QB0uRAZM

1import os
2from pprint import pprint
3import sqlite3
4from sqlite3 import OperationalError
5
6import pydicom
7from pydicom.filereader import read_dicomdir
8
9base_folder = os.environ.get("NELSON", "/media/lubuntu/EOL/archive")
10
11# There are the fields we wish to extract.
12headers_to_extract = [
13    "AccessionNumber",
14    "AcquisitionDate",
15    "AcquisitionNumber",
16    "AcquisitionTime",
17    "ContentDate",
18    "ContentTime",
19    "InstanceNumber",
20    "PatientBirthDate",
21    "PatientID",
22    "PatientName",
23    "PatientSex",
24    "SeriesDate",
25    "SeriesDescription",
26    "SeriesInstanceUID",
27    "SeriesNumber",
28    "SeriesTime",
29    "StudyDate",
30    "StudyDescription",
31    "StudyID",
32    "StudyInstanceUID",
33    "StudyTime",
34]
35
36
37# Connnect to database.
38conn = sqlite3.connect("records.sql")
39c = conn.cursor()
40
41
42# Try to create table, if it not yet exists.
43def create_table():
44    header_name_type = (f"{header} TEXT" for header in headers_to_extract)
45    c.execute("CREATE TABLE dicom_files ({})".format(",".join(header_name_type)))
46    conn.commit()
47
48
49try:
50    create_table()
51except OperationalError:
52    pass
53
54# Collect all files in this folder.
55#files = [
56#    os.path.join(base_folder, file)
57#    for file in os.listdir(base_folder)
58#    if os.path.isfile(os.path.join(base_folder, file))
59#]
60filelist = '/home/lubuntu/Desktop/archive.lst'
61files = [os.path.join(base_folder, line) for line in open(filelist).read().splitlines()]
62
63def get_header_values(dcm):
64    """
65    Extract header values, when available.
66    """
67    values = []
68    for key in headers_to_extract:
69        try:
70            value = getattr(dcm, key)
71        except AttributeError:
72            value = ""
73        values.append(str(value))
74    return values
75
76
77def header_values_to_database(values):
78    """
79    Store the key-value pairs in the database.
80    """
81    query = "INSERT INTO dicom_files ({})".format(",".join(headers_to_extract))
82    query += " VALUES ({})".format(",".join(("?",) * len(values)))
83    c.execute(query, values)
84    conn.commit()
85
86
87previous_patient_id = None
88MAX_ITER = 10
89# Extract headers from DICOM file and store in database.
90for i, file in enumerate(files):
91    # 1) Read dicom file.
92    try:
93        dcm = pydicom.dcmread(file)
94    except Exception as e:
95        print(e)
96        continue
97
98    values = get_header_values(dcm)
99
100    # 2) Store in database.
101    header_values_to_database(values)
102
103    # 3) Print progress.
104    if dcm.PatientID != previous_patient_id:
105        print("")
106        print(f"{dcm.PatientID}", flush=True, end="")
107        previous_patient_id = dcm.PatientID
108    else:
109        print(".", flush=True, end="")
110
111    if MAX_ITER > 0 and i >= MAX_ITER:
112        print('')
113        print('Maximum reached')
114        break
115
116
117c.close()