Revisit how measurements are fetched and if they should be displayed

[fedorov]

This is the relevant part of code:

dicom-microscopy-viewer/src/annotation.js

Line 511 in 020ea81

async function _fetchMeasurements ({

Need to discuss if and how it might make sense to display measurements in Slim UI, and also investigate what to do when they cannot be fetched due to size for example.

@CPBridge @DanielaSchacherer @dclunie

[dclunie]

I took a quick look at ANN in the current version of the production server for a TCGA example, and found that the metadata contained measurements:

% curl -X GET -H "Authorization: Bearer $(gcloud auth print-access-token)" https://proxy.imaging.datacommons.cancer.gov/current/viewer-only-no-downloads-see-tinyurl-dot-com-slash-3j3d9jyp/dicomWeb/studies/2.25.234091009181871968405618923996208732870/metadata | jq

...
                "00660132": {
                  "vr": "SQ",
                  "Value": [
                    {
                      "00660125": {
                        "vr": "OF",
                        "BulkDataURI": "https://proxy.imaging.datacommons.cancer.gov/bulk/current/viewer-only-no-downloads-see-tinyurl-dot-com-slash-3j3d9jyp/dicomWeb/studies/2.25.234091009181871968405618923996208732870/series/1.2.826.0.1.3680043.10.511.3.17792724052651219846150671595663920/instances/1.2.826.0.1.3680043.10.511.3.11732146924117907800700667565204709/bulkdata/006A0002/0/00660121/0/00660132/0/00660125"
                      }
                    }
                  ]
                }
...

and had no trouble retrieving the bulkdata containing the OF values:

% curl -o /tmp/crap.dat -X GET -H "Authorization: Bearer $(gcloud auth print-access-token)" https://proxy.imaging.datacommons.cancer.gov/bulk/current/viewer-only-no-downloads-see-tinyurl-dot-com-slash-3j3d9jyp/dicomWeb/studies/2.25.234091009181871968405618923996208732870/series/1.2.826.0.1.3680043.10.511.3.17792724052651219846150671595663920/instances/1.2.826.0.1.3680043.10.511.3.11732146924117907800700667565204709/bulkdata/006A0002/0/00660121/0/00660132/0/00660125

% hexdump -C /tmp/crap.dat | more
00000000  2d 2d 34 63 64 61 31 34  35 31 66 37 36 37 32 61  |--4cda1451f7672a|
00000010  34 33 34 66 37 33 39 36  65 37 63 33 31 62 37 34  |434f7396e7c31b74|
00000020  63 31 66 37 39 66 65 33  62 66 36 36 38 61 36 61  |c1f79fe3bf668a6a|
00000030  33 30 36 65 32 63 35 62  36 66 65 35 32 66 0d 0a  |306e2c5b6fe52f..|
00000040  43 6f 6e 74 65 6e 74 2d  54 79 70 65 3a 20 61 70  |Content-Type: ap|
00000050  70 6c 69 63 61 74 69 6f  6e 2f 6f 63 74 65 74 2d  |plication/octet-|
00000060  73 74 72 65 61 6d 3b 20  74 72 61 6e 73 66 65 72  |stream; transfer|
00000070  2d 73 79 6e 74 61 78 3d  31 2e 32 2e 38 34 30 2e  |-syntax=1.2.840.|
00000080  31 30 30 30 38 2e 31 2e  32 2e 31 0d 0a 0d 0a dc  |10008.1.2.1.....|
00000090  9a 44 41 a4 ba 04 40 f4  fa df 40 f5 85 46 43 9e  |.DA...@...@..FC.|
000000a0  c4 98 43 85 97 b0 42 8a  7d eb 42 f3 dd 78 40 0d  |..C...B.}.B..x@.|
...

FYI, the Items of Measurements Values Sequence (of which there can be multiple) (a) has a float payload and (b) can be one measurement for each of every annotation, or one for each of a specified list

[dclunie]

% curl -o /tmp/crap.dat -X GET -H 'Accept: application/octet-stream' -H "Authorization: Bearer $(gcloud auth print-access-token)" https://proxy.imaging.datacommons.cancer.gov/bulk/current/viewer-only-no-downloads-see-tinyurl-dot-com-slash-3j3d9jyp/dicomWeb/studies/2.25.234091009181871968405618923996208732870/series/1.2.826.0.1.3680043.10.511.3.17792724052651219846150671595663920/instances/1.2.826.0.1.3680043.10.511.3.11732146924117907800700667565204709/bulkdata/006A0002/0/00660121/0/00660132/0/00660125

also works to fetch a single part response that doesn't require extracting from its MIME container:

% hexdump -C /tmp/crap.dat | more
00000000  dc 9a 44 41 a4 ba 04 40  f4 fa df 40 f5 85 46 43  |..DA...@...@..FC|
00000010  9e c4 98 43 85 97 b0 42  8a 7d eb 42 f3 dd 78 40  |...C...B.}.B..x@|
00000020  0d ad 92 43 b8 bc 50 43  bc 1a 20 41 91 40 24 41  |...C..PC.. A.@$A|
00000030  0a af 0e 41 e4 ba cd 3f  8d e9 ee 3f 19 d2 39 40  |...A...?...?..9@|

% hexdump -e '1/4 "%4f\n"' /tmp/crap.dat | more
12.287807
2.073892
6.999384
198.523270
305.536072
...

[fedorov]

Another consideration for adding support of measurements is its impact on the memory footprint - even without supporting measurements, we already have some instances in IDC where we run out of memory just trying to display annotations.

[DanielaSchacherer]

Another consideration for adding support of measurements is its impact on the memory footprint - even without supporting measurements, we already have some instances in IDC where we run out of memory just trying to display annotations.

Is there still room / ideas to make the code retrieving and managing the annotations more efficient or are we already out of options there?

[DanielaSchacherer]

In general I think it's valuable to be able to display measurements accompanying the annotations in the future (maybe when hovering over them?).

[fedorov]

Is there still room / ideas to make the code retrieving and managing the annotations more efficient or are we already out of options there?

I don't think we have any good idea on the table. I think the main issue is that the entire array needs to be fetch, both for the points and measurements, before anything can be done, and this introduces latency (it is not unusual for some large annotations load to add in the order of tens of seconds latency) and memory pressure. You cannot selectively fetch only the annotations needed for a specific region, unless we fetch that array in chunks (assuming that is possible at all) and build something client-side lookup entity. Definitely we can/should discuss again.