Hey @venkatesh71097, do you have any more specifics about the database and files than "p00"? One thing to note is that version >=3.0.0 shifted from Physiobank to Physionet which means it is using the most updated versions of the databases and, in some cases, some content may have changed. The underlying functionality in the code should still be the same but let's see what we can figure out here! Thanks for posting!

Hey, so I'm trying to check the number of patients having 'PLETH' channel in p00 file of the mimic matched subset database that was released a month back, as I want only the pleth channel from the signals. It is from physionet.org. I've p00 files placed in my directory. Let me know if you need any further details.

Hey @venkatesh71097, do you have any code that can reproduce this issue? I want to check the specific differences between the two versions when replicating this bug that you are receiving. Thanks!

import pandas as pd
import os
from IPython.display import display
import matplotlib.pyplot as plt
%matplotlib inline
import numpy as np
import os
import shutil
import posixpath
import time
import wfdb

count_hea = 0
count_pleth = 0
for subfile in sorted(os.listdir('physionet.org/files/mimic3wdb-matched/1.0/p00')):
count_hea = 0
if subfile.startswith('p'):
for subsubfile in sorted(os.listdir(os.path.join('physionet.org/files/mimic3wdb-matched/1.0/p00/' , subfile))):
if subsubfile.startswith('p') and count_hea == 0 and not subsubfile[:-4].endswith('n'):
try:
i = i + 1
print(subsubfile)
count_hea = count_hea + 1
strip = subsubfile[:-4]
record = wfdb.rdrecord('physionet.org/files/mimic3wdb-matched/1.0/p00/' + subfile + '/' + strip, channel_names = ['PLETH'], return_res = 16)
r = record.dict
print(r['sig_name'])
if r['sig_name'] != None:
if r['sig_name'][0] == 'PLETH':
count_pleth = count_pleth + 1
print(count_pleth)
except Exception:
pass
print(count_pleth)

Hey lucas , you can try this out. Sorry for the delay. I missed the notification!

Hey @venkatesh71097, I just downloaded the files locally and ran your code and reproduced both numbers.

One file it failed on was p00/p000298/p000298-2148-01-02-18-16 which gave an error of ValueError: Samples were not loaded correctly when the return_res=16, ValueError: cannot reshape array of size 375 into shape (2) when the return_res=32, and loads correctly when the return_res=64. The return_res should only be changing the significant figures so I have to see what's going on further.

For now, you can catch the try statement with ValueError and try it again with return_res=64 and that should work though I haven't tested it myself. Will give it a go now.

I made the fix and got the same number for both the current Github version and 2.2.1. Hope this helps!