Standard Magnetic Resonance Imaging (MRI) treats the brain as a collection of pixels, or "voxels." In a typical scan, each voxel represents a few cubic millimeters of tissue—a volume that contains millions of neurons, glial cells, and blood vessels. Traditional techniques like Diffusion Tensor Imaging (DTI) provide a single, averaged measurement of how water moves within that volume.
The problem is that this average is often misleading. If water diffusion appears slow, is it because the cells are packed tightly together, or because the fibers are fanning out in different directions? DTI cannot tell the difference. Multi-compartment biophysical modeling solves this by "unmixing" the signal, decomposing the total voxel measurement into distinct biological pools.
The Voxel as a Biological Soup
To understand why this works, we have to view tissue not as a uniform block, but as a collection of environments that restrict water in different ways. Researchers like Daniel C. Alexander and Hui Zhang pioneered the idea of representing these environments using geometric primitives:
- Sticks: Water trapped inside long, thin axons or dendrites (neurites). Movement is restricted to one dimension.
- Spheres: Water trapped inside cell bodies (somas). Movement is restricted in all directions.
- Balls: Water in the extracellular space, which moves more freely but is "hindered" by the surrounding cells.
By applying different magnetic gradients (b-values), we can observe how the signal decays. Water in "balls" (extracellular) disappears quickly as the gradient increases, while water in "sticks" (intracellular) persists. By fitting the signal to these mathematical shapes, models like CHARMED and NODDI can estimate the actual density of nerve fibers independently of their orientation Brain Microstructure assessed by CHARMED and NODDI in the newborn.
Untangling Orientation from Density
One of the most significant hurdles in neuroimaging is "orientation dispersion"—the fanning or bending of nerve fibers. In standard imaging, a region where fibers fan out looks identical to a region where fibers are simply dying off.
Hui Zhang and Daniel C. Alexander addressed this in Axon Diameter Mapping in the Presence of Orientation Dispersion with Diffusion MRI by incorporating a Watson distribution into the model. This allows the algorithm to account for the "spread" of the fibers first, then calculate the density of the axons within that spread. This distinction is vital for diseases like ALS, where Rebecca Broad and colleagues used NODDI to show that the primary driver of white matter change is actual axonal loss, not just a change in fiber organization Neurite orientation and dispersion density imaging (NODDI) detects cortical and corticospinal tract degeneration in ALS.
Seeing Through the Fluid
In many pathologies, such as stroke, Alzheimer’s, or brain tumors, the tissue is flooded with extracellular fluid (edema). This "free water" masks the underlying health of the neurons.
Ofer Pasternak developed a bi-tensor approach to "see through" this fluid. By modeling a separate "free-water" compartment, the algorithm can mathematically subtract the signal of the swelling, leaving behind a "tissue-corrected" metric that reflects the true integrity of the axons Neuronal Fiber Delineation in Area of Edema from Diffusion Weighted MRI. This has proven critical in Alzheimer’s research, where Fang Ji and Ofer Pasternak demonstrated that even in "pure" Alzheimer's cases, hidden vascular-related fluid increases contribute significantly to cognitive decline Distinct white matter microstructural abnormalities and extracellular water increases relate to cognitive impairment in Alzheimer’s disease with and without cerebrovascular disease.
Virtual Biopsies in Cancer
The same principles of signal decomposition are revolutionizing oncology. Standard MRI can find a lump in the prostate, but it struggle to tell if that lump is an aggressive cancer or a benign inflammation.
The VERDICT (Vascular, Extracellular and Restricted Diffusion for Cytometry in Tumours) framework, developed by researchers including Eleftheria Panagiotaki and Edward W. Johnston, uses a three-compartment model to estimate cell radius and intracellular volume fraction INNOVATE: A prospective cohort study combining serum and urinary biomarkers with novel diffusion-weighted magnetic resonance imaging for the prediction and characterization of prostate cancer. Because cancer cells are typically smaller and more tightly packed than healthy cells, VERDICT acts as a "virtual biopsy," quantifying epithelial cell density without a needle Validating the biological origin of in vivo fractional intracellular volume from VERDICT-MRI in the prostate.
Adding the Soma
While early models focused on the "wires" of the brain (axons), the "cell bodies" (somas) were often ignored. This led to inaccuracies in gray matter, where somas are abundant. Marco Palombo introduced SANDI (Soma And Neurite Density Imaging), which explicitly models somas as restricted spheres SANDI: a compartment-based model for non-invasive apparent soma and neurite imaging by diffusion MRI. By accounting for the signal coming from these spheres, SANDI allows for the non-invasive mapping of brain cytoarchitecture that matches traditional histological staining.
Why It Works: The Power of Joint Estimation
The most recent evolution in this field is the move toward multi-dimensional MRI. Instead of looking only at water diffusion, researchers like Marco Palombo and Paddy J. Slator are jointly modeling diffusion and relaxation (T1 and T2) Joint estimation of relaxation and diffusion tissue parameters for prostate cancer grading with relaxation-VERDICT MRI.
This works because different biological compartments have different "magnetic signatures." For example, blood in a vessel decays at a different rate than water inside a cell. By correlating these two physical properties, models can disentangle overlapping signals that would be impossible to separate using diffusion alone, providing a more granular and biologically specific view of human anatomy.