MMV_Im2Im: an open-source microscopy machine vision toolbox for image-to-image transformation

Area of Science:

• Computational Biology
• Medical Imaging
• Machine Learning

Background:

• Deep learning (DL) has rapidly advanced computer vision and biomedical image analysis.
• Existing DL tools often require significant engineering expertise, hindering research focus.

Purpose of the Study:

• Introduce MMV_Im2Im, an open-source Python package for generic image-to-image transformations in bioimaging.
• Provide a user-friendly framework for diverse biomedical image analysis tasks.

Main Methods:

• Developed a generic image-to-image transformation framework using state-of-the-art machine learning engineering.
• Implemented MMV_Im2Im as an open-source Python package with comprehensive documentation and tutorials.

Main Results:

• Demonstrated the effectiveness and broad applicability of MMV_Im2Im on over 10 biomedical problems.
• Showcased its utility for semantic segmentation, instance segmentation, image restoration, and image generation.

Conclusions:

• MMV_Im2Im offers a valuable starting point for computational and experimental biomedical researchers.
• Facilitates the integration of deep learning-based image-to-image transformation into assay development for novel biomedical studies.