Real-time processing of medical image data
Modern endoscopes and microscopes as well as CTs and MRIs provide high-resolution image and video data. Fraunhofer FOKUS is developing algorithms specifically tailored to its customer’s needs for real-time processing of such data on GPUs (Graphic Processing Units).
As a result of their high degree of parallelization, GPUs are especially well-suited for execution of computationally intensive processes such as real-time image processing. Great data volumes can be analyzed swiftly and efficiently because of parallel processing of large numbers of processes. Real-time processing of high-resolution images is for instance necessary for a realistic view of minimally invasive operative areas. Fraunhofer FOKUS' software generates 360-degree panoramic views from the endoscope’s partial images that can be enriched with patient data, such as MRIs. The researchers are also assisting manufacturers of medical technology by integrating contactless user interfaces like head, view, voice, or gesture commands.
High intensity focused ultrasound (HIFU) minimally invasive surgery techniques need performant organ/tumor tracking in order to be usable in the body (outside of the head area) to compensate for the respiratory induced motion. Fraunhofer FOKUS' motion tracking and sensor fusion algorithms offer very precise tracking, with continuous mm-level accuracy.