mioXpektron.baseline.baseline_base

Functions

baseline_correction(intensities[, method, ...])

Baseline-correct a 1‑D spectrum with pybaselines or custom filters.

baseline_method_names()

Return a sorted list of available baseline algorithms.

compute_metrics(y_corr, y_raw, baseline, x)

Compute RFZN, NAR, SNR, BBI, BR, NBC for a single corrected spectrum.

read_spectrum_table(path)

Read a ToF‑SIMS table from CSV/TSV/TXT with flexible separators.

small_param_grid_preset([n_points])

A compact parameter grid for common methods.

Classes

MetricResult(rfzn, nar, snr, bbi, br, nbc)

mioXpektron.baseline.baseline_base.baseline_method_names()[source]

Return a sorted list of available baseline algorithms.

Based on pybaselines.Baseline public callables, plus two custom filters (“median_filter”, “adaptive_window”) and a ‘poly’ alias. A few methods that are not 1‑D safe or impractically slow are removed.

Return type:

List[str]

mioXpektron.baseline.baseline_base.small_param_grid_preset(n_points=None)[source]

A compact parameter grid for common methods.

Keys must match pybaselines.Baseline method names (plus ‘poly’ and our two filters).

Parameters:

n_points (int, optional) – Number of data points in spectrum. If provided, window_size will be calculated adaptively as a percentage of data size. If None, uses moderate defaults suitable for ~100K point spectra.

Returns:

Parameter grid with method names as keys

Return type:

dict

Notes

Window sizes are calculated as: - Small: 0.05% of data (min 51) - Medium: 0.10% of data (min 101) - Large: 0.20% of data (min 501)

This adaptive scaling ensures that filter methods perform consistently across datasets of different sizes. Fixed window sizes work poorly: - For 10K points: window=101 is 1.0% (OK) - For 1M points: window=101 is 0.01% (too small, causes jagged baselines)

Examples

>>> # Auto-scale for 938K point spectrum
>>> grid = small_param_grid_preset(n_points=938000)
>>> grid['median_filter']
[{'window_size': 469}, {'window_size': 938}, {'window_size': 1876}]

>>> # Use defaults for unknown size
>>> grid = small_param_grid_preset()
>>> grid['median_filter']
[{'window_size': 501}, {'window_size': 1001}, {'window_size': 2001}]
mioXpektron.baseline.baseline_base.baseline_correction(intensities, method='airpls', window_size=101, poly_order=4, clip_negative=True, return_baseline=False, **kwargs)[source]

Baseline-correct a 1‑D spectrum with pybaselines or custom filters.

Parameters:

  • intensities (array-like) – Raw y values.

  • method (str) – Algorithm name; see baseline_method_names().

  • window_size (int) – Kernel width for the two custom filters.

  • poly_order (int) – Polynomial order for the ‘poly’ alias.

  • clip_negative (bool) – If True, negative corrected values are set to 0.

  • return_baseline (bool) – If True, also return the estimated baseline.

  • **kwargs – Forwarded to the chosen algorithm (e.g. lam=1e6, p=0.01).

Return type:

corrected or (corrected, baseline)

mioXpektron.baseline.baseline_base.read_spectrum_table(path)[source]

Read a ToF‑SIMS table from CSV/TSV/TXT with flexible separators.

Parameters:

path (str | Path)

Return type:

DataFrame

class mioXpektron.baseline.baseline_base.MetricResult(rfzn: 'float', nar: 'float', snr: 'float', bbi: 'float', br: 'float', nbc: 'float')[source]

Bases: object

Parameters:
rfzn: float
nar: float
snr: float
bbi: float
br: float
nbc: float
mioXpektron.baseline.baseline_base.compute_metrics(y_corr, y_raw, baseline, x, noise_mask=None, topk=5, raw_noise_quantile=0.2)[source]

Compute RFZN, NAR, SNR, BBI, BR, NBC for a single corrected spectrum.

Notes

  • RFZN: RMS of y_corr in baseline-only region (noise_mask). If mask is not supplied, it is derived from the raw intensities (bottom-q).

  • NAR: sum(-y_corr[y<0]) / sum(|y_corr|); lower is better.

  • SNR: median prominence of top-K peaks divided by noise sigma.

  • BBI: median(y_corr[noise_mask]); lower magnitude is better.

  • BR: RMS of d²(baseline)/dx² in baseline-only regions; requires baseline and x.

  • NBC: fraction of points where y_corr < 0 (before clipping).

Parameters:
Return type:

MetricResult