Abstract
<title>Abstract</title> <p>Mitochondria power brain function and cognition, yet no label-free, non-invasive method has existed to explore their relationship to ageing, disease, and cognition in humans. The MitoBrainMap framework predicts mitochondrial features from magnetic resonance data alone, potentially bridging cellular biology with macroscale brain organization. Here we tested whether it captures meaningful age- and disease-related variation across individuals. MR-predicted mitochondrial density and tissue respiratory capacity declined with age, whereas intrinsic mitochondrial respiratory capacity was relatively preserved. Correlations among predicted features matched known mitochondrial biology, supporting preliminary construct validity. In patients with genetically confirmed mitochondrial diseases, predicted maps revealed region-specific alterations, notably the expected compensatory upregulation of nuclear- encoded complex II. Predicted features were further associated with the energetic stress marker GDF15 and with cognitive performance, linking brain mitochondrial estimates to systemic physiology and behavior. These findings introduce a first-generation, label-free neuroimaging-based mitochondrial mapping as a non-invasive window into living human brain mitochondria.</p>