diff --git a/src/fundus/publishers/de/mdr.py b/src/fundus/publishers/de/mdr.py
index 9a1e93bfd..3fe464dc2 100644
--- a/src/fundus/publishers/de/mdr.py
+++ b/src/fundus/publishers/de/mdr.py
@@ -3,6 +3,7 @@
 from typing import List, Optional, Pattern
 
 from lxml.cssselect import CSSSelector
+from lxml.etree import XPath
 
 from fundus.parser import ArticleBody, BaseParser, ParserProxy, attribute
 from fundus.parser.utility import (
@@ -17,7 +18,13 @@
 class MDRParser(ParserProxy):
     class V1(BaseParser):
         _author_substitution_pattern: Pattern[str] = re.compile(r"MDR \w*$|MDR \w*-\w*$|MDRfragt-Redaktionsteam|^von")
-        _paragraph_selector = CSSSelector("div.paragraph")
+        # regex examples: https://regex101.com/r/2DSjAz/1
+        _source_detection: str = r"^((MDR (AKTUELL ){0,1}\(([A-z]{2,3}(\/[A-z]{2,3})*|[A-z, ]{2,50}))\)|(Quell(e|en): (u.a. ){0,1}[A-z,]{3,4})|[A-z]{2,4}(, [A-z]{2,4}){0,3}( \([A-z]{2,4}\)){0,1}$|[A-z]{2,4}\/[A-z(), \/]{3,10}$)"
+        _paragraph_selector = XPath(
+            f"//div[@class='paragraph '] "
+            f"/p[not(re:test(em, '{_source_detection}') or re:test(text(), '{_source_detection}'))]",
+            namespaces={"re": "http://exslt.org/regular-expressions"},
+        )
         _summary_selector = CSSSelector("p.einleitung")
         _subheadline_selector = CSSSelector("div > .subtitle")
         _author_selector = CSSSelector(".articleMeta > .author")
@@ -33,7 +40,10 @@ def body(self) -> ArticleBody:
 
         @attribute
         def topics(self) -> List[str]:
-            return generic_topic_parsing(self.precomputed.meta.get("news_keywords"))
+            if self.precomputed.meta.get("news_keywords") is not None:
+                return generic_topic_parsing(self.precomputed.meta.get("news_keywords"))
+            else:
+                return generic_topic_parsing(self.precomputed.meta.get("keywords"))
 
         @attribute
         def publishing_date(self) -> Optional[datetime.datetime]: