Project description

Motivation

The analysis and processing of medical images requires the visual control of intermediate results. It is important that real world coordinates are taken into account as well as pixel-accurate gray value information. Especially in medicine, image data is often available in DICOM format, which causes difficulties for conventional viewers. An easy-to-use software like miaViewer for the visualization of medical image data is used in teaching as well as in research.The analysis and processing of medical images requires the visual control of intermediate results. It is important that real world coordinates are taken into account as well as pixel-accurate gray value information. Especially in medicine, image data is often available in DICOM format, which causes difficulties for conventional viewers. An easy-to-use software like miaViewer for the visualization of medical image data is used in teaching as well as in research.

To support machine learning of neural networks, the annotation of DICOM data sets with labels is supported. The annotations are stored as grayscale images and in JSON format.

Goals and procedure

miaViewer is based on the libraries ITK, VTK and Qt/QML. It is especially adapted to the requirements of ReMIC and is continuously extended and updated. For example, methods for quick manual segmentation are necessary to make use of the medical expert knowledge by manual input of DICOM data. Further extensions such as the application of lightweight image processing algorithms are also integrated into the package.

Team ReMIC

• Prof. Dr. Christoph Palm
• Martin Zorn