4. Getting Started with Your Own Data¶

Run the workflow on the hosted study, or replace one directory with your own aligned NIfTI dose and masks.

In [1]:

from dosemetrics import Dose
from dosemetrics.io import load_structure_set
from dosemetrics.metrics import conformity, homogeneity
from dosemetrics.utils import (
    download_example_data,
    dose_statistics_table,
    plot_dose_slice,
    plot_metric_values,
)

custom_data_dir = None  # For example: Path("/data/my-study")
study_dir = custom_data_dir or download_example_data("test_subject")

from dosemetrics import Dose from dosemetrics.io import load_structure_set from dosemetrics.metrics import conformity, homogeneity from dosemetrics.utils import ( download_example_data, dose_statistics_table, plot_dose_slice, plot_metric_values, ) custom_data_dir = None # For example: Path("/data/my-study") study_dir = custom_data_dir or download_example_data("test_subject")

Load and validate the study¶

In [2]:

dose = Dose.from_nifti(study_dir / "Dose.nii.gz", name=study_dir.name)
structures = load_structure_set(study_dir, format="nifti")
incompatible = [name for name, item in structures if not dose.is_compatible_with_structure(item)]
assert not incompatible, f"Incompatible structures: {incompatible}"
print(f"{dose.shape} dose grid with {len(structures)} compatible structures")

dose = Dose.from_nifti(study_dir / "Dose.nii.gz", name=study_dir.name) structures = load_structure_set(study_dir, format="nifti") incompatible = [name for name, item in structures if not dose.is_compatible_with_structure(item)] assert not incompatible, f"Incompatible structures: {incompatible}" print(f"{dose.shape} dose grid with {len(structures)} compatible structures")

(128, 128, 128) dose grid with 16 compatible structures

Compute reference-free structure metrics¶

In [3]:

dose_statistics_table(dose, structures).round(2)

dose_statistics_table(dose, structures).round(2)

Out[3]:

	Volume (cc)	Mean dose (Gy)	Minimum dose (Gy)	Maximum dose (Gy)	D98 (Gy)	D95 (Gy)	D50 (Gy)	D02 (Gy)
Structure
OpticNerve_L	0.83	30.34	24.17	44.98	24.42	24.80	29.10	42.93
Cochlea_R	0.08	59.48	59.34	59.62	59.35	59.36	59.48	59.62
CTV	254.51	59.26	44.33	63.87	50.72	53.93	60.00	61.52
Lens_L	0.22	17.66	15.43	19.16	15.56	15.74	17.52	19.16
OpticNerve_R	1.07	54.64	36.32	58.72	37.72	41.53	57.97	58.59
Cochlea_L	0.22	35.35	34.29	36.91	34.36	34.46	35.14	36.88
Lens_R	0.26	16.49	14.90	18.70	14.97	15.09	16.01	18.57
PTV	343.03	58.13	31.72	63.90	45.08	48.17	59.81	61.48
LacrimalGland_L	0.46	21.22	18.68	22.86	19.31	19.61	21.26	22.82
Eye_R	10.14	24.38	4.53	49.21	10.40	12.91	23.51	44.47
LacrimalGland_R	0.45	33.03	19.93	54.92	21.98	22.50	31.18	50.31
Chiasm	0.96	54.46	43.19	60.52	43.86	45.23	56.66	59.81
GTV	74.50	60.29	51.97	63.87	58.37	58.96	60.36	61.89
Brain	971.04	34.14	14.38	62.40	20.32	21.41	31.02	60.59
Eye_L	9.43	20.21	4.77	25.37	13.29	15.27	20.61	24.39
Brainstem	31.06	58.42	41.11	61.65	47.93	50.66	60.00	61.19

Compute target-quality metrics¶

In [4]:

target = structures["PTV"]
prescription_dose = 60.0
quality = {
    "Coverage": conformity.compute_coverage(dose, target, prescription_dose),
    "Spillage": conformity.compute_spillage(dose, target, prescription_dose),
    "Conformity index": conformity.compute_conformity_index(dose, target, prescription_dose),
    "Homogeneity index": homogeneity.compute_homogeneity_index(dose, target),
}
quality

target = structures["PTV"] prescription_dose = 60.0 quality = { "Coverage": conformity.compute_coverage(dose, target, prescription_dose), "Spillage": conformity.compute_spillage(dose, target, prescription_dose), "Conformity index": conformity.compute_conformity_index(dose, target, prescription_dose), "Homogeneity index": homogeneity.compute_homogeneity_index(dose, target), } quality

Out[4]:

{'Coverage': 0.4398190256302619,
 'Spillage': 0.556988489546629,
 'Conformity index': 0.4430115104533709,
 'Homogeneity index': 0.2742403973788274}

In [5]:

plot_dose_slice(
    dose,
    structures=structures,
    structure_names=["PTV", "Brainstem", "Chiasm"],
    contour_colors={"PTV": "white", "Brainstem": "lime", "Chiasm": "cyan"},
    prescription_dose=prescription_dose,
    cmap="turbo",
);
plot_metric_values(quality, title="Target-quality metrics", ylim=(0, 1.1));

plot_dose_slice( dose, structures=structures, structure_names=["PTV", "Brainstem", "Chiasm"], contour_colors={"PTV": "white", "Brainstem": "lime", "Chiasm": "cyan"}, prescription_dose=prescription_dose, cmap="turbo", ); plot_metric_values(quality, title="Target-quality metrics", ylim=(0, 1.1));