import numpy as np
from PIL import Image
import tempfile
import pytesseract
import os
import sys
import re
from pathlib import Path
"""
Extract text from image
@Param
filename : Image file name
@return
file containing extracted text
"""
def ocr(filename,APP_ROOT):
target = os.path.join(APP_ROOT, 'ocr/')
if not os.path.isdir(target):
os.mkdir(target)
text = str(((pytesseract.image_to_string(Image.open(filename),lang='fra'))))
simple_filename = Path(filename).stem
ocr_path = "/".join([target, simple_filename])
file = open(ocr_path,"w")
file.write(text)
file.close()
return ocr_path
"""
extract medical data from ocr text
@Param
textFile : text file
@return
data structure of medical data
"""
def extract_medical_data(textfile,APP_ROOT):
data = {}
medical_item_data ={}
medical_data ={}
all_medical_data =[]
#counter for doctor and patient match
doctor_match_counter = 0
patient_match_conter = 0
print("app root",APP_ROOT)
target = os.path.join(APP_ROOT, 'medicaldata/')
print("medical targer",target)
if not os.path.isdir(target):
os.mkdir(target)
pass
pattern_labels = os.path.join(APP_ROOT, 'dictionnary/medical_dict.txt')
dictionnaireLines = open(pattern_labels, encoding="utf8").read()
patterns = dictionnaireLines.split(",")
print("text file",textfile)
print(type(textfile))
try:
with open(textfile, encoding="utf8") as f:
#with open('/home/kommit/Documents/cours/tx52/medical-record-nlp/ocr/output.txt', encoding="utf8") as f:
lines = f.readlines()
for line in lines:
#regex serch for patient and doctor
patient_last_first_name_match = re.search ("(?P<title>(?:MR|MONSIEUR|MADEMOISELLE|MLLE|MADAME|MME) (?P<name>.*?)\s*$)", line, re.IGNORECASE)
doctor_last_first_name_match = re.search ("(?P<title>(?:DR) (?P<name>.*?)\s*$)", line, re.IGNORECASE)
if(patient_last_first_name_match):
print(patient_last_first_name_match.group(1).split("\n")[0])
#regex can match many patient and doctor so when filter then with counter and get the first
if patient_match_conter == 0: data['patient_name'] = patient_last_first_name_match.group(1).split("\n")[0]
patient_match_conter +=1
if(doctor_last_first_name_match):
print(doctor_last_first_name_match.group(1))
if doctor_match_counter == 0:data['doctor_name'] = doctor_last_first_name_match.group(1).split("\n")[0]
doctor_match_counter +=1
#regex search for medical entity
for pattern in patterns:
if re.search(pattern, line, re.I):
valeurItems = re.findall ("<?\d\d?.+[0-9]*", line)
print(pattern,valeurItems,"-----------------------\n")
str_valeur_items = list_to_str(valeurItems)
medical_item_data_match = re.search("^(\S+) (\S+) (\S+ \S+ \S+) ?(|(\S+))$",str_valeur_items)
if(medical_item_data_match):
group_len = len(medical_item_data_match.groups())
print(group_len,"************")
medical_item_data['code'] = pattern
medical_item_data['value'] = medical_item_data_match.group(1)
medical_item_data['unity'] = medical_item_data_match.group(2)
if group_len == 5 : medical_item_data['normal'] = medical_item_data_match.group(3)
medical_item_data['history'] = medical_item_data_match.group(4)
print('pattern',pattern)
p = pattern
#medical_data[pattern] = medical_item_data
print("medicaldata",medical_data)
all_medical_data.append(medical_item_data)
medical_item_data ={}
print("all----------",all_medical_data,"all")
data['medical_data'] = all_medical_data
print(data)
return data
except IOError:
print("Le fichier pas etre ouvert")
pass
def list_to_str(str_list):
res = ' '.join([str(elem) for elem in str_list])
return res
