MaZda was originally developed in 1996 at the Institute of Electronics, Technical University of Lodz (TUL), Poland, by Michal Strzelecki
and Piotr Szczypinski
, for texture analysis of mammograms. The statistical parameters computed by the early version of the program were derived from the co-occurrence matrix. Consequently, the name of the program is an acronym derived from 'Macierz Zdarzen' that is the Polish counterpart of the English term 'co-occurrence matrix'. Thus MaZda has no connotation with the famous Japanese car maker.
During the first COST B11 workshops in 1998, it became apparent that MaZda might be useful as a tool for carrying out quantitative analysis of magnetic resonance image texture. It was decided that it would be a successor of the NMRWin program developed at the German Cancer Research Centre (DKFZ) in Heidelberg, Germany, by Lothar Schad and Michael Friedlinger. A short-term COST-supported mission to Heidelberg was then carried out by Andrzej Materka
, Michal Strzelecki and Piotr Szczypinski of TUL to transfer the extensive image file support of NMRWin to MaZda and to further expand the list of statistical texture parameters computed by the program. At present, almost 300 parameters can be computed by MaZda for each ROI, for a given image normalization and quantization option selected. MaZda User's Manual was by Andrzej Materka. The main author of MaZda program code is Piotr Szczypinski, while the image file loader was written by Michael Friedlinger, the wavelet analysis module by Marcin Kociolek
and the feature selection module by Michal Strzelecki. Further feature analysis is performed by B11 software, also developed in terms of the COST B11 project. The B11 program code has been written by Andrzej Materka in Delphi and compiled for PC computers that use MS Windows®
9x/NT/2000/XP operating systems. It can be run from the MaZda or started independently. B11 feature analysis covers several techniques which provides feature reduction by converting input data into another space of lower dimension. It also allows for feature classification and visualisation.