flattened radiological gradings

SPIDER.py

@@ -37,6 +37,7 @@ def import_csv_data(filepath: str) -> List[Dict[str, str]]:
 
 # Define constants
 N_PATIENTS = 257
+MIN_IVD = 0
 MAX_IVD = 9
 
 # TODO: Add BibTeX citation
@@ -181,96 +182,15 @@ class SPIDER(datasets.GeneratorBasedBuilder):
                 "WindowWidth": datasets.Value(dtype="string"),
             },
             "rad_gradings": {
-                "IVD1": {
-                    "Modic": datasets.Value(dtype="string"),
-                    "UP endplate": datasets.Value(dtype="string"),
-                    "LOW endplate": datasets.Value(dtype="string"),
-                    "Spondylolisthesis": datasets.Value(dtype="string"),
-                    "Disc herniation": datasets.Value(dtype="string"),
-                    "Disc narrowing": datasets.Value(dtype="string"),
-                    "Disc bulging": datasets.Value(dtype="string"),
-                    "Pfirrman grade": datasets.Value(dtype="string"),
-                },
-                "IVD2": {
-                    "Modic": datasets.Value(dtype="string"),
-                    "UP endplate": datasets.Value(dtype="string"),
-                    "LOW endplate": datasets.Value(dtype="string"),
-                    "Spondylolisthesis": datasets.Value(dtype="string"),
-                    "Disc herniation": datasets.Value(dtype="string"),
-                    "Disc narrowing": datasets.Value(dtype="string"),
-                    "Disc bulging": datasets.Value(dtype="string"),
-                    "Pfirrman grade": datasets.Value(dtype="string"),
-                },
-                "IVD3": {
-                    "Modic": datasets.Value(dtype="string"),
-                    "UP endplate": datasets.Value(dtype="string"),
-                    "LOW endplate": datasets.Value(dtype="string"),
-                    "Spondylolisthesis": datasets.Value(dtype="string"),
-                    "Disc herniation": datasets.Value(dtype="string"),
-                    "Disc narrowing": datasets.Value(dtype="string"),
-                    "Disc bulging": datasets.Value(dtype="string"),
-                    "Pfirrman grade": datasets.Value(dtype="string"),
-                },
-                "IVD4": {
-                    "Modic": datasets.Value(dtype="string"),
-                    "UP endplate": datasets.Value(dtype="string"),
-                    "LOW endplate": datasets.Value(dtype="string"),
-                    "Spondylolisthesis": datasets.Value(dtype="string"),
-                    "Disc herniation": datasets.Value(dtype="string"),
-                    "Disc narrowing": datasets.Value(dtype="string"),
-                    "Disc bulging": datasets.Value(dtype="string"),
-                    "Pfirrman grade": datasets.Value(dtype="string"),
-                },
-                "IVD5": {
-                    "Modic": datasets.Value(dtype="string"),
-                    "UP endplate": datasets.Value(dtype="string"),
-                    "LOW endplate": datasets.Value(dtype="string"),
-                    "Spondylolisthesis": datasets.Value(dtype="string"),
-                    "Disc herniation": datasets.Value(dtype="string"),
-                    "Disc narrowing": datasets.Value(dtype="string"),
-                    "Disc bulging": datasets.Value(dtype="string"),
-                    "Pfirrman grade": datasets.Value(dtype="string"),
-                },
-                "IVD6": {
-                    "Modic": datasets.Value(dtype="string"),
-                    "UP endplate": datasets.Value(dtype="string"),
-                    "LOW endplate": datasets.Value(dtype="string"),
-                    "Spondylolisthesis": datasets.Value(dtype="string"),
-                    "Disc herniation": datasets.Value(dtype="string"),
-                    "Disc narrowing": datasets.Value(dtype="string"),
-                    "Disc bulging": datasets.Value(dtype="string"),
-                    "Pfirrman grade": datasets.Value(dtype="string"),
-                },
-                "IVD7": {
-                    "Modic": datasets.Value(dtype="string"),
-                    "UP endplate": datasets.Value(dtype="string"),
-                    "LOW endplate": datasets.Value(dtype="string"),
-                    "Spondylolisthesis": datasets.Value(dtype="string"),
-                    "Disc herniation": datasets.Value(dtype="string"),
-                    "Disc narrowing": datasets.Value(dtype="string"),
-                    "Disc bulging": datasets.Value(dtype="string"),
-                    "Pfirrman grade": datasets.Value(dtype="string"),
-                },
-                "IVD8": {
-                    "Modic": datasets.Value(dtype="string"),
-                    "UP endplate": datasets.Value(dtype="string"),
-                    "LOW endplate": datasets.Value(dtype="string"),
-                    "Spondylolisthesis": datasets.Value(dtype="string"),
-                    "Disc herniation": datasets.Value(dtype="string"),
-                    "Disc narrowing": datasets.Value(dtype="string"),
-                    "Disc bulging": datasets.Value(dtype="string"),
-                    "Pfirrman grade": datasets.Value(dtype="string"),
-                },
-                "IVD9": {
-                    "Modic": datasets.Value(dtype="string"),
-                    "UP endplate": datasets.Value(dtype="string"),
-                    "LOW endplate": datasets.Value(dtype="string"),
-                    "Spondylolisthesis": datasets.Value(dtype="string"),
-                    "Disc herniation": datasets.Value(dtype="string"),
-                    "Disc narrowing": datasets.Value(dtype="string"),
-                    "Disc bulging": datasets.Value(dtype="string"),
-                    "Pfirrman grade": datasets.Value(dtype="string"),
-                },
+                "IVD label": datasets.Sequence(datasets.Value("string")),
+                "Modic": datasets.Sequence(datasets.Value("string")),
+                "UP endplate": datasets.Sequence(datasets.Value("string")),
+                "LOW endplate": datasets.Sequence(datasets.Value("string")),
+                "Spondylolisthesis": datasets.Sequence(datasets.Value("string")),
+                "Disc herniation": datasets.Sequence(datasets.Value("string")),
+                "Disc narrowing": datasets.Sequence(datasets.Value("string")),
+                "Disc bulging": datasets.Sequence(datasets.Value("string")),
+                "Pfirrman grade": datasets.Sequence(datasets.Value("string")),
             }
         })
 
@@ -444,10 +364,10 @@ class SPIDER(datasets.GeneratorBasedBuilder):
             patient_grades = [
                 x for x in grades_data if x['Patient'] == str(patient_id)
             ]
-            # Pad radiological gradings so that data for all patients have
-            # the same dimensions
-            if len(patient_grades) < MAX_IVD:
-                for i in range(len(patient_grades) + 1, MAX_IVD + 1):
+            # Pad so that all patients have same number of IVD observations
+            IVD_values = [x['IVD label'] for x in patient_grades]
+            for i in range(MIN_IVD, MAX_IVD + 1):
+                if str(i) not in IVD_values:
                     patient_grades.append({
                         "Patient": f"{patient_id}",
                         "IVD label": f"{i}",
@@ -460,9 +380,19 @@ class SPIDER(datasets.GeneratorBasedBuilder):
                         "Disc bulging": "",
                         "Pfirrman grade": "",
                     })
-            assert len(patient_grades) == MAX_IVD, "Rad. gradings not padded correctly"
-            grades_dict[str(patient_id)] = patient_grades
-        assert all([len(x) == MAX_IVD for x in grades_dict.values()])
+            assert len(patient_grades) == (MAX_IVD - MIN_IVD + 1), "Radiological\
+                gradings not padded correctly"
+        
+            # Convert to sequences
+            df = (
+                pd.DataFrame(patient_grades)
+                .sort_values("IVD label")
+                .reset_index(drop=True)
+            )
+            grades_dict[str(patient_id)] = {
+                col:df[col].tolist() for col in df.columns
+                if col not in ['Patient']
+            }
 
         # Import image and mask data
         image_files = [
@@ -534,14 +464,7 @@ class SPIDER(datasets.GeneratorBasedBuilder):
             image_overview = overview_dict[scan_id]
 
             # Extract patient radiological gradings corresponding to patient
-            patient_grades_dict = {}
-            for item in grades_dict[patient_id]:
-                key = f'IVD{item["IVD label"]}'
-                value = {
-                    k:v for k,v in item.items() 
-                    if k not in ['Patient', 'IVD label']
-                }
-                patient_grades_dict[key] = value
+            patient_grades_dict = grades_dict[patient_id]
 
             # Prepare example return dict
             return_dict = {'patient_id':patient_id, 'scan_type':scan_type}
@@ -555,4 +478,4 @@ class SPIDER(datasets.GeneratorBasedBuilder):
                 return_dict['rad_gradings'] = patient_grades_dict
             
             # Yield example
-            yield (scan_id, return_dict)
+            yield scan_id, return_dict