A Comprehensive Protocol for Manual Segmentation of the Medial Temporal Lobe Structures

The overall goal of this procedure is to illustrate how to manually trace the human medial temporal lobe structures on high resolution magnetic resonance images. This is accomplished by first identifying the amygdala in the brain and tracing it systematically from the anterior to the posterior borders. The second step is to locate the longitudinal extent and the general borders of the hippocampus, and then trace it slice by slice in the anterior posterior direction.

Next, trace the anterior structures of the para hippocampal gyrus, namely the interal and perineal cortices in parallel to better accommodate variations in their mutual borders. The final step is to trace the posterior segment of the para hippocampal gyrus, namely the para hippocampal cortex proper in the same anterior posterior fashion. Ultimately, this manual tracing procedure can be used to obtain a 3D rendering and volumetric statistics of all medial temporal lobe structures.

The main advantage of this tracing protocol over existing protocols, as well as automatic segmentation algorithms, is that it identifies all of the regions of the medial temporal lobe relative to their neighboring structures. This can increase localization accuracy and manual tracing, as well as help in evaluating and correcting if necessary, the automatic segmentation results in a region of the brain that can have considerable variability. The current protocol is used for not only for anatomical investigations in medial temporal lobe, but can also be applied to the data analysis from other imaging modalities.

For example, the functional magnetic resonance imaging to identify functional regions of interest with high accuracy. This protocol provides unique illustration of the intertemporal lobe regions, which can help both advance and novice researchers to accurately identify localize these regions in a human brain. The first demonstration covers tracing of the amygdala.

Begin by localizing the structure in the brain. Identify the first slice of the amygdala in which the Lyman insula appears. That is where the white matter connection between the frontal and temporal lobes is continuous.

The optic chiasm is another important landmark. Then use the angular bundle as the inferolateral border of the amygdala to identify the last slice of the amygdala in the coronal view. Look for a level where the structure is superior to the medial extension of the temporal horn of the lateral ventricle and lateral to the uncinate gyrus, the protuberance of the head of the hippocampus.

Now to trace the amygdala, identify the first slice in which the anterior commissure is continuous throughout both hemispheres where the amygdala is visible in its typical shape, trace the amygdala using the interal sulcus as the super medial border, the imaginary line from the fundus of the semi-annual sulcus along the white matter to the inferior tip of the amygdala as the infra medial border, the temporal stem as the lateral border and follow the superolateral curvature to complete the tracing on posterior slices, draw an imaginary line from the level of the fundus of the inferior circular sulcus of the insula along the semi-annual gyrus to the optic tract as the superior border of the amygdala. Be sure to exclude the uncinate gyrus. Use the alius and the temporal horn of the lateral ventricle for the inferolateral delineation and the temporal stem for the lateral border.

Follow the curvature super laterally to complete the tracing. Pay special attention to distinguish the amygdala in its early slices from the surrounding areas, onus and enteral cortex by using sagittal and axial views as guides, then systematically perform a slice by slice tracing of the amygdala in an anterior to posterior direction. Using these guidelines, the next structure in the medial temporal lobe to be localized is the hippocampus begin tracing the hippocampus when the temporal horn of the lateral ventricle appears along the inferolateral border of the amygdala and the tracing of the hippocampus with its last appearance, infra medial to the trigone of the lateral ventricle always utilize the alternative viewing orientations to help localize the hippocampus and its borders.

To define the borders of the hippocampus begin by delineating the alius and the FIA at the superior aspect. The temporal horn at the lateral aspect inferiorly trace along the angular bundle to define the inferior border. Continue tracing the hippocampus using the same definitions throughout.

Additionally include the ulu into the segmentation such that it immediately borders the white matter bundle of the para hippocampal gyrus. Superiorly aligns with the curve of the onus and primarily extends horizontally from the hippocampus. Pays special attention when tracing the following three structures.

First, include the posterior onus in the segmentation. Second, be sure to omit the choroid plexus above the alius from the segmentation, although this may not be possible on lower resolution images. And third, refer to the two alternative views to avoid the inclusion of the tail of the caudate nucleus.

The next area to be traced is the para hippocampal gyrus. A term being used to include the perineal cortex on the lateral side. The trinal cortex on the medial side and the para hippocampal cortex proper in the anterior slices define the first slice of the perineal cortex with the appearance of the collateral sulcus before the onset of the trinal cortex trace the perineal cortex from the medial edge of the lateral bank of the collateral sulcus to the lateral fundus of the gyrus of albe or that of the medial one.

If two gyri of albe are present or the midpoint of the dorsal temporal polar surface. In the absence of this gyrus start to trace the interal cortex about five millimeters anterior to the Lyman insula, using the fundus of the medial temporal polar sulcus as the superior end and the peel surface as the infra medial border. After the amygdala appears, the fundus of the semi annular sulcus becomes the superior border, or the point where the imaginary extension of the angular bundle meets the ventricular cavity.

If the semi annular sulcus is indistinguishable, the border between the perineal and the trinal cortex may vary from slice to slice. When the collateral sulcus is deep, that is greater or equal to 1.5 centimeter trace the perineal cortex from the medial edge of the medial bank of this sulcus to the midpoint of its lateral bank. In cases with a regular collateral sulcus that is with the depth ranging from one to 1.5 centimeters, trace the perineal cortex as the area from the midpoint of the medial bank of the collateral sulcus to the medial end of the lateral bank of the sulcus.

In case of a shallow collateral sulcus that is less than one centimeter trace the perineal cortex from the fundus of this sulcus to the midpoint of the crown of the fusiform gyrus. When the collateral sulcus is interrupted, usually at the level of the uncle apex by a small gyrus emerging from its fundus, trace the perineal cortex to the fundus of the lateral sulcus. Trace the enteral cortex until about 1.5 millimeters posterior to the oncle apex.

Extend the tracing of the perineal cortex medially to occupy the place of the enteral cortex after its termination, where the definitions for the ladder continue to apply over a length of three millimeters. Beyond this, the perineal cortex is substituted by the para hippocampal cortex. Proper start to trace the para hippocampal cortex on the slice immediately posterior to the end of the perineal cortex and end tracing about four millimeters posterior to the end of the hippocampal tail.

Trace the para hippocampal cortex proper in the same way described for the posterior portion of the perineal cortex until the calcine sulcus appears restricting it to the inferior edge of the sulcus. Should a mini sulcus appear at around this level, include it in the segmentation. A result of tracing the MTL structures is shown here in the 3D rendering and in 2D images showing the location of representative slices in the brain on the right side.

These coronal images of the amygdala show an extreme example of possible differences between results of manual tracing and automatic segmentation. As seen in the right side panel, a common automatic segmentation software has only identified a small portion of the left amygdala while neglecting more than half of the tissue that is identifiable as part of the amygdala to an expert human eye. Similar underestimation, but to a lesser extent also occurred in the right amygdala.

Shown here are representative volumetric results of the bilateral amygdala and the hippo campi from a single subject yielded by manual tracing. Using the present protocol and automatic segmentation, it can be observed that automatic segmentation has misestimated the volume of each of the four structures compared. Following this procedure volume statistics can be collected in order to answer questions such as whether medial temporal lobe structures show volumetric differences between two age groups or between healthy and clinical groups While attempting this procedure.

In order to ensure accuracy and efficiency while tracing it is important to refer to each of the three visual planes, the coronal, axial, and sagittal. After watching this video, you should have a good understanding of how to accurately identify and manually trace or adjust automatic segmentations of the MIT temporal lobe regions based on knowing the anatomical borders of the structures.