Purpose

This section explains the rationale behind SlicerCART.

Rationale:

• Curating large, high-quality datasets is a resource-intensive foundation critical to developing artificial intelligence models that deliver high performance. However, many commonly used open-source solutions do not offer a way to easily navigate through a list of cases, perform version control or customize the annotation workflow to specific tasks. Some software suffer from complex installation processes and lack intuitive image visualization features.

• To address these challenges, SlicerCART—Slicer Configurable Annotation Review Tool—was developed as a solution to lower the barrier to dataset creation.

• This Slicer module provides a time-efficient, open-source platform for manual segmentation and classification across large datasets, with tools for quality control and structured data output.

• Moreover, when performing quality control of existing segmentation (whether manual or automatic), it would be great to have the possibility to directly modify the existing segments and get a new corrected versions, efficiently. Only few existing open-source solutions may offer yet this option, except SlicerCART.