diff --git a/clinica/iotools/converters/ixi_to_bids/ixi_to_bids_utils.py b/clinica/iotools/converters/ixi_to_bids/ixi_to_bids_utils.py
index 467e53e22..e42ecc1e7 100644
--- a/clinica/iotools/converters/ixi_to_bids/ixi_to_bids_utils.py
+++ b/clinica/iotools/converters/ixi_to_bids/ixi_to_bids_utils.py
@@ -1,18 +1,28 @@
 # todo : verify typing compatibility with user input
+import re
 from pathlib import Path
+from typing import List
 
 import pandas as pd
 
 
 def get_subjects_list_from_data(data_directory: Path) -> List[str]:
-    img_list = list(data_directory.rglob(pattern="*.nii.gz"))
-    return
+    return list(
+        dict.fromkeys(
+            re.match(r"IXI\d{3}", path.name).group(0)
+            for path in data_directory.rglob(pattern="IXI*.nii.gz")
+        )
+    )
 
 
 def filter_subjects_list(
     subjects_list: List[str], clinical_data: pd.DataFrame
 ) -> List[str]:
-    return [subject for subject in subjects_list if subject in clinical_data["IXI_ID"]]
+    return [
+        subject
+        for subject in subjects_list
+        if subject in clinical_data["IXI_ID"].values
+    ]
 
 
 def read_ixi_clinical_data(clinical_data_path: Path) -> pd.DataFrame:
@@ -22,3 +32,22 @@ def read_ixi_clinical_data(clinical_data_path: Path) -> pd.DataFrame:
         lambda x: "IXI" + "0" * (3 - len(str(x))) + str(x)
     )
     return clinical_data
+
+
+def get_data_df(data_directory: Path) -> pd.DataFrame:
+    # todo : works for all except DTI !!! (I mean we should drop it after this)
+    df = pd.DataFrame(
+        {"img_path": [path for path in data_directory.rglob(pattern="IXI*.nii.gz")]}
+    )
+    df = (
+        df.assign(img_name=lambda df: df.img_path.apply(lambda x: x.name))
+        .assign(img_name_no_ext=lambda df: df.img_name.apply(lambda x: x.split(".")[0]))
+        .assign(subject=lambda df: df.img_name_no_ext.apply(lambda x: x.split("-")[0]))
+        .assign(hospital=lambda df: df.img_name_no_ext.apply(lambda x: x.split("-")[1]))
+        .assign(modality=lambda df: df.img_name_no_ext.apply(lambda x: x.split("-")[3]))
+        .assign(session="ses-M000")
+    )
+    return df
+
+
+# todo : question DTI - sessions différentes ou images différentes ?