# battery_xct_workflows: extracting quality metrics from X-ray computed tomography of Li-ion cells

## TL;DR

This paper introduces an open-source Python method for analyzing X-ray images of Li-ion batteries to ensure quality and detect defects.

## Contribution

The novel contribution is an integrated workflow with pre-trained models for automated quality assurance in Li-ion battery manufacturing.

## Key findings

- The workflow addresses electrode overhang, canister alignment, and winding uniformity in Li-ion cells.
- The method uses pre-trained U-Net models for segmentation and provides reproducible QA metrics.
- It lowers the barrier for adopting XCT-based QA in both research and industrial settings.

## Abstract

battery_xct_workflows is an open, Python notebook-based method for deriving quantitative quality assurance (QA) metrics from X-ray computed tomography (XCT) of cylindrical Li-ion cells. The workflows target three recurring industrial QA questions: (i) are electrode overhang regions within design tolerance, (ii) is the canister geometry and alignment acceptable, and (iii) is the internal winding uniform and free from gross defects. The method combines preprocessing, segmentation (including optional pre-trained U-Net models), and metric calculation into a series of executable notebooks that can be run locally or via Binder using public example datasets.

By packaging data, code, models, and narrative explanations together, this method lowers the barrier to adopting XCT-based QA in both research and industrial settings. Users can reproduce the provided examples, adapt individual steps to their own scanners and cell formats, and extend the notebooks to new metrics while retaining a transparent audit trail.

Image, graphical abstract