bids-specification
bids-specification copied to clipboard
bids-standard
BEP032 Modality, Datatype, and Suffix Name(s)
Brief summary of the discussion started here:
The modality for BEP032 is to be called ephys, an abbreviation of “Electrophysiology”.
A proposal has been made to change it to something more specific.
Counterarguments to the vanilla name:
- (C1.) Other modalities which we have, i.e. EEG, can be considered subsets of electrophysiology. There may be additional counterarguments which I can't parse properly, @dorahermes if so please feel free to list.
Arguments for sticking to the vanilla name:
- (A1) Electrophysiology is a well consolidated field which is distinct from encephalography. Point in case being that the BEP document considers the modality shorthand
ephysto be unambiguous (e.g. notXephys) - (A2) Rather than clarifying, an additional prefix may compromise the integrity of the modality as presently defined in BEP032.
- (A2.1) The original proposal, “Animal Electrophysiology”, while coherent with the BEP title, needlessly constrains the scope — by species, unlike any other modality. The techniques covered by the BEP have been applied in humans and ex vivo.
- (A2.2) The updated proposal, “Microelectrode Electrophysiology” uses terminology not at all present in the BEP.
@yarikoptic @bids-standard/bep032
In the past we have used Wikipedia as a more neutral source of definitions to clarify potential terminology discrepancies between our respective neuroscience subdomains.
When I look at the https://en.wikipedia.org/wiki/Electrophysiology lemma, would say that "classic electrophysiology" as listed there with the three principle electrode types matches the use as considered in BEP032. However, it just as well matches any of the 25 other "ExG" methods that is listed there.
The wikipedia lemma also mentions four "principal preparations". That makes me realize that "intracranial electrophysiology" (as considered somewhere under #1705) is not a good term, as invertebrates do not have a cranium.
(Sorry, no solution here yet; just some thoughts.)
I agree with @robertoostenveld, was typing at the same time, listing the counterarguments to the 'vanilla'/'overarching' term:
-
C2. Future backwards compatibility. In case of disagreements, wikipedia is a neutral source, and electrophysiology is a general term that includes >20 more modalities than are captured in the BEP, see here for an extensive list, including e.g. EEG, iEEG, EMG, ECG and EOG. Therefore, using electrophysiology as a modality name for one of the 20 subsets of data types is simply not correct.
-
C3. Consistency with other BIDS use cases. For example, anat, dwi, pet, func etc are all imaging types, and all use more descriptive terms. None of these state 'imaging' as the description. Say, if calcium imaging were to come in with a new BEP, it should have a more specific description than 'imaging'.
Maybe @TheChymera can propose some alternatives? (imho, the choice for a suffix (_ephys) is independent of the Modality Name that is discussed here)
Hi guys,
What about "Extracellular electrophysiology" or "ecephys" for short?
Hi guys,
What about "Extracellular electrophysiology" or "ecephys" for short?
it's too restrictive, BEP032 also supports intracellular electrophysiology (patchclamp & co)...
Thanks for launching this discussion! I agree with the aforementioned arguments, "electrophysiology" is too broad, "animal electrophysiology" not enough...
I don't have a good suggestion at this time, but two question related to the "microelectrode electrophysiology" proposal:
- is there a definition of "microelectrode" somewhere that we use to support this option? EDIT: there's one on wikipedia (https://en.wikipedia.org/wiki/Microelectrode); is it satisfactory? (from what I see, it's simple and seems adequate, no?)
- what's the status of BEP026 ("microelectrode recordings")? we'd discussed a year ago (at that time, the activity around BEP026 was pretty much stopped if I remember correctly) to try to see whether BEP032 could actually cover what was proposed in BEP026...
@dorahermes
the choice for a suffix (_ephys) is independent of the Modality Name
As BIDS currently states: “When applicable, the modality is indicated in the suffix” I assume that would be the case here.
propose some alternatives
I am still proposing that we just call it “Electrophysiology”, I don't think there's any ambiguity. If we ever get different variations called electrophysiology but necessitating separate organization, we can split it then according to the specific scopes at hand.
@robertoostenveld
In the past we have used Wikipedia as a more neutral source of definitions to clarify potential terminology discrepancies between our respective neuroscience subdomains.
Well, I'm not fully certain that that's a good idea. Wiki is a collaborative encyclopedia... as such it's more of an aggregation of eclectic terminology than an effort of systematization. Point in case being the article describing electrophysiology which you linked:
- Starts off by describing what we mean by electrophysiology here in the field of neuroscience.
- Then adds that it “may” “also” be used to refer to a broader category.
- Later it expands on all sorts of modalities which may be considered part of the electrophysiology superset — and which are clearly not useful as BIDS “modalities”. This is not under the heading “Electrophysiological modalities” but rather “Electrographic modalities”, but then the paragraph body mentions that this may not be a good term 🤔
I am not sure wikipedia has really helped us here.
Sorry to chime in late. I might be missing something, but using ephys for animal or microelectrode electrophysiology (as suggested in https://github.com/bids-standard/bids-specification/tree/bep032/src/schema/objects/modalities.yaml#L38-40 ) does not make much sense, it is too broad. It is as if there was a bids directory but only used for MRI (because MRI is a brain imaging modality).
@TheChymera, interestingly, the question seems NOT to be the one you raised before:
@dorahermes @yarikoptic what other types of electrophysiology are in BIDS?
Quite to the contrary, IMHO, the question is what other ephys modalities do not have BIDS data type and/or modality yet.
Let me give an example. Until the EMG-BIDS (issue #1371) goes into effect, EMG data should go under the beh directory with _beh suffix.
I think everyone would agree that electromyography is an Electrophysiology modality.
Let's assume that BEP032 goes to the spec while EMG-BIDS does not. Then having an ephys directory but forcing EMG to the beh directory does not make sense. You can replace EMG with any of the ExG modalities that are not represented.
I would gladly support a catch-all ephys data type and modality to represent all the electrophysiology modalities.
I consider "Microelectrode Electrophysiology" to be good term. It is not perfect, but we also have to consider the 80/20 principle. Contrary to "Electrophysiology" it does not lead to the risk of confusing future data curators about whether their EEG or IEEG data should go under those, or under the umbrella "Electrophysiology" term. And an upcoming EMG extension would also not be confused with "Microelectrode Electrophysiology". I believe that micro versus macro defines the distinction quite well.
(S0) Ok.... while the notion of confusing EEG with electrophysiology is new to me, I see there is a considerable majority in favour of adding a more specific prefix. I would still urge you to consider if any of this is really necessary.
Three additional issues, if this is what you really want:
(S1) The name... “Microelectrode Electrophysiology” (note emphasis) verges on a pleonasm. Electrode electrophysiology... I would propose instead “Microelectrode Recording” or “Microelectrode Physiology”.
(S2) In any case, if we go this route, we should update the modality shorthand which would also be the suffix stub. If “Electrophysiology” is confusing, so is ephys. If upon considering this, ephys is deemed unambiguous, then we should go back to the original proposal and stick with the non-prefixed modality name and shorthand.
(S3) whatever we rename it to, at the end the BEP document should be updated with our choice, since thus far it just refers to “Electrophysiology”, probably for historical reasons relating to point (S0).
I would not object to "micro electrophysiology" without the electrode in between.
I don't see an problem with the suffix ephys. As argued before we don't have a one-to-one mapping between modalities and suffixes, and that does not seem to cause confusion so far (like "blood" for PET is not confused with func/BOLD).
I would prefer "microelectrode electrophysiology" (although I find it somewhat lengthy). It is a shortcut for "electrophysiology using microelectrodes", which is exactly what it is about. The two terms are immediately recognizable and unambiguous, whereas without "electro" they are not.
However, if there is a strong aversion against having "electro" twice, I would suggest "Microelectrode Neurophysiology"
I also like "ephys" as suffix, it is a common short term for such recordings. But maybe we should make sure that this isn't the case in the EEG field.
Disclaimer: The following are only my views as a junior scientist working on human Electrophysiology (ExG). Take it for what is worth.
Considering the literal meaning of electrophysiology (electro + physiology), studying the electrical features of biological entities, I agree that both electrophysiology and ephys terms should be used by the broadest definition that includes most (if not all) electrophysiology data.
Also, BIDS Common Principles differentiates between data type and modality, with the latter being a subset of the former. For example, the anat data type for structural imaging supports T1w, T2W, angio, etc data types (source: MRI-BIDS). With this distinction, the description of ephys data type and modality could be:
- Considering
ephysas a data type means that all electrophysiological data that does not currently have a specific data type. - Considering
_ephysas a modality does not make much sense since it is very broad, but it could also mean any modality that has not yet been defined. For example, EMG-BIDS currently does not consider fine wire and/or intramuscular EMG. Such a dataset must go into theephysdirectory with a_ephysmodality/suffix.
What does this mean for BEP032:
I am not an expert on inter/extracellular ephys at all. I believe ephys can be used by BEP032 as a data type, but ephys can't be exclusive to BEP032. Also, since BEP032 is about a specific electrophysiology modality, _ephys should not be used as a modality for BEP032.
Thanks.
I understand this may be a somewhat controversial suggestion at this stage, but I believe the current debate is actually indicative of a deeper issue. It seems practical to consider splitting BEP 32 into two separate modalities: "extracellular electrophysiology" (ecephys) and "intracellular electrophysiology" (icephys). These are the terms currently used in NWB, and they have proven effective with minimal confusion about what they are referring to.
Currently, the spec mixes properties unique to each modality, requiring us to designate fields as optional that could be mandatory if the modalities were distinct. By splitting them, we can tailor each specification to its respective procedures more effectively. For example, specific icephys procedures such as those used in current/voltage clamp techniques, which are seldom applicable to ecephys, could be addressed more directly. ecephys often necessitates details on multielectrode probe layouts, electrode localization relative to various coordinate spaces, and surgical information—factors not typically relevant to icephys.
The fundamental nature of the experiments, their objectives, the types of data collected, and the data processing methods differ significantly between these two fields. Considering these differences, ecephys shares more similarities with ECoG/iEEG than it does with icephys.
It seems practical to consider splitting BEP 32 into two separate modalities: "extracellular electrophysiology" (ecephys) and "intracellular electrophysiology" (icephys).
I agree, intracellular and extracellular elecrophysiology recordings are quite different in the metadata and experimental designs that it makes sense to treat them separately.
@twachtler
The two terms are immediately recognizable and unambiguous [...]
The issue we are facing is that we have now tentatively established a majority opinion that “Electrophysiology” can be done without microelectrodes. For the proposed term “Microelectrode Electrophysiology” to make sense we would need to also establish that “Microelectrodes” can be used for data acquisition other other than electrophysiology, otherwise the second term is redundant. A case could be made that microelectrodes can also be used for voltammetry ... normally I would have agreed that this isn't ephys, but if our systematization holds that electroencephalography is electrophysiology, I am unsure why voltammetry wouldn't be.
I also like "ephys" as suffix
While BIDS uses shorthand/abbreviations, nowhere does it happen that a prefix to more general methodology is dropped from the abbreviation. Due to english word order it's the first word which gives the most specificity, so if you had to pick one from “microelectrode physiology” you'd be better served by the first. An example would be “T2 weighted image”, which we abbreviate as T2w, dropping the imaging part and losing fairly little information, compare that to abbreviating it to wi for weighted imaging. Additionally, for multiple-word abbreviations, the general precedent is to concatenate the initials.
I suspect what is happening here is that nobody is actually confused about what ephys means in this context (hence the preference for the unspecific abbreviation), but we've gotten into this broader disambiguation/systematization effort which I continue to feel adds fairly little.
@bendichter
It seems practical to consider splitting BEP 32 into two separate modalities: "extracellular electrophysiology" (ecephys) and "intracellular electrophysiology" (icephys)
It also seems practical to me, however, it does not really address this specific issue. The discussion started with the claim that “electrophysiology” as a modality name is underdetermined since EEG (a separate modality) can also be considered part of electrophysiology. Sadly the more specific term “extracellular electrophysiology” remains just as underdetermined in that regard. If we consider EEG part of “electrophysiology” it would also make EEG part of “extracellular electrophysiology” (since it is extracellular).
I think that for anyone in the field it would be clear that extracellular electrophysiology doesn't refer to EEG, even though "technically" also EEG is extracellular :)
It seems practical to consider splitting BEP 32 into two separate modalities: "extracellular electrophysiology" (ecephys) and "intracellular electrophysiology" (icephys)
It also seems practical to me, however, it does not really address this specific issue.
I agree with both comments.
@bendichter maybe worth either opening a separate issue for this so as to not distract from the main topic.
FYI I had created the ephys label before this conversation ever started and I am happy to change it to whatever people agree to in the end.
I promise this is not trolling. 😉
@alejoe91 agreed, and it certainly isn't intracellular anything... It therefore follows that it should be clear in the context of the field that “electrophysiology” also doesn't refer to EEG. This is in fact demonstrated by the fact that in writing BEP032 not one of the numerous authors thought to include any disambiguation. Yet we now have numerous posts supporting a systematization based on how the term is presented on e.g. wikipedia and not on how experts use it in the field.
This is in fact demonstrated by the fact that in writing BEP032 not one of the numerous authors thought to include any disambiguation. Yet we now have numerous posts supporting a systematization based on how the term is presented on e.g. wikipedia and not on how experts use it in the field.
@TheChymera, I find these types of comments which is becoming a tone on this issue borderly not ok (here, comparing what numerous experts say in a BEP to what numerous posts support based on a point in Wikipedia, and not how experts use it in the field). BIDS is a community of experts, and we should respect each other's perspectives and expertise.
What @robertoostenveld mentioned was a "lemma," an indication; it is not proof. Wikipedia is a community-driven effort, much like BIDS, but on a much larger scale, so I think it is a good tool to check if our terminology makes sense on a broader horizon. If this "lemma" does not look ok, we should criticize it with evidence. For example, we could search for "electrophysiology" in Google Scholar or Web of Science and see what types of research are being represented by this term.
Looking at Web of Science for articles that used "electrophysiology" in their title and sorting based on the number of citations (link to the search results, requires paid access), the first article is on intracellular recordings of pyramidal cells (cited 1633 times), the second is on EEG in memory and language processing (cited 1501 times), and the third is on cardiac electrophysiology, ECG (cited 1162 times). The same diverse trends continue through the search results (my favorite title is the classic Electrophysiology of Cognitive Processing by Hillyard, 1983).
This result indicates how experts use the term electrophysiology in their scientific domains (although using the Wikipedia lemma was much simpler). Based on the results, most of the underlying data for these papers are within the BIDS scope. So, we should be mindful of other fields that use similar terminology to avoid future ambiguity.
The BIDS community is substantial yet very small compared to the scientific community that could/should be represented. Even if 20 experts (which is impressive) did not mention a point once in a BEP document, that does not reduce the validity of the point.
The elegance of the BEP process and then presenting the BEP to the community as a pull request is to seek expertise outside of the highly specialized community that prepared the BEP. I don't think anyone expects experts in other fields to put their voluntary time on an unrelated BEP, but the hope is that the broader BIDS community provides its feedback on the pull requests.
I like extra- and intracellular. The "cellular" communicates that we are observing and considering the electrophysiological phenomena at the micro level rather than the macro level. The microelectrode then does not need to mentioned.
TR;DR: I think, ideally similarly to "Common derivatives", we should eventually aim for "Common Electrophysiology" to describe common entities and metadata fields. For this BEP we might want to make it indeed "Cellular Electrophysiology" and cellephys/ for this ATM.
Kudos to @robertoostenveld to be able to generalize above discussion in arriving to the common denominator of "cellular"! ;-)
EEG, iEEG are also electrophys but were established before this attempt at generalization. BEP032 is heavily based on iEEG which is an indicator of commonality between ephys and iEEG and others.
But even if we come up with some very well generalizable way to describe all electrophysiology data, it does make sense for a user to be able to quickly tell the underlying specific (sub)modality. As wikipedia hints -- some (sub)modalities acquired their name based on the body part + measuring technique combination, e.g. EEG and iEEG. For some we do not have defined concise abbreviations and no body part association per se (intracellular, extracellular).
So, insofar ephys/ seems to be the "kitchen sink" for all those.
Would we find any sensible additional "specializations" which we could use solely for descriptive purposes while overall maintaining otherwise the generalized approach to composition of data files/metadata?
I do not think that BIDS should become the "ontology" or even a "controlled vocabulary" defining body. We should find some which would provide us desired level of description. Unfortunately on schema.org I found only https://schema.org/MedicalImagingTechnique as an enum over some "datatypes" (such as MRI) we care about.
In DANDI we pretty much adopted the groupping NWB came up with: https://github.com/NeurodataWithoutBorders/nwb-schema/tree/dev/core -- icephys, ecephys, ophys (BIDS micr/ although could be with "functional" pretty much AFAIK), etc as "suffixes". As .nwb files could potentially contain multiple, we composed them (so could have _icephys+ecephys in case both intra- and extra- cellular recordings present in the same nwb). But what indeed brings it all together is "cellular" and "electrophysiology".
edit: statistics on suffixes we use across dandisets in DANDI
Overall across all
dandi@drogon:/mnt/backup/dandi/dandisets$ for d in 0*; do find $d -iname sub*_*.nwb | sed -e 's,.*_,,g' -e 's,\.nwb$,,g' | grep -v - | sort ; done | sort | uniq -c | sort -n -r
38219 ophys
11108 icephys
8459 image
6958 behavior+image
6306 ecephys
4012 behavior
3467 image+ophys
2205 behavior+ecephys
2010 behavior+image+ophys
1612 behavior+ophys
1440 behavior+image+ogen
843 ecephys+ogen
404 behavior+ecephys+ogen
404 behavior+ecephys+image
253 ecephys+image
230 behavior+image+ogen+ophys
187 behavior+icephys+ogen
125 behavior+task
124 spikesorting
106 icephys+image
74 behavior+ecephys+image+ogen
40 ogen
32 behavior+icephys
26 ecephys+ophys
2 ecephys+icephys
Detailed per dandiset which has them:
> $ for d in 0*; do dts=$(find $d -iname sub*_*.nwb | sed -e 's,.*_,,g' -e 's,\.nwb$,,g' | grep -v - | sort | uniq -c); [ -z "$dts" ] && continue; echo -e "$d:\n$dts"; done
000003:
97 behavior+ecephys
4 ecephys
000004:
87 ecephys+image
000005:
18 behavior+ecephys
9 behavior+icephys
121 behavior+icephys+ogen
000008:
1328 icephys
000009:
36 behavior+icephys+ogen
21 ecephys
116 ecephys+ogen
000010:
14 behavior+ecephys
85 behavior+ecephys+ogen
59 behavior+ophys
000011:
1 behavior
91 behavior+ecephys+ogen
000012:
297 icephys
000013:
22 behavior+icephys
30 behavior+icephys+ogen
000020:
4435 icephys
000021:
182 ecephys
000022:
143 ecephys
000023:
318 icephys
000028:
2 ecephys
000029:
3 behavior+ecephys
000034:
4 ecephys
000035:
185 icephys
000036:
57 behavior+image+ophys
000037:
100 behavior+image+ophys
50 behavior+ophys
000039:
100 behavior+ophys
000041:
22 ecephys
000043:
94 icephys
000044:
8 behavior+ecephys
000045:
595 behavior
182 behavior+ecephys
5822 behavior+image
11 image
000048:
1 ecephys+ophys
000049:
78 behavior+ophys
000050:
56 behavior+ophys
000053:
339 behavior
20 behavior+ecephys
000054:
79 behavior+ophys
6 ophys
000055:
55 behavior+ecephys
000056:
40 behavior+ecephys
000059:
50 behavior+ecephys
50 ecephys
000060:
98 behavior+ecephys+ogen
000061:
40 ecephys+image
000065:
1 behavior+ecephys
000067:
10 behavior+ecephys
18 ecephys
000068:
1 icephys
000069:
1 behavior+icephys
000070:
9 behavior+ecephys
000109:
350 icephys
000114:
14 ecephys
16 ophys
000115:
57 behavior+ecephys
000117:
197 icephys
000126:
4 icephys
000127:
1 behavior+ecephys
1 ecephys
000128:
1 behavior+ecephys
1 ecephys
000129:
1 behavior+ecephys
1 ecephys
000130:
2 ecephys
000138:
1 behavior+ecephys
1 ecephys
000139:
1 behavior+ecephys
1 ecephys
000140:
1 behavior+ecephys
1 ecephys
000142:
717 icephys
000144:
2 ophys
000147:
10 ecephys
000148:
46 icephys
000149:
4 behavior+ecephys+image
000165:
572 behavior+ecephys
000166:
19 ecephys
000167:
1 image+ophys
4 ophys
000206:
1 behavior+ophys
000207:
19 ecephys
000209:
291 icephys
000212:
1013 behavior
000213:
67 behavior+ecephys
000218:
74 behavior+ecephys
24 ecephys
000219:
31 behavior+image+ophys
30 behavior+ophys
1 ophys
000220:
34 icephys
000221:
263 ecephys+ogen
000223:
20 ecephys+ophys
000226:
60 behavior
000228:
91 icephys
000230:
9 behavior+ecephys
000231:
44 behavior+ecephys+image
71 behavior+image
000232:
130 behavior
49 behavior+ecephys
000233:
345 ecephys
000234:
19 ophys
000235:
8 behavior+ophys
000236:
9 behavior+ophys
000237:
8 behavior+ophys
000238:
6 behavior+ophys
000239:
180 behavior
000240:
6 ophys
000242:
6 behavior+ecephys
000244:
33 ophys
000245:
25 icephys
000246:
1027 ophys
000248:
66 ecephys
12 ogen
000249:
777 ophys
000250:
3 behavior
000252:
12 ecephys
000253:
84 ecephys
14 ogen
000254:
1 behavior+ophys
000288:
36 icephys
000289:
221 behavior+ophys
000291:
1 behavior
000292:
11 icephys
000293:
121 icephys
000294:
2 ecephys+icephys
000295:
26 icephys
000296:
1278 ophys
000297:
118 icephys
000299:
1 ecephys
000301:
14 behavior+ecephys
000302:
11 behavior+ecephys
21 behavior+ecephys+ogen
000336:
314 image+ophys
2120 ophys
000337:
21 icephys
000339:
66 ecephys
000341:
681 icephys
106 icephys+image
000342:
2 behavior
000345:
10 ophys
000347:
1 image+ophys
8 ophys
000350:
12 ophys
000352:
17 behavior
000353:
4 behavior
000354:
12 behavior
000355:
7 behavior
000356:
5 behavior
000357:
6 behavior
000358:
12 behavior
000360:
1 image
000361:
32 behavior
000363:
5 behavior+ecephys
1 behavior+ecephys+image
70 behavior+ecephys+image+ogen
98 behavior+ecephys+ogen
000397:
3 ecephys
000398:
42 ecephys
000399:
105 ecephys
000402:
19 behavior+image+ophys
000407:
156 icephys
000408:
228 ecephys
000409:
347 behavior
24 behavior+ecephys
305 behavior+ecephys+image
1 ecephys+image
000410:
22 behavior+ecephys
000411:
1 ecephys
000447:
5 behavior+ecephys
000448:
18 image
000454:
4 ophys
000458:
24 behavior+ecephys
000459:
42 behavior+image+ophys
000461:
14 ophys
000462:
14 ophys
000463:
29 ecephys
000465:
36 ecephys
000467:
14685 ophys
000469:
41 ecephys+image
000470:
1 ophys
000471:
61 behavior
000472:
10 ophys
000473:
1 behavior+ecephys+ogen
24 ecephys+ogen
000477:
77 behavior+ophys
000481:
1 ecephys
000482:
1 ecephys
000483:
128 behavior+ophys
000484:
29 behavior
000485:
37 behavior
000486:
36 behavior
000488:
43 behavior+image+ophys
000489:
18 image
000491:
14 ophys
000493:
3 ecephys
1 ecephys+ogen
000526:
2 ophys
000527:
8 ophys
000528:
5 ophys
000529:
27 ecephys
000533:
141 behavior+ophys
000535:
115 behavior+ophys
000537:
125 image
000538:
11 image
000540:
495 behavior+image
000541:
21 ophys
000544:
2 behavior+ecephys+ogen
000546:
1 ecephys
000547:
70 ophys
000548:
19 image
000549:
26 image
000550:
17 image
000552:
104 behavior+ecephys
13 ecephys
000553:
4 ecephys
000554:
36 ecephys
000559:
111 behavior+image
1440 behavior+image+ogen
230 behavior+image+ogen+ophys
153 behavior+image+ophys
5 image+ophys
3 ophys
000562:
7 ecephys
72 ophys
000563:
81 ecephys
14 ogen
000565:
21 ophys
000566:
1 ophys
000568:
2 behavior+ecephys
50 behavior+ecephys+image
4 behavior+ecephys+image+ogen
000569:
103 icephys
000570:
155 icephys
000572:
289 ecephys+ogen
000573:
1 behavior
000574:
45 behavior+ecephys
000575:
18 behavior+ecephys
000576:
9 ecephys
9 ecephys+image
000579:
300 behavior+ophys
8 ophys
000580:
116 ophys
000582:
118 behavior+ecephys
000617:
889 image+ophys
992 ophys
000618:
124 ecephys
000620:
48 behavior+task
47 ecephys
47 spikesorting
000623:
29 behavior+ecephys
000624:
8 behavior
19 image
000625:
3 behavior+ophys
000628:
679 behavior
679 image
000629:
114 behavior+ecephys
000630:
210 icephys
000631:
15 image
000632:
24 image
000633:
2 image
000634:
10 image
000635:
81 icephys
000636:
706 icephys
000637:
292 ecephys
000673:
44 ecephys+image
000675:
3 ecephys
000681:
57 behavior+ophys
000683:
1 ophys
000684:
3 behavior+ecephys
000685:
1 ecephys
000686:
18 image
000687:
11 ecephys
000688:
111 behavior+ecephys
000690:
106 ecephys
000691:
1 image+ophys
000692:
9 ophys
000696:
6 ecephys
000697:
15 behavior
000698:
16 behavior
000699:
17 behavior
000700:
15 behavior
000701:
15 behavior
000702:
16 behavior
000703:
22 behavior
000704:
20 behavior
000705:
20 behavior
000706:
16 behavior
000707:
16 behavior
000708:
17 behavior
000709:
29 behavior
000710:
6 ecephys
000711:
4079 image
1936 image+ophys
000713:
869 ecephys
3271 image
000714:
9 ophys
000715:
10 ophys
000717:
1 behavior+ecephys
1 behavior+ophys
1 icephys
000721:
1 ecephys
000727:
9 behavior+image+ophys
000728:
1518 behavior+image+ophys
1518 ophys
000732:
22 ecephys
000767:
14 behavior+ecephys
14 ecephys
000768:
51 icephys
000769:
177 icephys
000770:
44 behavior+task
44 ecephys
44 spikesorting
000771:
18 behavior+task
18 ecephys
18 spikesorting
000772:
15 behavior+task
15 ecephys
15 spikesorting
000773:
15246 ophys
000774:
10 ecephys
000775:
57 behavior
000776:
38 behavior+image+ophys
000777:
1 ecephys
000778:
1 ecephys
000779:
1 ecephys+image
000780:
1 ecephys+image
000781:
1 ecephys+image
000782:
1 ecephys+image
000784:
6 ecephys
000785:
41 ophys
000786:
1 ecephys
000787:
1 ecephys
000788:
1 ecephys
000789:
1 ecephys
000792:
1 ecephys+image
000793:
1 ecephys+image
000794:
1 ecephys
000795:
1 ecephys
000796:
1 ecephys
000797:
1 ecephys
000798:
1 ecephys
000799:
1 ecephys
000800:
1 ecephys+image
000801:
1 ecephys+image
000802:
1 ecephys+image
000803:
1 ecephys+image
000804:
1 ecephys+image
000805:
1 ecephys+image
000806:
1 ecephys+image
000807:
1 ecephys+image
000808:
1 ecephys
000809:
1 ecephys
000810:
1 ecephys
000811:
1 ecephys
000812:
1 ecephys
000813:
1 ecephys
000814:
1 ecephys
000815:
1 ecephys
000816:
1 ecephys
000817:
1 ecephys
000818:
1 ecephys
000819:
1 ecephys
000820:
1 ecephys
000821:
1 ecephys
000822:
1 ecephys
000823:
1 ecephys
000824:
1 ecephys
000825:
1 ecephys
000826:
1 ecephys
000827:
1 ecephys
000828:
1 ecephys
000829:
1 ecephys
000830:
1 ecephys+image
000831:
1 ecephys+image
000832:
1 ecephys+image
000833:
1 ecephys+image
000834:
1 ecephys
000835:
1 ecephys
000836:
1 ecephys
000837:
1 ecephys
000838:
1 ecephys
000839:
1 ecephys
000840:
1 ecephys
000841:
1 ecephys
000842:
1 ecephys
000843:
1 ecephys
000844:
1 ecephys
000845:
1 ecephys
000846:
1 ecephys
000847:
1 ecephys
000848:
1 ecephys
000849:
1 ecephys
000850:
1 ecephys
000851:
1 ecephys
000852:
1 ecephys
000853:
1 ecephys
000854:
1 ecephys
000855:
1 ecephys
000856:
1 ecephys
000857:
1 ecephys
000858:
1 ecephys
000859:
1 ecephys
000860:
1 ecephys
000861:
1 ecephys
000862:
1 ecephys+image
000863:
1 ecephys+image
000864:
1 ecephys
000865:
1 ecephys
000866:
1 ecephys+image
000867:
1 ecephys+image
000868:
1 ecephys
000869:
1 ecephys
000871:
284 image+ophys
000873:
6 behavior+ecephys
000875:
9 image
000876:
1 ecephys
000882:
16 ecephys
000884:
12 ecephys
000888:
52 behavior
000889:
52 behavior
000890:
125 behavior+ecephys
000891:
87 image
000892:
1 ecephys
000894:
133 ecephys+ogen
000895:
17 ecephys+ogen
000896:
5 ophys
000897:
6 behavior+ecephys
6 ecephys
000932:
238 ecephys
000933:
28 icephys
000934:
13 icephys
000935:
1 ecephys
5 ecephys+ophys
000937:
1 behavior+ecephys
000939:
21 behavior+ecephys
8 behavior+ecephys+ogen
000941:
12 behavior+ecephys
000944:
84 behavior+ophys
000945:
60 ecephys
000946:
1885 ecephys
000947:
272 ecephys
000948:
376 ecephys
000951:
459 behavior+image
36 image+ophys
000953:
20 behavior+ecephys
000957:
9 ecephys+image
000960:
1 behavior+ecephys
000962:
2 behavior+ecephys
18 ecephys
000966:
18 ecephys
I think this issue has fallen into the classic BIDS trap of using the term "modality", which is almost never productive.
We have two "interesting" concepts that get confused with modality:
- Datatype, or the directory that sits below the subject/session level
- Suffix, or the alphanumeric string that goes after the final underscore
One misleading concept:
- The
mod-<label>entity, where the label is the suffix of some other file.
And one confused concept that explains why the others are frequently confused for it:
- Modality, that thing which is neither a datatype nor a suffix. It might be a recording instrument? It's defined in the schema as a grouping of datatypes (https://github.com/bids-standard/bids-specification/blob/master/src/schema/rules/modalities.yaml) and its primary purpose at this point is to allow OpenNeuro to categorize datasets for search purposes using the output of the validator.
I think the two questions that have been addressed by various people have centered on the two interesting concepts, which is at least a good sign that we're not wasting our time with the other two:
- Is
ephys/is an acceptable datatype for intracellular and extracellular microelectrode data, given that other types of electrophysiological recordings may justifiably claim a space in this directory at some point? - Is
_ephysan acceptable suffix for intracellular microelectrode recordings and extracellular microelectrode recordings, or is something narrower needed?
If this post is about datatype, can we please update the title to make this clear? Or if it's about both, please say it's about both. In the latter case, I would suggest addressing the two questions separately.
@effigies
If this post is about datatype, can we please update the title to make this clear? Or if it's about both, please say it's about both.
The discussion started specifically with an edit on the schema/objects/modalities.yaml file. I have tried my best to keep repeating the word modality to avoid this from turning into a datatype vs. modality discussion. In truth BEP032 uses “ephys” for both the datatype, and the suffix.
I wanted to keep them together to not split it into separate discussions. While we could split the two up differently, at present ephys is the name in both fields, and in both fields it can collide with eeg (which is the collision that prompted this discussions) — which is similarly both a datatype and suffix.
Regarding the two questions you listed, I believe the answer is yes to both, for reasons stated in (A1).
@robertoostenveld
The "cellular" [...]
This might be a good solution, particularly in differentiating the discussion as suggested by @effigies . The hierarchy would then be {intra,extra}.cellular.electrophysiology. The last two terms could be used for the datatype and the modality, and all three (and more, ofc) could go into the suffix. In that context, this BEP would deal with cellular electrophysiology, while EEG would be dealing with tissue electrophysiology.
Thanks @effigies !
Just a thought: besides {intra,extra}.cellular.electrophysiology, we now also have {scalp,intracranial}.electroencephalography (where "scalp" is so far implicit in "EEG"). I like that symmetry.
Back to the topic at hand: With BIDS we in general try to avoid reinventing things, and I suspect that icephys + ecephys have in the past also had considerable discussion in the NWB team, so I think those would be suitable as suffixes here as well. As modality I would be fine with ephys.
Besides technical details in the schema, there is also to consider how we phrase things in the text of the specification and how we guide people (including novices in BIDS and people from tangent domains) through the specification, starter kit, etc. The words that appear under the menu item "Modalities" on the left of https://bids-specification.readthedocs.io/en/stable/ are relevant for that. There I would refrain from the too general "Electrophysiology" and prefer "Micro Electrophysiology" or "Microelectrode Electrophysiology". That then sits at the same level as "Magnetic Resonance Imaging".
Ok, so given the now expanded scope where we differentiate datatype, modality, and suffix, how about:
Datatype: cephys — Cellular Electrophysiology
Modality: cephys — Cellular Electrophysiology
Suffixes: {icephys, ecephys} — {Intra, Extra}cellular Electrophysiology
Illustratively:
dataset_description.json
tasks.tsv
tasks.json
participants.tsv
sub-A/
ses-20220101/
cephys/
sub-A_ses-20220101_task-nosepoke_ecephys.nix
sub-A_ses-20220101_task-nosepoke_ecephys.json
sub-A_ses-20220101_task-nosepoke_events.tsv
[...]
The schema and the content of the BEP032 document would need to be significantly updated to reflect this.
Cellular Electrophysiology
This name risks being misleading/confusing when recording local field potentials, which are a tissue-level, not cell-level signal, so if we go with the label "cephys" the documentation will have to be very clear that this does not imply that the recorded signals can always be attributed to an individual cell.
I would support using the terms "Microelectrode Electrophysiology / mephys". I understand the concerns about redundancy in the term, but it has the merit of not being misleading.
We're almost there!!! But the last point of @apdavison is a good one! How about the following?
- modality = "Microelectrode Electrophysiology" (the above discussion showed some consensus with this wording)
- datatype = "microephys" (I prefer
microephystomephys, which I find difficult to read/pronounce ) - suffixes = "_icephys" or "_ecephys"
dataset_description.json
tasks.tsv
tasks.json
participants.tsv
sub-A/
ses-20220101/
microephys/
sub-A_ses-20220101_task-nosepoke_ecephys.nix
sub-A_ses-20220101_task-nosepoke_ecephys.json
sub-A_ses-20220101_task-nosepoke_events.tsv
[...]
Is there a downside to:
dataset_description.json
tasks.tsv
tasks.json
participants.tsv
sub-A/
ses-20220101/
ecephys/
sub-A_ses-20220101_task-nosepoke_ecephys.nix
sub-A_ses-20220101_task-nosepoke_ecephys.json
sub-A_ses-20220101_task-nosepoke_events.tsv
[...]
and
dataset_description.json
tasks.tsv
tasks.json
participants.tsv
sub-A/
ses-20220101/
icephys/
sub-A_ses-20220101_task-nosepoke_icephys.nix
sub-A_ses-20220101_task-nosepoke_icephys.json
sub-A_ses-20220101_task-nosepoke_events.tsv
[...]
the two are rarely done in the same experiment, and when they are it should be fine to have an icephys and ecephys folder in the same session folder.
The advantage of this approach is it allows us to avoid minting a new term to act an a datatype umbrella for icephys and ecephys. I also feel that having the suffixes, data type, and modality match as much as is practical makes the standard simpler and easier to learn and understand.
I personally do not see downside: they are indeed sufficiently different and AFAIK require different preprocessing (e.g. spikesorting for ecephys). Extra benefit is that it would make it right away aligned with a potential future refactoring:
- https://github.com/bids-standard/bids-2-devel/issues/72
and would be more aligned with other modalities such as
eegandieegin such a correspondence.