Try to retrieve the procedures of CT scan with contrast

I am trying to retrieve the procedures of CT scan with contrast.
The procedures_icd file contains the CT scan with contrast records.
I used procedure icd_code to get the CT scan with contrast records.

df = pd.read_csv("d_icd_procedures.csv")
#get the icd_code of CT scan with contrast
df1 = df.loc[df['long_title'].astype(str).str.contains('contrast',na=False, flags=re.IGNORECASE)]
df2 = df1.loc[df1['long_title'].astype(str).str.contains('(CT Scan)',na=False, flags=re.IGNORECASE)]

#icd_code of CT scan with contrast to a list
a = df2['icd_code'].tolist()
separator = '|'
b = separator.join(a)

df2 = pd.read_csv("procedures_icd.csv")
df2 = df2[['subject_id','hadm_id','icd_code']]

#use re to parse the icd_code of CT scan with contrast
df3 = df2.loc[df2['icd_code'].astype(str).str.contains(b,na=False, flags=re.IGNORECASE)]

df3.to_csv("procedure_contrast_id.csv",index=False)
print(df3['hadm_id'].nunique())

There are only 15 records of CT scan with contrast in all MIMIC-IV database! This doesn't make sense. The ratio of CT contrast is too low!

Do I miss anything? or how could I get the CT scan with contrast records?
Thank you very much.

Start by running the shown pandas filtering against d_icd_procedures.csv and procedures_icd.csv, focusing on how the ICD codes are combined and matched. Compare the resulting rows with the source procedure records; done means explaining the unexpectedly low count and documenting a reproducible way to retrieve the CT-with-contrast records.