The default threshold values are not corrected computed for 6 bit images or less

[vbaderks]

The JPEG-LS standard ISO/IEC 14495-1, paragraph C.2.4.1.1.1 defines how to compute the default threshold values. 2 set of formulas are defined. One needs to switch between the two sets based on the MAXVAL >= 128 or not, MAXVAL = 2 ^ bits_per_sample.

CharLS only uses the set of formulas defined for MAXVAL >= 128. This may cause incorrect encoding and decoding of images with 2,4,5 or 6 bits per sample when the default threshold values are used. The typical case is to use the default threshold values.

[vbaderks]

Impact of usage of CharLS in the DICOM domain is practical absent as almost all IOD use 8 bits per sample or more. Only the legacy secondary capture IOD could use less.