BioSR+: Paired Low- and High-Resolution Images of Subcellular Structures

BioSR+ extends the BioSR dataset with image pairs for five distinct subcellular structures: CCPs, ER, MTs, F-actin, and Myosin-IIA. Each region of interest includes eight different signal levels, providing a controlled benchmark for super-resolution algorithm development. The dataset is associated with peer-reviewed publications in Nature Methods and Nature Biotechnology.

Use Cases

• Training super-resolution neural networks on paired low- and high-resolution biological images.
• Benchmarking algorithm performance across different subcellular structures like microtubules and actin filaments.
• Evaluating the robustness of super-resolution methods to varying signal levels within a region of interest.
• Developing methods for sustained live imaging of rapid subcellular processes.

Strengths

• Contains paired low- and high-resolution images, providing ground truth for supervised learning.
• Includes data for five specific biological structures, enabling structure-specific analysis.
• Provides eight signal levels per ROI, allowing for controlled studies on signal-to-noise robustness.

Limitations

• Key metadata such as exact row count, file size, and specific column/feature names are not provided in the available sources.
• The license is specified for non-commercial use only, which restricts commercial applications.
• The last update date is unknown, making it difficult to assess the dataset's currency.

Provenance

Source

Chang Qiao, Tsinghua University

Collection Method

Likely contains microscopy images gathered for super-resolution research, as described in associated Nature Methods and Nature Biotechnology papers.