Apply suggestions from code review

Co-authored-by: bthirion <bertrand.thirion@inria.fr>

doc/glm/glm_intro.rst

@@ -33,7 +33,7 @@ One way to analyze times series consists in comparing them to a *model* built fr
 
 One expects that a brain region involved in the processing of a certain type of event (e.g. the auditory cortex for sounds) would show a time course of activation that correlates with the time-diagram of these events. If the :term:`fMRI` signal directly showed neural activity and did not contain any noise, we could just look at it in various voxels and detect those that conform to the time-diagrams.
 
-But we know from previous measurements that the :term:`BOLD` signal does not follow the exact time course of stimulus processing and the underlying neural activity. The BOLD response reflects changes in blood flow and concentrations in oxy-deoxy haemoglobin, all together forming a `haemodynamic response`_ which is sluggish and long-lasting, as can be seen in the following figure showing the response to an impulsive event (for example, an auditory click played to the participants).
+But we know from previous measurements that the :term:`BOLD` signal does not follow the exact time course of stimulus processing and the underlying neural activity. The :term:`BOLD` response reflects changes in blood flow and concentrations in oxy-deoxy haemoglobin, all together forming a `haemodynamic response`_ which is sluggish and long-lasting, as can be seen in the following figure showing the response to an impulsive event (for example, an auditory click played to the participants).
 
 .. figure:: ../images/spm_iHRF.png
 

doc/glm/second_level_model.rst

@@ -7,7 +7,7 @@ Second level models
 .. topic:: **Page summary**
 
    Second level models in Nilearn are used to perform group-level analyses on :term:`fMRI` data. Once individual
-   subjects have been processed in a common space (e.g. MNI, Talairach, or subject average), the data can
+   subjects have been processed in a common space (e.g. :term:`MNI`, Talairach, or subject average), the data can
    be grouped and statistical tests  performed to make broader inferences on :term:`fMRI` activity. Some common
    second level models are one-sample (unpaired or paired) and two-sample t-tests.
 

examples/04_glm_first_level/plot_localizer_surface_analysis.py

@@ -24,7 +24,7 @@
   was obtained.
 
 * Another possibility, used here, is to project
-  the normalized :term:`fMRI` data to an MNI-coregistered mesh,
+  the normalized :term:`fMRI` data to an :term:`MNI`-coregistered mesh,
   such as fsaverage.
 
 The advantage of this second approach is that it makes it easy to run

nilearn/interfaces/bids/glm.py

@@ -17,7 +17,7 @@ def save_glm_to_bids(
     out_dir=".",
     prefix=None,
 ):
-    """Save :term:`GLM` results to BIDS-like files.
+    """Save :term:`GLM` results to :term:`BIDS`-like files.
 
     .. versionadded:: 0.9.2