237 240 per instance results by ErikGro · Pull Request #247 · BrainLesion/panoptica

ErikGro · 2026-04-10T17:18:18Z

This PR implements 237 240 respectively.

Global and averaged metrics are defined in the sample row.
Instance rows only define their instance metrics
The truncated example output looks like following:

subject_name	endplate-n_ref_instances	endplate-n_pred_instances	endplate-tp	endplate-fp	endplate-fn	endplate-prec	endplate-rec	endplate-rq	endplate-sq	endplate-sq_std	endplate-pq	endplate-sq_dsc	endplate-sq_dsc_std	endplate-pq_dsc	endplate-global_bin_volume_pred	endplate-global_bin_volume_ref	endplate-global_bin_dsc	endplate-global_bin_rvd	endplate-computation_time	ivd-n_ref_instances	ivd-n_pred_instances	ivd-tp	ivd-fp	ivd-fn	ivd-prec	ivd-rec	ivd-rq	ivd-sq	ivd-sq_std	ivd-pq	ivd-sq_dsc	ivd-sq_dsc_std	ivd-pq_dsc	ivd-global_bin_volume_pred	ivd-global_bin_volume_ref	ivd-global_bin_dsc	ivd-global_bin_rvd	ivd-computation_time	sacrum-n_ref_instances	sacrum-n_pred_instances	sacrum-tp	sacrum-fp	sacrum-fn	sacrum-prec	sacrum-rec	sacrum-rq	sacrum-sq	sacrum-sq_std	sacrum-pq	sacrum-sq_dsc	sacrum-sq_dsc_std	sacrum-pq_dsc	sacrum-global_bin_volume_pred	sacrum-global_bin_volume_ref	sacrum-global_bin_dsc	sacrum-global_bin_rvd	sacrum-computation_time	vertebra-n_ref_instances	vertebra-n_pred_instances	vertebra-tp	vertebra-fp	vertebra-fn	vertebra-prec	vertebra-rec	vertebra-rq	vertebra-sq	vertebra-sq_std	vertebra-pq	vertebra-sq_dsc	vertebra-sq_dsc_std	vertebra-pq_dsc	vertebra-global_bin_volume_pred	vertebra-global_bin_volume_ref	vertebra-global_bin_dsc	vertebra-global_bin_rvd	vertebra-computation_time
sample1	7	7	4	3	3	0.5714285714285714	0.5714285714285714	0.5714285714285714	0.7757607768614974	0.020970196469791622	0.4432918724922842	0.8735651300987373	0.013297939489101358	0.49918007434213557	15653	15446	0.6793787581594264	0.01340152790366438	0.03076695799973095	7	7	7	0	0	1.0	1.0	1.0	0.9356428989017713	0.02678689285966417	0.9356428989017713	0.9665515338708124	0.014455749243101751	0.9665515338708124	78770	77137	0.9423566613429801	0.02117012587992792	0.03775324999878649	1	1	1	0	0	1.0	1.0	1.0	0.9761063077643852	0.0	0.9761063077643852	0.9879087009935987	0.0	0.9879087009935987	45329	45836	0.9698239455931553	-0.011061174622567414	0.024147957999957725	7	7	7	0	0	1.0	1.0	1.0	0.9658619603341371	0.007220487600135381	0.9658619603341371	0.9826208051627631	0.0037474563305194635	0.9826208051627631	237614	239174	0.9631786034883428	-0.006522448092183933	0.06640583399712341
sample1-endplate_inst_0									0.799546485260771			0.8886088709677419
sample1-endplate_inst_1									0.7554438860971524			0.8606870229007634
sample1-endplate_inst_2									0.7937043282743114			0.8849890316515199
sample1-endplate_inst_3									0.7543484078137543			0.8599755948749237
sample1-ivd_inst_0																												0.9554469273743017			0.9772159131490608
sample1-ivd_inst_1																												0.932447646926368			0.9650431134933581
sample1-ivd_inst_2																												0.970298615002802			0.9849254398439722
sample1-ivd_inst_3																												0.9504240501965111			0.974581963446106
sample1-ivd_inst_4																												0.9419905213270142			0.9701288559156579
sample1-ivd_inst_5																												0.9158805031446541			0.9560935576528519
sample1-ivd_inst_6																												0.8830120283407481			0.9378718935946797
sample1-sacrum_inst_0																																															0.9761063077643852			0.9879087009935987
sample1-vertebra_inst_0																																																																		0.9528324247452002			0.9758465833232188
sample1-vertebra_inst_1																																																																		0.9676380887834632			0.9835529148368208
sample1-vertebra_inst_2																																																																		0.9703972165845173			0.9849762356715078
sample1-vertebra_inst_3																																																																		0.9572474921713285			0.9781568207394954
sample1-vertebra_inst_4																																																																		0.9744055360604648			0.9870368759243838
sample1-vertebra_inst_5																																																																		0.9684790482253002			0.9839871540400098

brainless-bot · 2026-04-10T17:18:45Z

🤖 Code Formatting Reminder

Hello there! 👋 It looks like the code in this pull request might benefit from some formatting improvements.
Fix the issues locally or use our auto format action by commenting /format on this PR!

ErikGro · 2026-04-10T17:27:02Z

/format

brainless-bot · 2026-04-10T17:27:16Z

🤖 I will now format your code with black. Check the status here.

Copilot

Pull request overview

This PR implements per-instance metric persistence/output in the aggregator TSV to support downstream distribution/meta-analysis (issues #237 and #240), while adding safeguards in statistics/plotting to avoid double-counting when instance rows are present.

Changes:

Add optional TSV output of per-instance rows alongside the existing per-sample “master” row.
Extend PanopticaResult.to_dict() to optionally return [master_dict, inst_0_dict, …] for aggregator consumption.
Add statistics helpers/flags intended to exclude instance rows from summaries/figures by default.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
`panoptica/panoptica_statistics.py`	Adds subject filtering helpers and “master-only” filtering for summaries/figures to avoid double counting with instance rows.
`panoptica/panoptica_result.py`	Switches several metric IDs to be derived from `Metric.*.result_id` and adds optional per-instance dict export from list-metrics.
`panoptica/panoptica_aggregator.py`	Adds `output_individual_instance_metrics` flag and writes additional `_inst_` rows to TSV when enabled.
`panoptica/metrics/metrics.py`	Introduces a `result_id` helper for metrics to standardize result key naming (`sq_*`).
`panoptica/instance_evaluator.py`	Minor cleanup: remove redundant `tp` local and clarify return type comment.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

brainless-bot · 2026-04-13T15:21:41Z

🤖 Code Formatting Reminder

Hello there! 👋 It looks like the code in this pull request might benefit from some formatting improvements.
Fix the issues locally or use our auto format action by commenting /format on this PR!

ErikGro · 2026-04-13T15:22:16Z

/format

brainless-bot · 2026-04-13T15:22:28Z

🤖 I will now format your code with black. Check the status here.

…/panoptica into 237-240-per-instance-results

brainless-bot · 2026-04-13T16:41:46Z

🤖 Code Formatting Reminder

Hello there! 👋 It looks like the code in this pull request might benefit from some formatting improvements.
Fix the issues locally or use our auto format action by commenting /format on this PR!

neuronflow · 2026-04-13T16:46:14Z

/format

brainless-bot · 2026-04-13T16:46:29Z

🤖 I will now format your code with black. Check the status here.

…/panoptica into 237-240-per-instance-results

ErikGro · 2026-04-13T17:14:37Z

What do you think about the export format?
The first row of each sample defines global and average (SQ) Metrics.
The following instance rows only set the values in the relevant semantic quality ("sq") columns.
The export is quite sparse and also it might be confusing that in SQ columns it is the average in the first and the individual values in the instance rows. Let me know what you think.
About the sparsity: an alternative format allowing nested values like in JSONL would be an option.

TODO: The Statistics package needs to be updated as well.

ErikGro · 2026-04-20T15:43:32Z

/format

brainless-bot · 2026-04-20T15:43:45Z

🤖 I will now format your code with black. Check the status here.

ErikGro · 2026-04-20T15:48:15Z

/format

brainless-bot · 2026-04-20T15:48:25Z

🤖 I will now format your code with black. Check the status here.

Copilot

Pull request overview

Copilot reviewed 9 out of 9 changed files in this pull request and generated 2 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

ErikGro · 2026-04-21T16:33:33Z

/format

brainless-bot · 2026-04-21T16:33:43Z

🤖 I will now format your code with black. Check the status here.

brainless-bot · 2026-04-21T16:33:44Z

🤖 Code Formatting Reminder

Hello there! 👋 It looks like the code in this pull request might benefit from some formatting improvements.
Fix the issues locally or use our auto format action by commenting /format on this PR!

Copilot

Pull request overview

Copilot reviewed 11 out of 11 changed files in this pull request and generated 2 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

brainless-bot · 2026-04-22T11:39:04Z

🤖 Code Formatting Reminder

Hello there! 👋 It looks like the code in this pull request might benefit from some formatting improvements.
Fix the issues locally or use our auto format action by commenting /format on this PR!

brainless-bot · 2026-04-22T11:43:01Z

🤖 Code Formatting Reminder

Hello there! 👋 It looks like the code in this pull request might benefit from some formatting improvements.
Fix the issues locally or use our auto format action by commenting /format on this PR!

ErikGro · 2026-04-22T11:43:09Z

/format

brainless-bot · 2026-04-22T11:43:22Z

🤖 I will now format your code with black. Check the status here.

ErikGro · 2026-04-22T11:52:35Z

+            raise NotImplementedError(
+                "PanopticaAUTCResult does not support per-instance output; "
+                "set output_individual_instance_metrics=False on the aggregator when using is_autc=True."
+            )


We omit instance wise reporting for AUTC for now to keep complexity of tsv managable:
Combining them gives a roughly thresholds × instances blow-up of cells per subject, plus the question of which cells are meaningful (unmatched refs at a given threshold) vs. empty.

ErikGro added 5 commits April 10, 2026 16:16

Optionally export all instances of a sample

f5dbbda

Replace hardcoded keys

be3bc14

WIP

41314b0

Merge branch 'main' into 237-240-per-instance-results

08cc640

Fix instance export

306327d

Autoformat with black

355730e

ErikGro marked this pull request as draft April 10, 2026 19:56

neuronflow requested a review from Copilot April 11, 2026 07:24

Copilot started reviewing on behalf of neuronflow April 11, 2026 07:24 View session

Copilot AI reviewed Apr 11, 2026

View reviewed changes

Comment thread panoptica/panoptica_statistics.py Outdated

Comment thread panoptica/panoptica_statistics.py Outdated

Comment thread panoptica/panoptica_aggregator.py Outdated

Comment thread panoptica/panoptica_result.py

aymuos15 reviewed Apr 12, 2026

View reviewed changes

Comment thread panoptica/metrics/metrics.py

Review fixes & tests added

3236c1e

brainless-bot Bot and others added 3 commits April 13, 2026 15:22

Autoformat with black

f3e0f3b

Revert sq ids for result

f6b8206

Merge branch '237-240-per-instance-results' of github.com:BrainLesion…

90f37fc

…/panoptica into 237-240-per-instance-results

Autoformat with black

3986ef6

aymuos15 reviewed Apr 13, 2026

View reviewed changes

Comment thread panoptica/panoptica_aggregator.py

ErikGro added 3 commits April 13, 2026 19:02

fix HD95 typo

e591aff

Merge branch '237-240-per-instance-results' of github.com:BrainLesion…

c9e7b7d

…/panoptica into 237-240-per-instance-results

Readd comment

b9b7848

Review fixes

9de86fd

ErikGro marked this pull request as ready for review April 20, 2026 15:43

Review fixes

4471d51

neuronflow requested a review from Copilot April 20, 2026 17:25

Copilot started reviewing on behalf of neuronflow April 20, 2026 17:25 View session

Copilot AI reviewed Apr 20, 2026

View reviewed changes

Comment thread panoptica/utils/serialization.py

Comment thread panoptica/panoptica_aggregator.py

Review fixes

6e908b8

Autoformat with black

41b9c1c

neuronflow requested a review from Copilot April 22, 2026 07:39

Copilot started reviewing on behalf of neuronflow April 22, 2026 07:40 View session

Copilot AI reviewed Apr 22, 2026

View reviewed changes

Comment thread panoptica/utils/serialization.py

Comment thread panoptica/utils/serialization.py

neuronflow approved these changes Apr 22, 2026

View reviewed changes

ErikGro added 2 commits April 22, 2026 13:20

Merge branch 'main' into 237-240-per-instance-results

1d7b5bb

Fix merge errors

c652398

Review fixes

ad9de19

Autoformat with black

011f67a

ErikGro commented Apr 22, 2026

View reviewed changes

Hendrik-code merged commit a3f6ab7 into main Apr 22, 2026
13 checks passed

Hendrik-code deleted the 237-240-per-instance-results branch April 22, 2026 12:06

Conversation

ErikGro commented Apr 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

brainless-bot Bot commented Apr 10, 2026

Uh oh!

ErikGro commented Apr 10, 2026

Uh oh!

brainless-bot Bot commented Apr 10, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

brainless-bot Bot commented Apr 13, 2026

Uh oh!

ErikGro commented Apr 13, 2026

Uh oh!

brainless-bot Bot commented Apr 13, 2026

Uh oh!

brainless-bot Bot commented Apr 13, 2026

Uh oh!

neuronflow commented Apr 13, 2026

Uh oh!

brainless-bot Bot commented Apr 13, 2026

Uh oh!

Uh oh!

ErikGro commented Apr 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ErikGro commented Apr 20, 2026

Uh oh!

brainless-bot Bot commented Apr 20, 2026

Uh oh!

ErikGro commented Apr 20, 2026

Uh oh!

brainless-bot Bot commented Apr 20, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

ErikGro commented Apr 21, 2026

Uh oh!

brainless-bot Bot commented Apr 21, 2026

Uh oh!

brainless-bot Bot commented Apr 21, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

brainless-bot Bot commented Apr 22, 2026

Uh oh!

brainless-bot Bot commented Apr 22, 2026

Uh oh!

ErikGro commented Apr 22, 2026

Uh oh!

brainless-bot Bot commented Apr 22, 2026

Uh oh!

ErikGro Apr 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

ErikGro commented Apr 10, 2026 •

edited

Loading

ErikGro commented Apr 13, 2026 •

edited

Loading

ErikGro Apr 22, 2026 •

edited

Loading