Modern medical image processing

Modern imaging techniques like computer tomography (CT) and magnetic resonance tomography (MRT/MRI) allow to grab three dimensional images of the inner structure of objects.

The tomographs reach resolutions of about 1mm³ or even higher and thus produce vast amounts of data. The images of the human abdomen referenced on this website typically take up around 400 megabytes of storage. (An image stack typically consists of about 800 images which have a resolution of 512x512 pixel. Due to the high dynamic range, each pixel takes up 2 bytes.)

Assessing datasets of this size manually by viewing each slice individually is tiring and not cost effective. Moreover, even an experienced viewer might miss some lesions hidden in the three dimensional image relationship that could have been made visible by an appropriate visualization.

Until recently it was rarely possible to do such visualizations due to the huge memory and processing power demands that were only met by specialized hardware. And even if such hardware was available, the processing was often not realtime capable and delivered video sequences that were rendered offline.

The Eccet software package was developed to process and visualize such huge 3D datasets and help with their interpretation. It is based on relatively cheap PC hardware which is nowadays able to satisfy the memory and processing power demands.

Developing our own system looked reasonable, as comercially available solutions didn't meet our requirements on pricing, functionality, flexibility and performance.

Eccet was developed as a visualization tool to allow us to judge the results of the filter methods developed at the department of computer science of the Heinrich-Heine-Universität Düsseldorf. Later we added (in cooperation with the univerisity hospital Düsseldorf and other research centers) filter and segmention functions, detection algorithms for medically interesting structures.
Thus Eccet now comprises a complete system for preprocessing, segmenting and visualizing three dimensional datasets.

The modular architecture allows to easily add features as well as to use multiple cooperating workstations to display multiple aspects of the same dataset at the same time.

Eccet employs multithreading (as far as the underlying algorithms allow) to utilize multiprocessor environments to full capacity.