ProstateNetLoaders 0.0.1

Prostate Net loading functions

Prostate Net Loader

Prostate Net Loader contains functions to assist the Data Mining and loading process of Patients originated from the ProCAncer-AI Horizon's 2020 project.
The package is under construction for the time being therefore any suggestion would be appreciated.

Installation

Install the project via pip or pull the repo

pip install PROSTATENETLOADER == 0.0.1

Usage/Examples Series Parser tool

Detailed explanation of the series parser tool is presented at the ParquetParser_Examples.ipynb

Usage/Examples PROSTATENETLOADER Module

Examples could be found in Module_Examples.ipynb regarding the package. An example for a single patient is presented below

Single Patient

a) Import Libraries

import pandas as pd
import SimpleITK as sitk
import ProstateNetLoaders

b) Set the patient folder path and the csv extracted by the sequence selector tool

pth = "PCa-..."
metadata= pd.read_csv("results.csv", 
                names=["patient_id", "study_uid", 
                "series_uid", "series_type", "series_type_heuristics"])

c) Execute loaders and pick orientation ("AX","COR", "SAG") and sequence ("T2","ADC","DWI") and whether to be AI sequence parser (Heuristics = False) or Heuristics = True

a = ProstateNetLoaders.ExecuteLoader.Execute(pth, metadata,  Heuristics = True) 
a.LoadArrays(orientation="AX", seq="T2")

d) Get dictionaries where keys are the series names, values are the Image numpy arrays

pat,ann = a.GetItems() 

Batch Loading

The structure of the folders should be like this

pth_batch = "Patients"
patients = {}
Sequence = "T2" # pick you sequence between "T2", "ADC", "DWI"
T2_absence = [] # Store the names of the failed patients
for patient in os.listdir(pth_batch):
    pat = os.path.join(pth_batch,patient)
    a = ProstateNetLoaders.ExecuteLoader.Execute(pat, metadata)
    try:
        a.LoadArrays(orientation="AX", seq=Sequence)
        pat,ann = a.GetItems()
        patients.update({patient:{Sequence:np.array(list(pat.values())[0]),"Lesion": np.array(list(ann.values())[0])}})
    except: 
        T2_absence.append(patient)
        continue

Authors

• Dimitris Zaridis
• Harry Kalantzopoulos
• Eugenia Mylona
• Nikolaos Tachos
• José Guilherme Almeida

Badges

Add badges from somewhere like: shields.io

License

MIT