Differential Aging Signals in Abdominal CT Scans.

Area of Science:

• Gerontology
• Medical Imaging
• Computational Biology

Background:

• Aging is characterized by significant changes in body tissue composition.
• Studying these age-related tissue changes in depth has been challenging.
• Computed tomography (CT) scans offer a potential avenue for observing these alterations.

Purpose of the Study:

• To investigate age-related changes in abdominal tissues using CT scans.
• To apply pattern recognition and machine learning for detecting and quantifying these changes.
• To establish a model-agnostic approach for analyzing aging in tissues.

Main Methods:

• Utilized CT scans from the Baltimore Longitudinal Study of Aging (BLSA) participants.
• Trained machine classifiers to differentiate between younger (50-70 years) and older (80-99 years) age groups.
• Assessed classification accuracy as a measure of the aging signal in different tissues.

Main Results:

• Achieved high classification accuracies in discriminating age groups: 0.76 for males and 0.72 for females.
• Adipose tissue showed the highest accuracy (0.79 males, 0.71 females), followed by soft tissue (0.70 males, 0.68 females) and bone (0.65 males, 0.64 females).
• Demonstrated the effectiveness of machine learning in identifying age-related tissue variations.

Conclusions:

• Explored age-related differences in abdominal tissue morphology and texture using medical imaging and machine learning.
• This imaging technology provides advantages over traditional biomarkers for tracking biological aging.
• Machine learning analysis of medical images offers valuable insights into aging-related tissue changes and their consequences.