Associating the morphology of helper T cells with cytokine activity through imaging flow cytometry and deep learning.

The immune system protects the body from foreign substances, such as viruses and bacteria. Among its key orchestrators are helper T cells, which coordinate immune responses by activating other immune cells to eliminate these threats. Helper T cell activation has been widely studied, often by stimulating cells in vitro and measuring changes in receptor expression. However, systematic assessment of activation at scale remains challenging because it requires both large single-cell imaging datasets and advanced computational methods to analyse complex morphological features. Here, using imaging flow cytometry and deep learning, we show that morphological information is crucial for identifying high-activity T cells. We also demonstrate that receptor clustering may associate with high cytokine activity. These findings provide a new perspective for assessing treatment options for patients with immune-related diseases.