Automated Quantitative Evaluation of Age-Related Thymic Involution on Plain Chest CT

Area of Science:

• Immunology
• Radiology
• Computational Biology

Background:

• The thymus is crucial for T-cell generation and its age-related decline (involution) is linked to pathologies.
• Current methods lack objective quantification of thymic involution from clinical images.

Purpose of the Study:

• To establish an objective, automated method for quantifying age-related thymic involution using plain chest computed tomography (CT) images.
• To define and segment the thymic region for quantification (TRQ) and develop an estimator of thymic volume (ETV).

Main Methods:

• Manual segmentation of the TRQ in 135 CT studies for neural network (NN) model training.
• Development of an NN model for automated TRQ segmentation.
• Calculation of ETV based on NN-segmented TRQ's Hounsfield unit (HU) values and volume in 853 healthy subjects.
• Quantile additive regression models used to analyze age and sex relationships with ETV.

Main Results:

• High correlation (r=0.996 for HU, r=0.986 for volume) between NN-segmented and manually segmented TRQ.
• ETV demonstrated an exponential decline with age (p < 0.001), confirming age-related thymic tissue loss.
• The developed method provides robust quantification of thymic involution.

Conclusions:

• The novel automated method accurately quantifies age-related thymic involution from chest CT scans.
• This approach may assist in predicting and classifying pathologies associated with thymic involution.
• The ETV serves as a reliable indicator of thymic tissue volume decline with age.