Automated estimation of image quality for coronary computed tomographic angiography using machine learning.

Area of Science:

• Radiology and Medical Imaging
• Artificial Intelligence in Healthcare
• Cardiovascular Imaging

Background:

• Coronary computed tomography angiography (CCTA) is crucial for diagnosing coronary artery disease.
• Accurate assessment of CCTA image quality (IQ) is essential for reliable interpretation and clinical trial outcomes.
• Manual IQ assessment is subjective and time-consuming, highlighting the need for automated solutions.

Purpose of the Study:

• To evaluate the efficacy of a fully automated machine learning (ML) method for assessing CCTA image quality (IQ).
• To compare the performance of the automated IQ assessment against traditional visual analysis.
• To determine the reproducibility and standardization potential of the automated IQ assessment method.

Main Methods:

• An ML model was trained on 75 CCTA studies, mapping image features to IQ scores derived from manual ground truth.
• The automated method was validated on 50 CCTA studies and tested on 172 CCTA studies.
• Performance was evaluated by comparing automated IQ scores with visual assessments using a 5-point Likert scale and Cohen's kappa statistic.

Main Results:

• The automated method achieved an area under the curve of 0.96 in the validation set.
• A Cohen's kappa statistic of 0.67 indicated substantial agreement between automated and visual IQ assessments in the test set (p < 0.01).
• The ML method accurately categorized most CCTA studies into good, fair, or poor visual IQ groups.

Conclusions:

• Fully automated IQ assessment of CCTA datasets using ML is reproducible and yields results comparable to visual analysis.
• The proposed ML method offers a standardized approach to IQ assessment, potentially improving consistency across different datasets and trials.
• Automated IQ assessment facilitates more reliable standardization of clinical trial results.