v2.2.4

• Added histogram of T-statistics from artifact detection to TICL_FR
• Added artifact detection T-statistics when hovering over post-stim EEG plot
• Added support for generating LocationSearch reports (prototype)

v2.2.3

Changes:

• Post-stim EEG plots now indicate channels that were rejected during artifact detection
• TICL_FR report now contains recall performance plot
• post-hoc classifier evaluations now use retrained classifier if the classifier has been retrained during the experiment

v2.2.1

Summary of Changes

• Added support for generating TICL_FR reports
• Begun using PTSA v2.0.0 (#237)
• Added support for generating LocationSearch experiment configs (#228)
• Made ramutils.test executable (#225)
• Updated CI testing to always build and use the conda package (#227)

v2.1.11

Summary of changes:

• Revamped algorithm for matching recall periods to deliberation periods during FR/catFR experiments
• Changed encoding sample weight parameter in light of new deliberation matching algorithm and additional data

v2.1.10

Summary of Changes:

• Updated parameters for TICL_FR ramulator configs
• Added additional diagnostic plots to stim and record-only reports
• Removed experiment name from saved classifiers since classifiers should be constant across experiments
• Removed explicit Python 2 support
• Fixed bug in deliberation period algorithm that implicitly assumed 1000HZ sampling rate

v2.1.9

This release does not include any changes that affect RAM_clinical use of the ramutils programs. For end-users the one change to note is that delta recall values in generated reports are now normalized by the baseline recall rate for that subject/experiment/session.

Summary of changes:

• Fixed logging to avoid duplicated messages #164
• Added support for broadcasting task updates over UDP socket #166
• Bugfix to identification of post-stim powers that had previously included artifact detection events #172
• Bugfix to how previous results were loaded that actually uses the given root path #157
• Documentation updates to make_report and make_aggregated_report pipelines #162
• Delta recall is now normalized by the in-session recall rate #182
• Multi-trace objects from Bayesian hmm models are now saved as part of session summary #167
• Avoid OS errors for long pathnames by not saving session numbers as part of file name for agg report data #183
• Bugfix to how stim items are identified for FR2/catFR2 #168
• Updated how PS4 reports are generated after RHINO folders were renamed to reflect the actual experiment name
• Added support for generating TICL_FR ramulator config files

v2.1.8

Summary of changes:

• Minor bugfix to return the path of the generated report instead of the html report itself
• Internal refactoring to clean up how previous results are loaded and passed around

v2.1.7

Summary of changes:

• Added support for building aggregated stim reports #149
• Added utility function for extracting metadata based on the file names in the report database #147
• Added ability to exclude sessions from reports and config generation #151
• Added ability to exclude specific contacts from classifier training #148, #150

v2.1.6

Summary of changes:

• Added support for building FR2/catFR2, FR3/catFR3 reports
• Added support for building preliminary FR6/catFR6 reports
• Updated documentation of serialized data structures
• Improved automatic montage identification based on subject alias, experiment, and session
• Bugfix to using math symbols that broke when building reports from python 2 environment
• Bugfix to allow re-localized subjects to be run through the pipeline
• Bugfix to stored data to be read using python 2 and 3 regardless of the python version used to generate the data
• Bufgfix to stored data so the dimension of the stored powers matches the dimension of the stored task events

v2.1.5

Summary of changes:

• Embed images rather than using absolute paths (#140)
• Added support for DBOY1 Ramulator config generation and using common reference
  for data collection (#141)
• Added checks for typos when specifying trigger pairs in PS5 (#142)
• Updated electrode config naming to specify NOSTIM for record-only
  experiments