A representative case from an independent internal patient cohort shows a 54-year-old male patient who had an IDH-mutant oligodendroglioma WHO grade 2 that was correctly classified by all algorithms. credit: cancer (2024). DOI: 10.3390/Cancer16061102
× close
A representative case from an independent internal patient cohort shows a 54-year-old male patient who had an IDH-mutant oligodendroglioma WHO grade 2 that was correctly classified by all algorithms. credit: cancer (2024). DOI: 10.3390/Cancer16061102
Machine learning (ML) techniques can be used to quickly and accurately diagnose mutations in glioma, a primary brain tumor. This is shown by a recent study by Karl Landsteiner University of Health Sciences (KL Krems).
In this study, canceranalyzed data from physiological metabolic magnetic resonance images and used ML to identify mutations in metabolic genes. Mutations in this gene have a major impact on the course of the disease, so early diagnosis is important for treatment.
This study also shows that currently the standards for acquiring physiometabolic magnetic resonance images are still inconsistent, hindering routine clinical use of this method.
Glioma is the most common primary brain tumor. Although the prognosis is still poor, personalized treatments can already significantly improve treatment success rates. However, the use of these advanced treatments is based on individual tumor data, which in the case of gliomas is not readily available due to their location in the brain.
Imaging techniques such as magnetic resonance imaging (MRI) can provide such data, but their analysis is complex, demanding, and time-consuming.
Therefore, the Central Laboratory for Medical Radiology at University Hospital St. Pölten, KL Krems' teaching and research hub, has been using machine learning and deep learning techniques for many years to automate and integrate such analyzes into routine clinical work. I've been developing it. Now, a further breakthrough has been achieved.
positive mutation
“Patients whose glioma cells carry a mutant form of the isocitrate dehydrogenase (IDH) gene actually have a better clinical outlook than patients with the wild type,” said Professor Andreas Stadlbauer, a medical physicist at the Central Research Institute. explains. “This means that the earlier we know the status of this mutation, the better we can personalize treatment.”
Differences in energy metabolism between mutant and wild-type tumors help achieve this. Thanks to previous work by Professor Stadlbauer's team, these can be easily measured using physiometabolic MRI, even without tissue samples. However, data analysis and evaluation is a very complex and time-consuming process that is difficult to incorporate into clinical routine, especially when rapid results are required due to poor patient prognosis.
In the current study, the team used ML techniques to analyze and interpret these data in order to obtain results more quickly and be able to initiate appropriate treatment steps. But how accurate are the results obtained? To assess this, the study first used data from 182 patients at the University Hospital St. Pölten, whose MRI data were collected according to a standardized protocol. Collected.
positive results
“We were very pleased when we saw the results of our ML algorithm. We achieved 91.7% accuracy and 87.5% accuracy in differentiating tumors with wild-type genes from tumors with wild-type genes.” explains Professor Stadlbauer. in a mutated form.
“We then compared these values to ML analysis of classic clinical MRI data and were able to show that using physiological metabolic MRI data as a basis yields significantly better results.”
However, this advantage was only maintained if the data collected in St. Polten were analyzed according to a standardized protocol. This was not the case when ML techniques were applied to external data, i.e. MRI data from other hospital databases. In this situation, ML methods trained using classical clinical MRI data proved to be more successful.
“The poor performance of ML analysis of physiometabolic MRI data here is due to the fact that the technology is still young and in an experimental development stage. Data collection methods still vary from hospital to hospital, which It leads to distortions in data collection and ML analysis,” Stadlbauer said.
But for scientists, the problem is simply one of standardization. Standardization will inevitably occur as the use of physiometabolic MRI increases in different hospitals. The time-saving evaluation of physiometabolic MRI data using ML techniques has proven itself to be an excellent technique.
Therefore, it is an excellent approach to preoperatively determine the IDH mutational status of glioma patients and individualize treatment options.
For more information:
Andreas Stadlbauer et al, Machine learning-based prediction of glioma IDH gene mutation status using physiometabolic MRI of oxygen metabolism and angiogenesis (two-center study), cancer (2024). DOI: 10.3390/Cancer16061102
Magazine information:
cancer
Provided by Karl Landsteiner University
