Merge pull request #28 from grandjeanlab/main

make orthogonal qc plots with nilearn

Author: maswendt

README.md

@@ -25,18 +25,20 @@ Main function: *ParsingData*
 See the full manual [here](https://github.com/Aswendt-Lab/AIDAqc/blob/main/docs/AIDAqc_v2_1.pdf).
 
 <h3>Docker/Apptainer Usage</h3>
-Build:
 
-```docker build aidaqc:2.1 .```
+```{bash}
+#Build
 
-Running the main ParsingData.py:
+docker build aidaqc:2.1 .
 
-```docker run -v /your/project/data:/data -v /your/project/qc aidaqc:2.1 -i /data -o /qc -f raw```
-=======
-For installation in a [apptainer](https://apptainer.org/) container for GNU/Linux:
+# Running the main ParsingData.py:
 
-```{bash}
+docker run -v /your/project/data:/data -v /your/project/qc aidaqc:2.1 -i /data -o /qc -f raw
 
+```
+
+For installation in a [apptainer](https://apptainer.org/) container for GNU/Linux:
+```{bash}
 # Download the repository
 git clone https://github.com/Aswendt-Lab/AIDAqc.git
 cd AIDAqc
@@ -45,7 +47,7 @@ cd AIDAqc
 apptainer build aidaqc.sif apptainer.def
 
 # Get into a bash shell in the container
-apptainer run aidaqc.sif
+apptainer shell aidaqc.sif
 
 ```
 
@@ -64,7 +66,7 @@ This tool has been validated and used in the following publication: [Publication
 A total of 23 datasets from various institutes were used for validation and testing. These datasets can be found via: [Datasets Link](https://gin.g-node.org/Aswendt_Lab/2023_Kalantari_AIDAqc)
 
 <h3>Download test dataset</h3>
-https://gin.g-node.org/Aswendt_Lab/testdata_aidaqc
+[Dataset Link](https://gin.g-node.org/Aswendt_Lab/testdata_aida)
 
 [<h3><b>CONTACT</h3></b>](https://neurologie.uk-koeln.de/forschung/ag-neuroimaging-neuroengineering/)
 Aref Kalantari (aref.kalantari-sarcheshmehATuk-koeln.de) and Markus Aswendt (markus.aswendtATuk-koeln.de)

apptainer.def

@@ -28,7 +28,6 @@ From: ubuntu:latest
 
         apt-get update -yy
         apt-get install -yy --no-install-recommends git python3 python3-pip fonts-unifont
-        apt-get install -yy --reinstall ttf-mscorefonts-installer
 
         git clone https://github.com/Aswendt-Lab/AIDAqc.git
         cd AIDAqc
@@ -44,4 +43,4 @@ From: ubuntu:latest
 
 %runscript
 
-        bash
+        ParsingData.py

scripts/FeatureCheck.py

@@ -4,8 +4,8 @@
 import glob
 import matplotlib.backends.backend_pdf
 import nibabel as nib
-import nibabel as nii
-import matplotlib.pyplot as plt
+from nilearn.plotting import plot_img
+from nilearn.image import index_img
 import pv_conv2Nifti as pr
 import alive_progress as ap
 import pv_parser as par
@@ -93,7 +93,14 @@ def CheckingRawFeatures(Path):
                     if os.path.isfile(CP_v) and os.path.isfile(CP_a):
                         try:
                             pv.read_2dseq(map_raw=False, pv6=False)
-                            input_file = nii.squeeze_image(pv.nim)
+                            input_file = nib.squeeze_image(pv.nim)
+                            # this is a workaround to convert the data for nilearn plotting
+                            affine = np.eye(4)
+                            affine[0, 0] = pv.nim.header.get('pixdim')[1]
+                            affine[1, 1] = pv.nim.header.get('pixdim')[2]
+                            affine[2, 2] = pv.nim.header.get('pixdim')[3]
+                            affine[3, 3] = pv.nim.header.get('pixdim')[4]
+                            input_file=nib.Nifti1Image(input_file.get_fdata(), affine=affine,dtype=int(32))
                         except ValueError:
                             ErorrList.append(tf+"_Value Error")
                             print(tf)
@@ -135,21 +142,21 @@ def CheckingRawFeatures(Path):
 #                         continue
 # =============================================================================
                     ########### Slice extraction 
-                    selected_img = Image_Selection(input_file)                    
                     qc_path = os.path.join(saving_path,"manual_slice_inspection")
                     if not os.path.isdir(qc_path):
                         os.mkdir(qc_path)
                     img_name = str.split(tf,os.sep)[-2]
                     full_img_name = str(N)+"_" + img_name+"_"+ str(dd)+".png".replace(".nii","").replace(".gz","")
                     img_names_new.append(full_img_name) 
 
-                    #plt.figure()          
-                    plt.axis('off')
-                    plt.imshow(selected_img,cmap='gray')
-                    #svg_path = os.path.join(qc_path,+ img_name+".png").replace(".nii","").replace(".gz","")
+                    # grandjean patch to output ortho representation of the image in manual slice inspection
                     svg_path = os.path.join(qc_path,str(N)+"_"+ img_name+"_"+ str(dd)+".png").replace(".nii","").replace(".gz","")
                     dd = dd +1
-                    plt.savefig(svg_path)
+                    if len(input_file.shape) == 4:
+                        input_file_img = index_img(input_file,0)
+                    else:
+                        input_file_img = input_file
+                    plot_img(input_file_img, title=full_img_name, output_file=svg_path)
                     ########### Slice extraction               
                     # other Features
                     SpatRes = ResCalculator(input_file)
@@ -210,7 +217,7 @@ def CheckingRawFeatures(Path):
             df["corresponding_img"] = img_names_new
             df['SpatRx'] = np.array(SpatRes_vec)[:,0]
             df['SpatRy'] = np.array(SpatRes_vec)[:,1]
-            df['Slicethick'] = np.array(SpatRes_vec)[:,2]
+            df['SpatRz'] = np.array(SpatRes_vec)[:,2]
             df['Ghosting'] = np.array(GMetric_vec)
 
 
@@ -340,7 +347,6 @@ def CheckingNiftiFeatures(Path):
                         continue
 
                     ########### Slice extraction 
-                    selected_img = Image_Selection(input_file)                    
                     qc_path = os.path.join(Path,"manual_slice_inspection")
                     if not os.path.isdir(qc_path):
                         os.mkdir(qc_path)
@@ -349,12 +355,14 @@ def CheckingNiftiFeatures(Path):
                     full_img_name = (str(N)+"_"+folder_name+"_"+img_name+"_"+str(dd)+".png").replace(".nii","").replace(".gz","")
                     img_names_new.append(full_img_name)
 
-                    #plt.figure()          
-                    plt.axis('off')
-                    plt.imshow(selected_img,cmap='gray')
-                    svg_path = os.path.join(qc_path,str(N)+"_"+folder_name+"_"+img_name+"_"+str(dd)+".png").replace(".nii","").replace(".gz","")
+                    # grandjean patch to output ortho representation of the image in manual slice inspection
+                    svg_path = os.path.join(qc_path,str(N)+"_"+ img_name+"_"+ str(dd)+".png").replace(".nii","").replace(".gz","")
                     dd = dd +1
-                    plt.savefig(svg_path)
+                    if len(input_file.shape) == 4:
+                        input_file_img = index_img(input_file,0)
+                    else:
+                        input_file_img = input_file
+                    plot_img(input_file_img, title=full_img_name, output_file=svg_path)
                     ########### Slice extraction               
                     # other Features
                     SpatRes = ResCalculator(input_file)
@@ -419,7 +427,7 @@ def CheckingNiftiFeatures(Path):
             df["corresponding_img"] = img_names_new
             df['SpatRx'] = np.array(SpatRes_vec)[:,0]
             df['SpatRy'] = np.array(SpatRes_vec)[:,1]
-            df['Slicethick'] = np.array(SpatRes_vec)[:,2]
+            df['SpatRz'] = np.array(SpatRes_vec)[:,2]
             df['Ghosting'] = np.array(GMetric_vec)
 
 

scripts/QC.py

@@ -96,25 +96,6 @@ def GhostCheck(input_file):
 
 #%% Res function
 
-def Image_Selection(input_file):
-    
-    img = input_file
-    img_data=img.get_fdata()
-    img_shape=img_data.shape 
-    middle=int(img_shape[2]/2)
-    if len(img_shape) == 3:
-            selected_img= img_data[:, :,middle]
-            
-    elif len(img_shape) > 3:
-        time_middle=int(img_shape[3]/2)
-        selected_img= img_data[:, :,middle,time_middle]
-        
-    selected_img= np.rot90(selected_img,k=-1)
-    return selected_img
-
-
-#%% 
-
 
 def ResCalculator(input_file):
 
@@ -553,7 +534,7 @@ def QCPlot(Path):
         D = Abook[nn]
         for cc, C in enumerate(COL):
             Data = list(D[C])
-            if C == 'SpatRx' or C == 'SpatRy' or C == 'Slicethick':
+            if C == 'SpatRx' or C == 'SpatRy' or C == 'SpatRz':
                 # Plot pie plots
                 labels = list(set(Data))
                 sizes = [Data.count(l) for l in labels]