MRI Image Quality and Diagnostic Performance for Carotid Plaques, 69 Patients

A study dataset comparing deep learning-reconstructed MRI (MRI_DLR) to conventional and fast MRI for carotid atherosclerotic plaque assessment. It contains results from 69 patients recruited between May 2022 and April 2024, evaluating objective metrics like SNR and CNR, subjective image quality, and diagnostic performance against surgical pathology. The data was authored by Xixiang Chen and shared under a CC-BY-4.0 license.

Use Cases

• Benchmarking deep learning-based MRI reconstruction algorithms based on reported signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) metrics.
• Training or validating models for diagnosing carotid plaque features like hemorrhage and lipid nuclei based on the described diagnostic performance comparisons.
• Analyzing the trade-off between MRI scan time reduction and image quality based on the reported 63.2% average time reduction.
• Studying inter-method agreement for medical image assessment using the mentioned Cohen’s κ statistic.

Strengths

• Includes specific, quantified results such as a 12.4-minute (63.2%) reduction in average scanning time.
• Compares three distinct MRI methodologies (MRI_DLR, MRI_C, MRI_Fast) across multiple objective and subjective metrics.
• Diagnostic performance is validated against a ground truth, specifically surgical pathology findings.

Limitations

• The dataset is very small at 20.8 KB; the actual data structure, column definitions, and row count are unknown.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data may reflect temporal and methodological bias inherent to the single-center study design described.

Provenance

Source

figshare, author Xixiang Chen.

Collection Method

Clinical study involving 69 patients imaged with a 3.0 T MRI system (SIGNA Architect, GE Healthcare).

Time Range

Patient recruitment occurred between May 2022 and April 2024.

Freshness

Last updated 2026-04-17 04:11:49; freshness should be verified.