We present a protocol and software for automatic deep learning-based segmentation of medical images guided by a graphical user interface (GUI) using the CODAvision algorithm. This GUI guided workflow simplifies the process of semantic segmentation by enabling users to train highly customizable deep learning models without extensive coding expertise. This protocol outlines best practices for creating robust training datasets, configuring model parameters, and optimizing performance for diverse biomedical image types. CODAvision enhances the usability of the CODA algorithm (Nature Methods, 2022) by streamlining parameter configuration, model training, and performance evaluation, automatically generating quantitative results and comprehensive reports.
We demonstrate robust performance across numerous medical image modalities and diverse biological questions. We provide sample results in data types including histology, magnetic resonance imaging (MRI), and computed tomography (CT), and in applications including quantification of metastatic burden in in vivo models, and deconvolution of spot-based spatial transcriptomics datasets.