Improve the level of the operating room in tumor diagnosis and AI have a way

Improve the level of the operating room in tumor diagnosis and AI have a way

In order to help the neurosurgeon more effectively in the diagnosis of brain tumors, the university of Michigan (UniVersity of Michigan)Researchers have developed a deep learning based imaging technology,Can surgery in tumor diagnosis from the process30 to 40Minutes to3Within minutes.

The technique is calledSRH(Stimulated Rama histologyFor the first time),Debut in 2017,It can help the neurosurgeon to evaluate tumor tissue faster in the operating room.Through deep learning technology and utilization of surgical resection by fresh specimens generated in the process of digital mapLike to classify tumors.

In a new oneThe nature of medical papersOf Michigan medical school,Michigan MedicineChief neurosurgeon)Todd HollonAnd the university of New York laingen health center (NYU Langone HealthAssociate professor of neurosurgeryDaniel OrringerAnd with the introduced their latest clinical trials about the technology.

HollonAs the first author of this article, introduces a way:"This is the first time in the operating roomwithAITo assess the prospective experiments, we have carried out based onAIThe workflowClinical application."

The researchers say, since the technology development for the first time,It has been as a clinical diagnosis of neurosurgery and otolaryngology tool applied to500A number of risk ofThe treatment.

The technology USESCUDA,NVIDIA RTX - the GPUandcuDNNTo speed up theTensorFlowThe depth of the studyXi convolution framework training neural network (CNN), theCNNThe output of the model classCover the U.S.90%Brain tumor diagnosis and classification accuracy94.6%.

Used in the modelCNNIs through theSRHTo obtain the415The patient250moreAll images images for training.

The senior author of the journal nature,OrringerSaid:"The results of our studyThe table of Michigan medical school9Years, the development and practice of achievements on brain tumor surgery,This method takes advantage of the advanced optics andAIDecisions to make safer in the operating room,Effective decisions.