About group

The amount of image data produced in medicine is growing very rapidly. Many previously analogue modalities (e.g. X-ray) now provide digital data, modalities providing 3D data (e.g. MRI or CT) are already routinely used in everyday clinical practice and their resolution is increasing year by year, thus increasing the size of the data produced.

Today, the speed with which a patient learns the result of an examination is no longer limited by technology, but by the availability of radiologists. Many hospitals around the world routinely send their data to China and India. The group is developing tools to make doctors' jobs easier and faster, for example by highlighting changes since the last scan or highlighting areas of potential interest. In the future, the computer will then be able to make a diagnosis completely independently, and in some fields it already has better results than the average doctor

Even in biology, the volume of data generated is growing at a dizzying pace - 3D microscopy is becoming more common, resolution is increasing, and robotic workstations can automatically prepare and image slides. However, it is not humanly possible, for example, to look at all the cells on a single slide and determine whether each one contains a parasite or a genetic anomaly. 

It takes a person months to draw nerve fibres in a piece of tissue smaller than a millimetre, so there is also great potential for computer algorithms that could make the evaluation of scanned data much faster. The advantage is that, unlike medical applications, absolute accuracy is often not required here.

What are we working on?

  • Analysis of ultrasound images to predict the development of atherosclerosis
  • Predicting the success of artificial insemination from microscopic images and videos
  • Analysis of the effect of substances on cell culture from microscopic images for the development of new drug substances
  • Algorithms for segmentation of nerve fibres (axons and dendrites) in 3D electron and optical microscopy images (SEM-FIB, OCT, 2-photon microscopy) and for mutual registration (comparison) of these images
  • Algorithms for image registration, i.e. for finding the geometric transformation between two images of the same or similar object, which is used to detect motion or changes and to compare images acquired by different instruments or at different times.
  • Research on new similarity criteria for registration and methods for estimating registration accuracy
  • Fast and robust methods for localizing thin ultrasound tools in 3D ultrasound sequences
  • Numerical methods for ultrasonic elastography using a conventional ultrasonic instrument,making it possible to determine the mechanical properties of tissues, which can be usedin tumor detection
  • Detection of pulmonary nodules from computed tomography images as lung cancer prevention
  • Reconstruction algorithms for parallel magnetic registration used to increase scanning speed
  • Spatial reconstruction of brain activity from EEG and MEG measurements
  • Automatic generation of breathing motion models to compensate for these movements for radiation therapy
  • Fast algorithms for 3D medical image segmentation
  • Diagnosis of malaria from microscopic images of blood
  • Quantification of cardiac angiography opacity for evaluation of cardiac perfusion
  • Automatic quantitative evaluation of DNA analysis by gel electrophoresis
  • Spatial reconstruction from monocular colonoscopy video for colon examination

Responsible person Ing. Mgr. Radovan Suk