CHANGELOG

Philipp Ehses 11.02.07, original version
[..]
Philipp Ehses 22.03.11, port to VD11
Felix Breuer  31.03.11, added support for noise & ref scans, speed fixes
Philipp Ehses 19.08.11, complete reorganization of code, added
                        siemens_data_obj class to improve readability
Wolf Blecher  15.05.12, readout of slice position parameters for VB Data sets
Wolf Blecher  11.06.12, added distinction between PATREF and PATREF PHASCOR
Philipp Ehses 02.08.12, again massive code reorganization. The new class
                        twix_map_obj.m now stores the memory position of
                        each dataset (coils are included - size: NCol*NCha)
                        The actual data is not read until it is demanded
                        by a "data_obj()" call. This makes it possible
                        to selectively read in only parts of the data
                        (e.g. to preserve memory).
Philipp Ehses 27.08.12, speed optimizations (avoiding of .-subsref calls)
07.09.12 Thanks to Stephen Yutzy for implementing support for raw data
         correction (currently only supported for VB software version).
15.01.13 Thanks to Zhitao Li for proper handling of SYNCDATA.
Philipp Ehses 28.08.13, added support for VD13 multi-raid files
Michael Völker May-Aug 15 * Better error tolerance with incomplete files.
                        * Swapped out parsing loop into a separate function
                          without access to twix object (no thousands of subsref calls).
                        * For parsing, use an as-minimalistic-as-possible loop
                          to gather all mdhs in binary form. They are all stored
                          in one array "mdh_blob".
                        * Translation of mdhs from binary to struct is vectorized
                          and almost instant. It's done in evalMDH(), replacing
                          evalMDHvb() and evalMDHvd(), no more freads inside!
                        * vectorized readMDH(), quasi-instant
                        * When parsing, actually read the entire file without jumps.
                          This is weirdly faster than fseek(), plus the entire file is
                          kept in the file system cache, if possible. Next read
                          is therefore faster, too.
                        * => Parsing speed improved by factor 3...7 or so
                        * Speed increase for reading data, esp. when slicing,
                          os-removal or reflected lines. Also for random acquisitions.