LUMINOUS

LUMINOUS provides a curated dataset of lumbar multifidus (LM) ultrasound images with manually segmented binary masks to support development and validation of automated segmentation algorithms for quantitative assessment of LM cross-sectional area (CSA) and echo intensity (EI).

Key Features:

• Imaging modality: B-mode ultrasound images of the lumbar multifidus (LM) muscle.
• Anatomical coverage: Left and right LM at the L5 vertebral level captured in prone and standing positions.
• Annotations: Manually segmented binary masks provided as ground truth reference standards.
• Population: Data acquired from 109 young athletic adults affiliated with Concordia University's varsity teams.
• Analytical focus: Enables extraction and evaluation of cross-sectional area (CSA) and echo intensity (EI) metrics.
• Computational intent: Prepared to support development and validation of deep learning–based automatic segmentation algorithms.
• Scientific target: Focuses on the lumbar multifidus, a paraspinal muscle relevant to lower back pain and rehabilitation studies.

Scientific Applications:

• Algorithm development and benchmarking: Training and validation of automated and deep learning segmentation methods for LM.
• Quantitative morphology: Measurement of LM cross-sectional area (CSA) for morphological analyses.
• Echo-intensity analysis: Evaluation of echo intensity (EI) as a tissue composition or quality metric.
• Comparative studies and standardization: Facilitation of uniform LM measurement protocols and comparative analyses across studies.
• Clinical research on back pain: Support for investigations linking LM morphology and function to lower back pain and rehabilitation outcomes.

Methodology:

Ultrasound images are paired with manually segmented binary masks that serve as ground truth for developing and validating deep learning–based automatic segmentation algorithms.