Diagnosis of Dementia is often made too late. Magnetic Resonance Imaging (MRI) together with high-performance computing already allows the reliable quantification of brain structures to facilitate early diagnosis of brain diseases. However, due to the need for high computational power, available systems have the disadvantage of being very expensive and producing results only hours, or even days, after a patient has left the office.
mediaire, the developer of AI-based software for image analysis, has now solved that issue sustainably. Its new software-solution, mdbrain 2.2, with a normal PC with graphic card enable every radiologist to perform state-of-the-art quantitative analyses of brain MRI in less than 5 min within their own centre. As such, the analysis report is ready to be discussed with the patient during the same visit.
“Since our foundation mediaire’s aspiration was always to provide radiologists with affordable cutting-edge technology that can improve patient outcomes. Our new volumetry approach, based on Deep Learning is a major milestone along that way.” comments Dr. Andreas Lemke, CEO of mediaire.
Technical basis of this innovation are new AI-based algorithms using Deep Learning techniques. These allow more precise and faster image analysis. Up until now, it was not possible to apply Deep Learning technology to brain volumetry for a licensed medical product, as this usually requires a huge set of labelled image data. The mediaire team managed to use available data more effectively (by using unlabelled data and exploiting domain knowledge from MRI physics) and enable brain volumetry in less than 5 min on a normal PC.