search

fNIRS / snirf-samples

Sample SNIRF datasets
Other
5 stars 1 forks source link

neuro_run01.snirf and Simple_Probe.snirf both have a single measurementList for multichannel data #10

Closed kdarti closed 5 months ago

kdarti commented 5 months ago

neuro_run01.snirf and Simple_Probe.snirf both have multiple channels, but only a single measurementList, this means all of the items, e.g. detectorgain are arrays instead of scalar. As far as I can tell, this is not according to spec. image

fangq commented 5 months ago

this should have also been fixed by the updated sample files in https://github.com/fNIRS/snirf-samples/commit/3a8505f19a21749bd35f37d8ac988d5e75ca1047, after the jsnirfy patch committed last year https://github.com/NeuroJSON/jsnirfy/commit/e196ebdd12733feb43d0033b95e6d1b53fc5706a

I see the measurementList field is now a group array matching the width of dataTimeSeries.

>> a=loadh5('Simple_Probe.snirf');
>> a.nirs.data1

ans = 

  struct with fields:

      dataTimeSeries: [1200×8 double]
    measurementList1: [1×1 struct]
    measurementList2: [1×1 struct]
    measurementList3: [1×1 struct]
    measurementList4: [1×1 struct]
    measurementList5: [1×1 struct]
    measurementList6: [1×1 struct]
    measurementList7: [1×1 struct]
    measurementList8: [1×1 struct]
                time: [1200×1 double]

Please be aware that snirf users had reported significant space overhead by using "arrays of struct" form for measurementList(k) in HDF5, and there is currently a drafted PR to change this to "struct of arrays" form as a new field name measurementLists, which effectively reverts back to the sample form that you reported

https://github.com/fNIRS/snirf/pull/115

kdarti commented 5 months ago

Thanks for the quick reply and for the information about the possible change!