| 856 | |
| 857 | |
| 858 | class PreprocessorFor2D_noNormalization(GenericPreprocessor): |
| 859 | def resample_and_normalize(self, data, target_spacing, properties, seg=None, force_separate_z=None): |
| 860 | original_spacing_transposed = np.array(properties["original_spacing"])[self.transpose_forward] |
| 861 | before = { |
| 862 | 'spacing': properties["original_spacing"], |
| 863 | 'spacing_transposed': original_spacing_transposed, |
| 864 | 'data.shape (data is transposed)': data.shape |
| 865 | } |
| 866 | target_spacing[0] = original_spacing_transposed[0] |
| 867 | data, seg = resample_patient(data, seg, np.array(original_spacing_transposed), target_spacing, 3, 1, |
| 868 | force_separate_z=force_separate_z, order_z_data=0, order_z_seg=0, |
| 869 | separate_z_anisotropy_threshold=self.resample_separate_z_anisotropy_threshold) |
| 870 | after = { |
| 871 | 'spacing': target_spacing, |
| 872 | 'data.shape (data is resampled)': data.shape |
| 873 | } |
| 874 | print("before:", before, "\nafter: ", after, "\n") |
| 875 | |
| 876 | if seg is not None: # hippocampus 243 has one voxel with -2 as label. wtf? |
| 877 | seg[seg < -1] = 0 |
| 878 | |
| 879 | properties["size_after_resampling"] = data[0].shape |
| 880 | properties["spacing_after_resampling"] = target_spacing |
| 881 | use_nonzero_mask = self.use_nonzero_mask |
| 882 | |
| 883 | assert len(self.normalization_scheme_per_modality) == len(data), "self.normalization_scheme_per_modality " \ |
| 884 | "must have as many entries as data has " \ |
| 885 | "modalities" |
| 886 | assert len(self.use_nonzero_mask) == len(data), "self.use_nonzero_mask must have as many entries as data" \ |
| 887 | " has modalities" |
| 888 | return data, seg, properties |
nothing calls this directly
no outgoing calls
no test coverage detected