Electroencephalography (EEG) provides direct, millisecond-resolution measures of neural activity and is widely used across ages and experimental settings. However, EEG signals—particularly from infants—are highly susceptible to noise and artifacts, complicating preprocessing and analysis. APICE-Py (Automated Preprocessing for Infants Continuous EEG) is an open-source pipeline reimplemented in Python on top of MNE-Python, designed to provide scalable, flexible preprocessing for developmental and adult EEG datasets. The pipeline is a port of a MATLAB toolbox originally intended for infant EEG and aims to extend applicability across a broad age range and diverse recording systems.
APICE-Py is organized around three core design principles:
Adaptive artifact detection: artifact identification operates on continuous data using data-driven thresholds rather than fixed cutoffs, enabling adaptation to variable signal characteristics across participants, age groups, and recording hardware.
Hierarchical artifact correction: the pipeline combines localized short-segment correction using Principal Component Analysis (PCA) with broader segment and continuous-data correction via Spherical Spline Interpolation (SSI), allowing targeted repair of brief contamination and larger-scale interpolation where necessary.
Transparent reporting: APICE-Py produces comprehensive quality logs and decision-tracking outputs to document preprocessing actions, support reproducibility, and inform downstream analyses.
The source describes a processing strategy that emphasizes continuous-data handling and layered artifact mitigation. Short contaminated segments are corrected with PCA-based methods that remove principal components associated with artifacts, while larger or continuous data gaps are addressed with SSI-based interpolation across channels. Artifact detection thresholds are derived from the data itself rather than imposed fixed values, an approach intended to account for differences in noise levels and signal properties across datasets and age groups.
The pipeline's modular structure leverages MNE-Python for data handling and processing, enabling integration with common EEG workflows and tools in the Python ecosystem. The preprint summarizes the underlying algorithms and provides an implementation compatible with common EEG file formats.
APICE-Py has been implemented to support common EEG formats and to run within the MNE-Python environment. The reimplementation in Python is intended to improve scalability and integration with Python-based analysis pipelines. Specific supported file formats and interfaces are documented in the project repository; the abstract confirms compatibility with common EEG formats but does not list each supported file type.
The authors demonstrate use of APICE-Py on three example datasets that cover a range of ages and recording contexts: neonates, 5-month-old infants, and child-parent hyperscanned data. These datasets include recordings acquired with high-density wet electrode systems as well as mobile gel-based EEG systems, illustrating the pipeline's applicability across hardware configurations and experimental paradigms.
The source emphasizes that, although APICE-Py was developed with early developmental EEG in mind, it generalizes to child and adult datasets as well.
APICE-Py was benchmarked against the original MATLAB implementation of APICE. According to the preprint abstract, the Python reimplementation achieved comparable levels of data quality and trial retention relative to the MATLAB version. The abstract does not provide detailed numerical metrics, statistical comparisons, or specific trial-retention figures—those details are not reported in the provided source excerpt.
A key emphasis of APICE-Py is transparent reporting: the pipeline generates thorough quality logs and decision-tracking outputs that document preprocessing steps and choices. The authors highlight reproducibility as a central design goal.
The APICE-Py source code, documentation, and related materials are freely available on GitHub at: https://github.com/neurokidslab/apice-py. The preprint declares no competing interests.
The abstract and provided source text summarize the pipeline's principles and broad performance claims but do not report detailed quantitative results from benchmarking comparisons, parameter settings, or the full list of supported EEG file formats. Specific performance metrics, runtime characteristics, and example commands or configuration templates are not included in the abstract and would need to be consulted in the full manuscript or repository documentation for implementation details.
The preprint lists funding from the European Research Council and the Agence Nationale de la Recherche. The work is presented as a preprint and has not been peer reviewed at the time of posting. The code is released under the project repository referenced above.