,000 cases. The results impressed the researchers themselves, since they were able to diagnose with 99.1% of success the brain tumors analyzed and identify more than 170 types of cancer – of different organs – with a accuracy of 97.8%. “This means that it can be used for cancers of all organs, in addition to the relatively rare brain tumors.” Philipp Euskirchen pointed out.
The crossn model was able to diagnose with 99.1% success the brain tumors analyzed and identify more than 170 types of cancer – of different organs – with a 97.8% precision
The decisive factor for future approvals in clinical applications is that the models are completely explainable, that is, it is possible to understand how decisions are made. “The most surprising thing is that, despite being a simpler model than others used so far, it offers more precise and more reliability diagnoses,” said Dr. Sören Lukassen, head of the “Medical Omics” working group at the Berlin Health Institute (BIH). This not only improves the effectiveness of the diagnosis, but makes it more transparent and understandable for doctors, a key requirement for clinical use.
Diagnosis without surgery: the real case of a patient
One of the beneficiaries of this technology was a patient who arrived with double vision that underwent a magnetic resonance that revealed a brain tumor located in a critical area. Since a surgical intervention was risky, doctors chose to analyze their cerebrospinal fluid with a genetic sequencing technique called “Nanopore”. Thanks to the AI model, they could diagnose a central nervous system lymphoma and quickly start the treatment with proper chemotherapy.
The researchers are already collaborating with the German Cancer Consortium (DKTK) to carry out clinical trials with Crossnn in the eight centers that are part of the consortium. They also plan to study their application during surgeries. The objective is to integrate this system into the usual medical practice as a faster, more safe and affordable alternative for the diagnosis of tumors.
Source: www.webconsultas.com