An algorithm, developed in Holland, allows a fairly accurate diagnosis based on the analysis of DNA segments: the analysis usually takes a week
L’
Artificial intelligence
it is also making its way into the operating room. This is demonstrated by a study published in
Nature
by researchers from UMC Utrecht and researchers, pathologists and neurosurgeons from the Princess Mxima Pediatric Oncology Center and UMC Amsterdam.
The team has developed a deep learning algorithm
called Sturgeon, which helps identify the type of brain tumor of a patient directly on the operating table within 1 hour and a half versus the week usually required for such analysis. The method involves scanning segments of DNA of the tumor by a computer and the detection of certain chemical changes (methylations) that can provide a detailed diagnosis of the type and even subtype of the brain tumor.
Deep learning algorithm
In brain tumors, Surgery is often the first treatment option. Currently, during surgery, neurosurgeons do not know precisely what type of brain tumor and what degree of aggressiveness they are dealing with. The exact diagnosis usually available only one week after surgery, after the tumor tissue has been analyzed visually and from a molecular point of view by the pathologist. Researchers at UMC Utrecht have developed a new deep learning algorithm, a form of artificial intelligence that significantly accelerates diagnosis.
Recently became available Nanopore sequencing: a technology that helps read DNA in real time – explains Jeroen de Ridder, leader of the research group at UMC Utrecht and the Oncode Institute –. To do this, we developed an algorithm that can learn from millions of simulated realistic “DNA snapshots.” With this algorithm we can identify the type of tumor fast enough to be able to directly adapt the surgical strategy if necessary.
Tested and trained with the biobank
Bastiaan Tops head of the Pediatric Oncology Laboratory of the Princess Mxima Center. It was he who brought together the new technology and the needs of the operating room. This was possible thanks to funding from the KiKa foundation and, more specifically, to the vast biobank which the Mxima Center has been managing for years. In this biobank among other things, tissues from children suffering from brain tumors are preserved. The algorithm was trained and tested using the biobank. The fact that we can now determine the type of brain tumor already during surgery demonstrates how technology can speed up diagnosis. And also how we can use an existing biobank to develop new technologies, explains Tops.
Used first on frozen specimens and then during surgery
The deep learning system was therefore tested for the first time on frozen tumor samples from previous brain cancer operations. It accurately diagnosed 45 of 50 cases within 40 minutes of starting genetic sequencing. In the other five cases he refrained from making a diagnosis because the information was not clear. The system was then tested during 25 brain surgeriesmost of them on children (in Utrecht) but also on adults (in Amsterdam) together with the standard method of examining tumor samples under the microscope.
The results of the experiment
The new approach provided 18 correct diagnoses and failed to reach the necessary confidence threshold in the other seven cases. D
total duration of the procedure: 60 to 90 minutes. The Princess Mxima Center has decided that the results of the technique are sufficiently reliable and he is already using it with children for whom the result can determine the surgical strategy. The Amsterdam UMC will also use the technique in daily practice, to speed up diagnosis.
Eelco Hoving, pediatric neurosurgeon and clinical director of Neuro-oncology at the Mxima Center, is enthusiastic about the possibilities of DNA analysis during surgery: During surgery, a small remnant of tumor tissue is sometimes deliberately left behind to prevent neurological damage. But if it were later discovered, for example, that the very aggressive tumor, a second surgery may still be necessary to remove the last residue. This will once again create risks and stress for patients and their families. With the new algorithm, this scenario can be avoided because we will already know during the first surgery what type of tumor we are dealing with.
Comparative research
To use the new technique in an even broader and more structured way, further research is needed. For example, more tumor types should be added to the algorithm. In this way, international standards will be respected, allowing data comparison. Furthermore, the results of the new and current (longer) method will be further compared, in collaboration with other (inter)national centres. It should clarify whether the new method will also contribute to the quality of life of patients in the long term.
Questions remain about the implementation of the model
Other medical centers have already begun applying the new method to surgical specimens, the study authors said, suggesting it may work in other people’s hands. But Dr. Sebastian Brandnerprofessor of Neuropathology at University College London, interviewed by the New York Times explains that Sequencing and classifying tumor cells often still require significant bioinformatics expertiseas well as experts who can manage, troubleshoot and repair the technology.
The implementation itself is less simple than often suggested, he said. Brain tumors are also the most suitable to be classified based on the chemical changes analyzed by the new method; not all cancers can be diagnosed this way, he adds. The new method is part of a broad movement aimed at bringing molecular precision to the diagnosis of tumors, potentially allowing scientists to develop targeted treatments that are less harmful to the nervous system. But translating deeper knowledge of tumors into new therapies has proven difficult. We have made some progress says Dr. Cohen again to the New York Times, but not many in treatment as well as in understanding the molecular profile of tumors.
October 12, 2023 (modified October 12, 2023 | 11:57)
© ALL RIGHTS RESERVED
#Brain #tumor #artificial #intelligence #find #type #operating #table #minutes