Small fixes for detecting the correct files in folders with multiple timestamps. [skip ci]

Author: dbouget

raidionicsval/Studies/AbstractStudy.py

@@ -433,6 +433,7 @@ def compute_and_plot_categorical_metric_over_metric_categories(self, class_name:
                                                           category=category,
                                                           study_name=study_name,
                                                           suffix=suffix + '_' + metric2 + '_' + cat)
+            suffix = "tp" + suffix if category == "True Positive" else suffix
             export_segmentation_df_to_latex_paper(folder=self.output_folder, class_name=class_name, study=study_name,
                                                   categories=list(optimal_results_per_cutoff.keys()), suffix=suffix)
         except Exception as e:

raidionicsval/Studies/SegmentationStudy.py

@@ -31,40 +31,40 @@ def run(self):
                                                                 metric2_cutoffs=[2.], category='All')
         """
         for c in self.class_names:
-            # super().compute_and_plot_overall(c, category='All')
-            # super().compute_and_plot_overall(c, category='True Positive')
-            #
-            # # Plotting the results based on the selection of dense parameters
-            # for s in SharedResources.getInstance().studies_selections_dense:
-            #     parsing = s.split(',')
-            #     metric1 = parsing[0]
-            #     metric2 = parsing[1]
-            #     if parsing[2] != '':
-            #         metric2_cutoff = [float(x) for x in parsing[2].split('-')]
-            #     else:
-            #         metric2_cutoff = None
-            #     category = parsing[3]
-            #     self.compute_and_plot_metric_over_metric_categories(class_name=c, metric1=metric1, metric2=metric2,
-            #                                                         metric2_cutoffs=metric2_cutoff, category=category)
-            #
-            # # Plotting the results based on the selection of categorical parameters
-            # for s in SharedResources.getInstance().studies_selections_categorical:
-            #     parsing = s.split(',')
-            #     metric1 = parsing[0].strip()
-            #     metric2 = parsing[1].strip()
-            #     if parsing[2].strip() != '':
-            #         metric2_cutoff = [x for x in parsing[2].split('-')]
-            #     else:
-            #         metric2_cutoff = None
-            #     category = parsing[3].strip()
-            #     self.compute_and_plot_categorical_metric_over_metric_categories(class_name=c, metric1=metric1,
-            #                                                                     metric2=metric2,
-            #                                                                     metric2_cutoffs=metric2_cutoff,
-            #                                                                     category=category)
-            #
-            # # Correlation matrix between all metrics
-            # super().compute_and_plot_metrics_correlation_matrix(class_name=c, category='All')
-            # super().compute_and_plot_metrics_correlation_matrix(class_name=c, category='True Positive')
+            super().compute_and_plot_overall(c, category='All')
+            super().compute_and_plot_overall(c, category='True Positive')
+
+            # Plotting the results based on the selection of dense parameters
+            for s in SharedResources.getInstance().studies_selections_dense:
+                parsing = s.split(',')
+                metric1 = parsing[0]
+                metric2 = parsing[1]
+                if parsing[2] != '':
+                    metric2_cutoff = [float(x) for x in parsing[2].split('-')]
+                else:
+                    metric2_cutoff = None
+                category = parsing[3]
+                self.compute_and_plot_metric_over_metric_categories(class_name=c, metric1=metric1, metric2=metric2,
+                                                                    metric2_cutoffs=metric2_cutoff, category=category)
+
+            # Plotting the results based on the selection of categorical parameters
+            for s in SharedResources.getInstance().studies_selections_categorical:
+                parsing = s.split(',')
+                metric1 = parsing[0].strip()
+                metric2 = parsing[1].strip()
+                if parsing[2].strip() != '':
+                    metric2_cutoff = [x for x in parsing[2].split('-')]
+                else:
+                    metric2_cutoff = None
+                category = parsing[3].strip()
+                self.compute_and_plot_categorical_metric_over_metric_categories(class_name=c, metric1=metric1,
+                                                                                metric2=metric2,
+                                                                                metric2_cutoffs=metric2_cutoff,
+                                                                                category=category)
+
+            # Correlation matrix between all metrics
+            super().compute_and_plot_metrics_correlation_matrix(class_name=c, category='All')
+            super().compute_and_plot_metrics_correlation_matrix(class_name=c, category='True Positive')
 
             # Cascading results based on a combination of the selected dense/categorical parameters
             self.compute_and_plot_metric_over_metric_cascading_categories(class_name=c, category='All')

raidionicsval/Validation/kfold_model_validation.py

@@ -116,6 +116,7 @@ def __compute_metrics(self):
             results_per_folds.append(results)
 
     def __compute_metrics_for_fold(self, data_list, fold_number):
+        data_list = ["5_1000_MR_FLAIR_post_4125", "5_1000_MR_FLAIR_pre_2366"]
         for i, patient in enumerate(tqdm(data_list)):
             uid = None
             try:
@@ -195,7 +196,7 @@ def __identify_patient_files(self, patient_metrics: PatientMetrics, folder_index
             for _, _, files in os.walk(detection_image_base):
                 for f in files:
                     if pred_suffix in f:
-                        if use_internal_convention and patient_metrics.patient_id.split('_')[1] in f:
+                        if use_internal_convention and patient_metrics.patient_id.split('_')[1] in f.split('_'):
                             detection_filename = os.path.join(detection_image_base, f)
                         elif not use_internal_convention:
                             detection_filename = os.path.join(detection_image_base, f)