preparation_data_IS.py

import glob
from MetricsReloaded.processes.overall_process import ProcessEvaluation
import os
import nibabel as nib
import pickle as pkl
import pandas as pd
from MetricsReloaded.utility.utils import MorphologyOps


list_reffile = glob.glob("examples/data/Ref/CorrectLesion*CLA*")
list_predfile = glob.glob("examples/data/Pred/CorrectLesion*RAP94*")
list_det = []
list_seg = []
for f in list_reffile:
    name = os.path.split(f)[1]
    name = name.split("Ref")[0]
    name = name.split("Lesion_")[1]
    if not os.path.exists("examples/results/Det94AvFin_%s.csv" % name):

        list_pospred = [c for c in list_predfile if name in c]
        if len(list_pospred) == 1:
            ref = nib.load(f).get_fdata()
            pred = nib.load(list_pospred[0]).get_fdata()
            # ref = nib.load('/Users/csudre/Data/B-RAPIDD/CLA66/CorrectLesion_B-RAP_0007_01_CLA66.nii.gz').get_fdata()
            # pred = nib.load('/Users/csudre/Data/B-RAPIDD/RAP66/CorrectLesion_B-RAP_0007_01_RAP66.nii.gz').get_fdata()
            ref_bin = ref >= 0.5
            pred_bin = pred >= 0.5
            print(f, list_pospred[0])
            list_ref, _, _ = MorphologyOps(ref_bin, neigh=6).list_foreground_component()
            list_pred, _, _ = MorphologyOps(
                pred_bin, neigh=6
            ).list_foreground_component()
            pred_prob = []
            pred_class = []
            ref_class = []
            for k in list_pred:
                pred_class.append(1)
                pred_prob.append(1)
            for k in list_ref:
                ref_class.append(1)

            list_values = [1]
            file = list_pospred
            dict_file = {}
            dict_file["pred_loc"] = [list_pred]
            dict_file["ref_loc"] = [list_ref]
            dict_file["pred_prob"] = [pred_prob]
            dict_file["ref_class"] = [ref_class]
            dict_file["pred_class"] = [pred_class]
            dict_file["list_values"] = list_values
            dict_file["file"] = file
            # f = open("TestDataBRAP_%s.pkl"%name, "wb")  # Pickle file is newly created where foo1.py is
            # pkl.dump(dict_file, f)  # dump data to f
            # f.close()
            PE = ProcessEvaluation(
                dict_file,
                "Instance Segmentation",
                localization="mask_iou",
                file=list_pospred,
                flag_map=True,
                assignment="greedy_matching",
                measures_overlap=[
                    "fbeta",
                    "numb_ref",
                    "numb_pred",
                    "numb_tp",
                    "numb_fp",
                    "numb_fn",
                ],
                measures_mcc=[],
                measures_pcc=[
                    "fbeta",
                    "numb_ref",
                    "numb_pred",
                    "numb_tp",
                    "numb_fp",
                    "numb_fn",
                ],
                measures_mt=[],
                measures_boundary=["masd", "nsd", "boundary_iou"],
                measures_detseg=["PQ"],
                thresh_ass=0.000001,
            )
            df_resdet, df_resseg, df_resmt, df_resmcc = PE.process_data()
            df_resdet["id"] = name
            df_resseg["id"] = name
            df_resdet.to_csv("examples/results/Det94AvFin_%s.csv" % name)
            df_resseg.to_csv("examples/results/Seg94AvFin_%s.csv" % name)

            list_det.append(df_resdet)
            list_seg.append(df_resseg)
df_resdetall = pd.concat(list_det)
df_ressegall = pd.concat(list_seg)
df_resdetall.to_csv("examples/results/Det94AvFin.csv")
df_ressegall.to_csv("examples/results/Seg94AvFin.csv")
print(df_resdet, df_resseg)