Add dataset data_normalize_metric_pam50

[mguaypaq]

Fixes #4044.

For testing, I successfully downloaded the data with:

sct_download_data -d data_normalize_metric_pam50

Question for the reviewer(s): is data_normalize_metric_pam50 the name we want to use for this dataset?

[sandrinebedard]

Will it be downloaded with the SCT installation?

Question for the reviewer(s): is data_normalize_metric_pam50 the name we want to use for this dataset?

Yes it is fine!

[mguaypaq]

Will it be downloaded with the SCT installation?

Not by default. Should it be? If so, we just need to add it to install_sct here:

spinalcordtoolbox/install_sct

Lines 684 to 689 in 158abd1

	# Download data
	print info "Installing data..."
	run mkdir -p "$SCT_DIR/$DATA_DIR"
	for data in PAM50 optic_models pmj_models deepseg_sc_models deepseg_gm_models deepseg_lesion_models c2c3_disc_models deepreg_models; do
	run sct_download_data -d "$data"
	done

and to install_sct.bat here:

spinalcordtoolbox/install_sct.bat

Lines 97 to 100 in 158abd1

	rem Install external dependencies
	echo:
	echo ### Downloading model files and binaries...
	FOR %%D IN (PAM50 optic_models pmj_models deepseg_sc_models deepseg_gm_models deepseg_lesion_models c2c3_disc_models binaries_win deepreg_models) DO sct_download_data -d %%D -k || goto error

[joshuacwnewton]

Question for the reviewer(s): is data_normalize_metric_pam50 the name we want to use for this dataset?

Some nitpicks:

• Elsewhere in sct_download_data, we use PAM50 (capitalized), so it might be good to unify that.
• Part of me wants to use a name that represents a noun, similar to the other dataset names? For example, rather than "data normalize metric pam50", we would say "PAM50-normalized metrics", which could then become the dataset name PAM50_normalized_metrics .

[sandrinebedard]

I am fine with renaming the repo and folder!

[joshuacwnewton]

Side-question: Would it be beneficial to combine this PR with #4003?

The reason I suggest this is because of the following scenario:

• We merge this PR to master.
• #4003 is updated with master to include the new dataset link.
• We start playing around with #4003 to compute compression metrics from the CSV dataset.
• During testing, we identify some flaw or concern with the dataset (perhaps some issue with how the metrics were generated via #3977).
• We need to regenerate the dataset and create a new release link, and thus a new change to master.

By comparison, if this dataset link were to be included as part of #4003, then it could potentially save a bit of clutter on master, and keep related development grouped in the same location, since any changes to the dataset would be part of #4003's development.

(But, maybe the scenario I describe is unlikely, or easy enough to handle if something goes wrong?)

[sandrinebedard]

yes, we can include this in #4003, I agree!

[sandrinebedard]

Will it be downloaded with the SCT installation?

Not by default. Should it be? If so, we just need to add it to install_sct here:

spinalcordtoolbox/install_sct

Lines 684 to 689 in 158abd1

	# Download data
	print info "Installing data..."
	run mkdir -p "$SCT_DIR/$DATA_DIR"
	for data in PAM50 optic_models pmj_models deepseg_sc_models deepseg_gm_models deepseg_lesion_models c2c3_disc_models deepreg_models; do
	run sct_download_data -d "$data"
	done

and to install_sct.bat here:

spinalcordtoolbox/install_sct.bat

Lines 97 to 100 in 158abd1

	rem Install external dependencies
	echo:
	echo ### Downloading model files and binaries...
	FOR %%D IN (PAM50 optic_models pmj_models deepseg_sc_models deepseg_gm_models deepseg_lesion_models c2c3_disc_models binaries_win deepreg_models) DO sct_download_data -d %%D -k || goto error

OK great! We discussed this in the sct_dev meeting and with the project team for the compression metrics, we want it within the sct installation!

[sandrinebedard]

Closing since I merged this branch into #4003. --> 124cf66