[ad_1]
If artificial intelligence is sometimes perceived as a threat, we must not forget its many benefits. In medicine in particular, where AI seems to have a bright future, through its capacity to considerably improve the quality of medical diagnoses. The DeepGlioma rapid brain tumor screening system, developed at the University of Michigan Medicine, is a good example. This AI would be able to detect genetic mutations in cancerous brain tumors in less than 90 seconds!
The goal of l’IA is not to replace man, but to assist him, by making his task easier. Recent studies have proven how the combination of AI-based systems can help doctors predict the onset of cancers be youlung and more recently of the pancreas et of the brain.
Diffuse glioma: a common brain tumor that is difficult to accurately diagnose
There are several types of diffuse gliomas, with each type having different genetic mutations. However, the effectiveness of treatments varies depending on this genetic makeup. It has thus been proven that patients suffering from a diffuse glioma called astrocytoma could gain five years of life expectancy after complete removal of the tumor, compared to other types of diffuse gliomas!
In this context, molecular classification techniques for tumors are therefore of considerable importance. Unfortunately, current methods are difficult to access, complicating surgical decision-making and the choice of chemotherapy treatments.
The method developed by the team of neurosurgeons and engineers at Michigan Medicine[1] is therefore welcome, as it paves the way for accurate and faster identification, allowing surgeons to distinguish the nature of diffuse glioma during surgery.
DeepGlioma, an AI system that exploits rapid imaging
DeepGlioma, the idea of which emerged in 2019, is a system that combines machine learning algorithms called Deep Neural Networks (DNN) and an optical imaging method. This technique known as stimulated Raman histology, also developed at the University of Michigan, provides a real-time image of brain tumor tissue.
The effectiveness of the system was tested in a study conducted on more than 150 patients with diffuse glioma. In a paper published in Naturethe researchers explain that the new DeepGlioma system made it possible to identify mutations used by the WHO to define molecular subgroups of the disease, with an accuracy of more than 90% on average.
In a press releaseTodd Hollon, a neurosurgeon at the University of Michigan Health and first author of the study, says that “this AI-powered tool has the potential to improve access and speed du diagnostic and care for patients with fatal brain tumors ».
According to him, “DeepGlioma paves the way for accurate and faster identification that would give providers a better chance of defining treatments and predicting patient prognosis”.
CHARM: a tool with similar objectives, developed at Harvard
Researchers at the University of Michigan are not the only ones working on the subject and other work is also underway around the world.
A team from Harvard Medical School has also just presented, in the revue Med, a glioma DNA sequencing tool using artificial intelligence. This one is called CHARM, an acronym that stands for “Cryosection Histopathology Assessment and Review Machine”. The tool also has the particularity of being available for free. He is downloadable online at this address.
[1] In collaboration with New York University, University of California, San Francisco and other research institutes.
[ad_2]
Source link