Sectional Neuroanatomy on MRI

Why neuroanatomy is a physics problem

Identifying a structure on MRI is not pattern recognition in the abstract; it is the joint solution of two problems. First, where am I in the head — which plane, which orientation convention, which slice level relative to a reproducible landmark. Second, what generates this voxel's signal — the proton density and relaxation times of the tissue, sampled by a particular pulse sequence. A confident read fuses geometry and contrast: the thalamus is not merely a gray oval in a certain location, it is a region whose $T_1$ and $T_2$ place it between white matter and CSF in a predictable way, bounded by the internal capsule whose dense myelin makes it brighter on $T_1$ and darker on $T_2$. This lesson develops both axes quantitatively, then sends you into real volumes to navigate them.

Imaging planes and orientation conventions

Brain MRI is acquired or reformatted in three orthogonal planes referenced to the anatomical axes. The axial (transverse) plane is perpendicular to the superior-inferior axis; the coronal plane is perpendicular to the anterior-posterior axis; the sagittal plane is perpendicular to the left-right axis. Modern 3D acquisitions sample an isotropic volume — typically $1\ \text{mm} \times 1\ \text{mm} \times 1\ \text{mm}$ voxels in a sequence such as MPRAGE — so any plane can be reconstructed after the fact without interpolation penalty. A 2D fast spin-echo stack, by contrast, is acquired in one prescribed plane with thicker slices (3 to 5 mm) and a slice gap, and reformats poorly.

Orientation convention is the single most common source of left-right error. In radiological convention, an axial image is displayed as if you are standing at the patient's feet looking toward the head: the patient's left is on the viewer's right. In neurological convention, used by many neurosurgeons and in much of the neuroscience literature, left is on the left. There is no signal difference between the two — only the display mapping changes — so the laterality of a finding depends entirely on which convention the workstation is using and on the image's left-right marker.

Standard space: MNI and Talairach

To compare brains across subjects, images are warped into a standard space with a defined origin and axes. The historical Talairach space sets its origin at the anterior commissure (AC) with the AC-PC line defining the y-axis, and uses a piecewise-linear scaling to a single post-mortem brain. The modern workhorse is the MNI space, built by averaging many MRI scans (the widely used MNI152 template is an average of 152 normal subjects). A point is reported as a triple $(x, y, z)$ in millimeters, with $x$ positive to the right, $y$ positive anterior, $z$ positive superior. The two systems are close but not identical; published approximate conversions exist, and you should always state which space a coordinate belongs to.

Tissue signal: the contrast coordinate

Every structure you name is identified partly by its intrinsic contrast behavior. The signal of a spin-echo voxel is governed by proton density $\rho$, longitudinal relaxation $T_1$, and transverse relaxation $T_2$, sampled through the repetition time $\text{TR}$ and echo time $\text{TE}$:

Choosing $\text{TR}$ and $\text{TE}$ selects which parameter dominates. $T_1$-weighting uses short TR (about 400 to 700 ms at 1.5 T) and short TE (about 10 to 15 ms): the $(1 - e^{-\text{TR}/T_1})$ term spreads tissues apart while the $e^{-\text{TE}/T_2}$ term stays near unity. $T_2$-weighting uses long TR (about 2500 to 4000 ms) and long TE (about 80 to 120 ms): the recovery term saturates near 1 for all tissues so differences in $T_2$ dominate. Proton-density (PD) weighting uses long TR and short TE so that both the recovery term $(1 - e^{-\text{TR}/T_1})$ and the decay term $e^{-\text{TE}/T_2}$ approach unity for every tissue: with both $T_1$ and $T_2$ effects suppressed, $\rho$ alone separates tissues — a subtle but powerful intermediate where white matter and gray matter differ by only a few percent in proton density yet CSF stands out.

Concretely, white matter has more lipid (myelin) and a shorter $T_1$ than gray matter, which is why it is bright on $T_1$. Gray matter has higher water content and a longer $T_2$, making it brighter than white matter on $T_2$. CSF is nearly free water: very long $T_1$ (so dark on $T_1$) and very long $T_2$ (so bright on $T_2$). Fat in the marrow and orbits is bright on both standard $T_1$ and fast-spin-echo $T_2$. The table below gives representative 1.5 T values.

| Tissue | T1 at 1.5 T (ms) | T2 (ms) | T1-weighted | T2-weighted | PD-weighted | | --- | --- | --- | --- | --- | --- | | White matter | 560 to 700 | 70 to 90 | Bright | Darker than GM | Intermediate | | Gray matter (cortex) | 900 to 1100 | 90 to 110 | Intermediate (darker than WM) | Brighter than WM | Slightly higher than WM | | CSF | 3000 to 4000 | 1500 to 2000 | Dark | Very bright | Intermediate to bright | | Fat (marrow, orbit) | 250 to 350 | 60 to 130 | Very bright | Bright (FSE) | Bright |

Supratentorial anatomy

The supratentorial compartment holds the cerebral hemispheres above the tentorium cerebelli. Each hemisphere is divided into frontal, parietal, temporal, and occipital lobes, plus the insula buried in the Sylvian fissure. Key landmarks: the central sulcus separates the precentral (motor) gyrus from the postcentral (sensory) gyrus and is located on axial images by the inverted-omega 'hand knob' of the precentral gyrus; the Sylvian (lateral) fissure separates the temporal lobe below from the frontal and parietal lobes above; the parieto-occipital sulcus and calcarine sulcus define the occipital lobe and primary visual cortex on sagittal and axial views.

Deep to the cortex lie the deep gray nuclei. The caudate nucleus runs along the lateral wall of the lateral ventricle (head, body, tail). The lentiform nucleus comprises the lateral putamen and medial globus pallidus. The thalamus is the large paired ovoid mass flanking the third ventricle. These nuclei are separated and connected by the internal capsule, a compact white-matter sheet with an anterior limb (between caudate head and lentiform nucleus), a genu, and a posterior limb (between thalamus and lentiform nucleus) that carries the corticospinal tract. Because the internal capsule is densely myelinated, it is conspicuously bright on $T_1$ and dark on $T_2$ relative to the adjacent gray nuclei — a contrast cue that lets you parse the basal ganglia even when borders are subtle.

The corpus callosum is the great commissure connecting the hemispheres, seen best on midline sagittal images as a C-shaped band: rostrum, genu, body, and splenium. It is heavily myelinated and therefore bright on $T_1$. The ventricular system — paired lateral ventricles, the midline third ventricle, the cerebral aqueduct, and the fourth ventricle — contains CSF and is your built-in contrast reference: follow it to read the sequence and to assess for mass effect, hydrocephalus, or atrophy.

Infratentorial anatomy and the basal cisterns

Below the tentorium lie the brainstem and cerebellum. The brainstem has three levels read on axial and sagittal images. The midbrain shows the paired cerebral peduncles anteriorly, the tegmentum, the colliculi posteriorly, and the cerebral aqueduct; on axial section it has a characteristic 'Mickey Mouse' silhouette. The pons bulges ventrally and connects to the cerebellum through the middle cerebellar peduncles. The medulla oblongata tapers to the cervicomedullary junction at the foramen magnum. The cerebellum consists of two hemispheres and a midline vermis, with a finely foliated cortex; its deep nuclei and peduncles are dense white matter.

Surrounding the brainstem are the basal cisterns, CSF-filled subarachnoid spaces that are bright on $T_2$ and outline the surface anatomy and the basal vessels and cranial nerves. Key examples: the suprasellar cistern above the pituitary (star-shaped on axial, containing the optic chiasm and circle of Willis), the interpeduncular and ambient cisterns around the midbrain, the prepontine cistern anterior to the pons, and the quadrigeminal cistern behind the colliculi. Effacement of these cisterns is an early and important sign of raised intracranial pressure or herniation.

White-matter tracts and the appearance of myelin

White matter is organized into three tract classes: association fibers (connecting regions within a hemisphere, for example the superior longitudinal fasciculus and the arcuate fasciculus), commissural fibers (connecting the two hemispheres, chiefly the corpus callosum and the anterior commissure), and projection fibers (connecting cortex to deep structures and cord, for example the corticospinal tract through the internal capsule and cerebral peduncle). On conventional MRI these tracts share the generic white-matter signature: bright on $T_1$, relatively dark on $T_2$, owing to the lipid bilayers of myelin and the resulting restricted water environment that shortens $T_2$ and $T_1$.

Myelin content also drives the developmental appearance. In the unmyelinated neonatal brain the gray-white contrast is essentially reversed relative to the adult: unmyelinated white matter has long $T_1$ and long $T_2$, so it is darker than gray matter on $T_1$ and brighter on $T_2$. Myelination proceeds caudal to rostral and central to peripheral over roughly the first two years, and the adult contrast pattern is established by about 18 to 24 months. Knowing the expected milestone protects against calling normal immature myelin a white-matter disease.

Guided navigation in real volumes

Now apply both coordinates to actual data. The viewer below loads four volumes: a $T_1$-weighted, a $T_2$-weighted, a proton-density dataset, and the MNI152 template for standard-space reference. Switch sequences on the same slice to watch tissues invert, and switch planes to triangulate a structure in three dimensions. Work through the checklist that follows the widget.

1. Read the sequence from CSF. On an axial slice through the lateral ventricles, note the ventricle signal: dark equals $T_1$, very bright equals $T_2$, intermediate with low overall gray-white contrast equals PD. Toggle all three and confirm the prediction from Eq. 16.2.
2. Find the central sulcus. On a high axial slice, locate the inverted-omega hand knob of the precentral gyrus; the sulcus just behind it is the central sulcus. Switch to sagittal to trace it from vertex toward the Sylvian fissure.
3. Parse the basal ganglia. On an axial slice through the third ventricle, identify caudate head, putamen, globus pallidus, and thalamus. Use the internal capsule as the divider — note it is bright on $T_1$ and dark on $T_2$, confirming dense myelin.
4. Trace the corpus callosum. Switch to midline sagittal and identify rostrum, genu, body, and splenium; confirm it is bright on $T_1$.
5. Step down the brainstem. On sagittal midline, identify midbrain, pons, and medulla; then go axial and recognize the midbrain 'Mickey Mouse' silhouette and the prepontine cistern.
6. Register to standard space. Open the MNI152 template, pick a deep structure such as the thalamus, and read its approximate $(x, y, z)$ MNI coordinate; appreciate that the template is an average, so individual sulci are blurred relative to the single-subject T1.

Putting the two coordinates together

A robust read is a short algorithm. Establish the plane and confirm the orientation convention from the L/R marker. Read the sequence off the CSF signal. Then localize structures by their joint geometric-and-contrast signature: deep gray nuclei separated by a $T_1$-bright, $T_2$-dark internal capsule; a $T_1$-bright corpus callosum on midline sagittal; CSF-bright basal cisterns hugging a three-tiered brainstem. When a finding's laterality or location matters clinically — a lacune in the posterior limb, an effaced suprasellar cistern, an asymmetric thalamus — you can defend it because you know both where you are and why the tissue looks the way it does.