Add adaptive mask plot to report

docs/outputs.rst

@@ -528,11 +528,30 @@ should not overly focus on carpet plots and should examine these results in cont
   :height: 400px
 
 
+**************************
+Adaptive Mask Summary Plot
+**************************
+
+Below the carpet plots is a summary plot of the adaptive mask.
+
+This figure overlays contours reflecting the boundaries of the following masks onto the mean optimally combined data:
+
+- **Base**: The base mask, either provided by the user or generated automatically using ``compute_epi_mask``.
+- **Optimal combination**: The mask used for optimal combination and denoising.
+  This corresponds to values greater than or equal to 1 (at least 1 good echo) in the adaptive mask.
+- **Classification**: The mask used for the decomposition and component classification steps.
+  This corresponds to values greather than or equal to 3 (at least 3 good echoes) in the adaptive mask.
+
+.. image:: /_static/adaptive_mask.png
+  :align: center
+  :height: 400px
+
+
 ************************
 T2* and S0 Summary Plots
 ************************
 
-Below the carpet plots are summary plots for the T2* and S0 maps.
+Below the adaptive mask plot are summary plots for the T2* and S0 maps.
 Each map has two figures: a spatial map of the values and a histogram of the voxelwise values.
 The T2* map should look similar to T2 maps and be brightest in the ventricles and darkest in areas of largest susceptibility.
 The S0 map should roughly follow the signal-to-noise ratio and will be brightest near the surface near RF coils.

tedana/reporting/data/html/report_body_template.html

@@ -180,21 +180,25 @@ <h1>Carpet plots</h1>
     </div>
   </div>
 </div>
+<div class="carpet-plots">
+  <h1>Adaptive mask</h1>
+  <img id="adaptiveMask" src="$adaptiveMask" style="height: 500px" />
+</div>
 <div class="carpet-plots">
   <h1>T2* and S0</h1>
   <h2>T2*</h2>
   <div class="carpet-plots-image">
-    <img id="t2starBrainPlot" src="$t2starBrainPlot" class="brainplot" />
+    <img id="t2starBrainPlot" src="$t2starBrainPlot" style="height: 500px" />
   </div>
   <div class="carpet-plots-image">
-    <img id="t2starHistogram" src="$t2starHistogram" />
+    <img id="t2starHistogram" src="$t2starHistogram" style="height: 500px" />
   </div>
   <h2>S0</h2>
   <div class="carpet-plots-image">
-    <img id="s0BrainPlot" src="$s0BrainPlot" class="brainplot" />
+    <img id="s0BrainPlot" src="$s0BrainPlot" style="height: 500px" />
   </div>
   <div class="carpet-plots-image">
-    <img id="s0Histogram" src="$s0Histogram" />
+    <img id="s0Histogram" src="$s0Histogram" style="height: 500px" />
   </div>
 </div>
 </div>

tedana/reporting/html_report.py

@@ -143,6 +143,9 @@ def _update_template_bokeh(bokeh_id, info_table, about, prefix, references, boke
     # Initial carpet plot (default one)
     initial_carpet = f"./figures/{prefix}carpet_optcom.svg"
 
+    # Adaptive mask image
+    adaptive_mask = f"./figures/{prefix}adaptive_mask.svg"
+
     # T2* and S0 images
     t2star_brain = f"./figures/{prefix}t2star_brain.svg"
     t2star_histogram = f"./figures/{prefix}t2star_histogram.svg"
@@ -165,6 +168,7 @@ def _update_template_bokeh(bokeh_id, info_table, about, prefix, references, boke
         about=about,
         prefix=prefix,
         initialCarpet=initial_carpet,
+        adaptiveMask=adaptive_mask,
         t2starBrainPlot=t2star_brain,
         t2starHistogram=t2star_histogram,
         s0BrainPlot=s0_brain,

tedana/reporting/static_figures.py

@@ -616,3 +616,82 @@ def plot_t2star_and_s0(
         annotate=False,
         output_file=os.path.join(io_generator.out_dir, "figures", s0_plot),
     )
+
+
+def plot_adaptive_mask(
+    *,
+    optcom: np.ndarray,
+    base_mask: np.ndarray,
+    io_generator: io.OutputGenerator,
+):
+    """Create a figure to show the adaptive mask.
+
+    This figure shows the base mask, the adaptive mask thresholded for denoising (threshold >= 1),
+    and the adaptive mask thresholded for classification (threshold >= 3),
+    overlaid on the mean optimal combination image.
+
+    Parameters
+    ----------
+    optcom : (S x T) :obj:`numpy.ndarray`
+        Optimal combination of components.
+        The mean image over time is used as the underlay for the figure.
+    base_mask : (S,) :obj:`numpy.ndarray`
+        Base mask used in tedana.
+        This is the original mask either provided by the user or generated with `compute_epi_mask`.
+    io_generator : :obj:`~tedana.io.OutputGenerator`
+        The output generator for this workflow.
+    """
+    from matplotlib.lines import Line2D
+    from nilearn import image
+
+    adaptive_mask_img = io_generator.get_name("adaptive mask img")
+    mean_optcom_img = io.new_nii_like(io_generator.reference_img, np.mean(optcom, axis=1))
+
+    # Concatenate the three masks used in tedana to treat as a probabilistic atlas
+    base_mask = io.new_nii_like(io_generator.reference_img, base_mask)
+    mask_denoise = image.math_img("(img >= 1).astype(np.uint8)", img=adaptive_mask_img)
+    mask_clf = image.math_img("(img >= 3).astype(np.uint8)", img=adaptive_mask_img)
+    all_masks = image.concat_imgs((base_mask, mask_denoise, mask_clf))
+    # Set values to 0.5 for probabilistic atlas plotting
+    all_masks = image.math_img("img * 0.5", img=all_masks)
+
+    cmap = plt.cm.gist_rainbow
+    discrete_cmap = cmap.resampled(3)  # colors matching the mask lines in the image
+    color_dict = {
+        "Base": discrete_cmap(0),
+        "Optimal combination": discrete_cmap(0.4),
+        "Classification": discrete_cmap(0.9),
+    }
+
+    ob = plotting.plot_prob_atlas(
+        maps_img=all_masks,
+        bg_img=mean_optcom_img,
+        view_type="contours",
+        threshold=0.2,
+        annotate=False,
+        draw_cross=False,
+        cmap=cmap,
+        display_mode="mosaic",
+        cut_coords=4,
+    )
+
+    legend_elements = []
+    for k, v in color_dict.items():
+        line = Line2D([0], [0], color=v, label=k, markersize=10)
+        legend_elements.append(line)
+
+    fig = ob.frame_axes.get_figure()
+    width = fig.get_size_inches()[0]
+
+    ob.frame_axes.set_zorder(100)
+    ob.frame_axes.legend(
+        handles=legend_elements,
+        facecolor="white",
+        ncols=3,
+        loc="lower center",
+        fancybox=True,
+        shadow=True,
+        fontsize=width,
+    )
+    adaptive_mask_plot = f"{io_generator.prefix}adaptive_mask.svg"
+    fig.savefig(os.path.join(io_generator.out_dir, "figures", adaptive_mask_plot))

tedana/tests/data/cornell_three_echo_outputs.txt

@@ -23,6 +23,7 @@ desc-adaptiveGoodSignal_mask.nii.gz
 desc-denoised_bold.nii.gz
 desc-optcom_bold.nii.gz
 figures
+figures/adaptive_mask.svg
 figures/carpet_optcom.svg
 figures/carpet_denoised.svg
 figures/carpet_accepted.svg

tedana/tests/data/fiu_four_echo_outputs.txt

@@ -61,6 +61,7 @@ sub-01_echo-4_desc-Rejected_bold.nii.gz
 sub-01_report.txt
 sub-01_tedana_report.html
 figures
+figures/sub-01_adaptive_mask.svg
 figures/sub-01_carpet_optcom.svg
 figures/sub-01_carpet_denoised.svg
 figures/sub-01_carpet_accepted.svg

tedana/tests/data/nih_five_echo_outputs_verbose.txt

@@ -82,6 +82,7 @@ sub-01_references.bib
 sub-01_report.txt
 sub-01_tedana_report.html
 figures
+figures/sub-01_adaptive_mask.svg
 figures/sub-01_carpet_optcom.svg
 figures/sub-01_carpet_denoised.svg
 figures/sub-01_carpet_accepted.svg

tedana/workflows/tedana.py

@@ -595,12 +595,13 @@
         t2s_limited_sec = utils.reshape_niimg(t2smap)
         t2s_limited = utils.sec2millisec(t2s_limited_sec)
         t2s_full = t2s_limited.copy()
-        mask = utils.reshape_niimg(mask)
+        mask = utils.reshape_niimg(mask).astype(int)
         mask[t2s_limited == 0] = 0  # reduce mask based on T2* map
     else:
         LGR.info("Computing EPI mask from first echo")
         first_echo_img = io.new_nii_like(io_generator.reference_img, catd[:, 0, :])
-        mask = compute_epi_mask(first_echo_img)
+        mask = compute_epi_mask(first_echo_img).get_fdata()
+        mask = utils.reshape_niimg(mask).astype(int)
         RepLGR.info(
             "An initial mask was generated from the first echo using "
             "nilearn's compute_epi_mask function."
@@ -899,6 +900,11 @@
 
         dn_ts, hikts, lowkts = io.denoise_ts(data_oc, mmix, mask_denoise, comptable)
 
+        reporting.static_figures.plot_adaptive_mask(
+            optcom=data_oc,
+            base_mask=mask,
+            io_generator=io_generator,
+        )
         reporting.static_figures.carpet_plot(
             optcom_ts=data_oc,
             denoised_ts=dn_ts,