Improved Assessment of Ex Vivo Brainstem Neuroanatomy With High-Resolution MRI and DTI at 7 Tesla

Authors

  • Guadalupe Soria,

    Corresponding author
    1. Department of Brain Ischemia and Neurodegeneration, Institut d'Investigacions Biomèdiques de Barcelona (IIBB)-Consejo Superior de Investigaciones Científicas (CSIC), Institut d'Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Rosselló 162, Barcelona 08036, Spain
    2. Experimental MRI 7T Unit, Institut d'Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Villarroel 170, Barcelona 08036, Spain
    • Institut d'Investigacions Biomèdiques de Barcelona (IIBB), Consejo Superior de Investigaciones Científicas (CSIC), Institut d'Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS)
    Search for more papers by this author
    • Fax: +34-93-3638301

  • Matteo De Notaris,

    1. Laboratory of Surgical Neuroanatomy (LSNA), Facultat de Medicina, Universitat de Barcelona, Casanova 143, Barcelona 08036, Spain
    2. Department of Neurological Sciences, Division of Neurosurgery, Università degli Studi di Napoli Federico II, Via Sergio Pansini 5, Naples 80131, Italy
    Search for more papers by this author
  • Raúl Tudela,

    1. Experimental MRI 7T Unit, Institut d'Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Villarroel 170, Barcelona 08036, Spain
    2. CIBER de Bioingeniería, Biomateriales y Nanomedicina (CIBER-BBN), Group of Biomedical Imaging of the University of Barcelona, Casanova 143, Barcelona 08036, Spain
    Search for more papers by this author
  • Gerard Blasco,

    1. IDI, Radiology Department, Hospital Universitario Dr. Josep Trueta. IDIBGI, Universitat de Girona, Av. Francia s/n, Girona 17007, Spain
    Search for more papers by this author
  • Josep Puig,

    1. IDI, Radiology Department, Hospital Universitario Dr. Josep Trueta. IDIBGI, Universitat de Girona, Av. Francia s/n, Girona 17007, Spain
    Search for more papers by this author
  • Anna M. Planas,

    1. Department of Brain Ischemia and Neurodegeneration, Institut d'Investigacions Biomèdiques de Barcelona (IIBB)-Consejo Superior de Investigaciones Científicas (CSIC), Institut d'Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Rosselló 162, Barcelona 08036, Spain
    2. Experimental MRI 7T Unit, Institut d'Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Villarroel 170, Barcelona 08036, Spain
    Search for more papers by this author
  • Salvador Pedraza,

    1. IDI, Radiology Department, Hospital Universitario Dr. Josep Trueta. IDIBGI, Universitat de Girona, Av. Francia s/n, Girona 17007, Spain
    2. Programa de Doctorado del Departament de Medicina de la Universitat Autònoma de Barcelona, Passeig Vall d'Hebron, 119, Barcelona 08035, Spain
    Search for more papers by this author
  • Alberto Prats-Galino

    1. Laboratory of Surgical Neuroanatomy (LSNA), Facultat de Medicina, Universitat de Barcelona, Casanova 143, Barcelona 08036, Spain
    Search for more papers by this author

Abstract

The aim of the present work was to provide the topography of the main gray nuclei and white matter tracts of the human brainstem at 7 Tesla (7 T) high-field magnetic resonance imaging (MRI) using structural imaging (T1) and diffusion tensor imaging (DTI). Both imaging techniques represent a new field of increasing interest for its potential neuroanatomic and neuropathologic value. Brainstems were obtained postmortem from human donors, fixated by intracarotid perfusion of 10% neutral buffered formalin, and scanned in a Bruker BioSpec 7 T horizontal scanner. 3D-data sets were acquired using the modified driven equilibrium Fourier transform (MDEFT) sequence and Spin Echo-DTI (SE-DTI) sequence was used for DTI acquisition. High-resolution structural MRI and DTI of the human brainstem acquired postmortem reveals its basic cyto- and myeloar-chitectonic organization, only visualized to this moment by histological techniques and higher magnetic field strengths. Brainstem structures that are usually not observed with lower magnetic fields were now topographically identified at midbrain, pons, and medullar levels. The application of high-resolution structural MRI will contribute to precisely determine the extension and topography of brain lesions. Indeed, the current findings will be useful to interpret future high-resolution in vivo MRI studies in living humans. Anat Rec, 2011. © 2011 Wiley-Liss, Inc.

The brainstem represents an essential structure for the neurotransmission of the whole central nervous system (CNS). It contains ascending sensory pathways which transfer information from the peripheral nervous system and the spinal cord to higher levels. Motor descending pathways carrying control commands from the brain to the spinal cord convey in the brainstem, supplying the peripheral nervous system, the muscles, and organs of the body. In addition, the brainstem is responsible for key vital functions (i.e., control of blood pressure, cardiac frequency, respiration), as well as consciousness.

The English translation of Ramón y Cajal's work (1909) remind us that the bases of our actual understanding of the neuroanatomy of the brainstem were established at the beginning of the last century. Later in 1949, Olszewski and Baxter's cytoarchitectonic studies introduced key advances in the structural knowledge of the brainstem, which were followed by the application of histochemical techniques based on chemical reactions to demonstrate the presence of acetylcholinesterase (Paxinos and Huang, 1995), and by immunohistochemical techniques for monoamines and other neurotransmitters localization (Blessing and Gai, 1997). Altogether, these studies have been particularly useful for interpreting Magnetic Resonance Imaging (MRI) of the brainstem, since the level of resolution reached by the latter has been far from that obtained by histological techniques. Nevertheless, the rapid development of MRI techniques in the last 3 decades has represented a major advance in the study of the brainstem, whose radiological structure was systematically described at 0.35 Tesla (T) (Flannigan et al., 1985), 1 T (Bradley, 1991), 1.5 T (Tamraz and Comair, 2000), and 3 T (Nagae-Poetscher et al., 2004). Mainly applied to the investigation of cortical systems, several studies have demonstrated that high-resolution MRI (≥3 T) is directly comparable with postmortem histological staining methods (Walters et al., 2003; Eickhoff et al., 2005; Bridge and Clare, 2006). Although the capacity for discriminating centers (nuclei) and nerve tracts within the brainstem has gradually increased with higher magnetic fields, it is still not comparable to histological techniques and can lead to misinterpretations in the clinical practice.

Neuronal tracer techniques such as transport of different molecules (HRP-WGA, fluorochromes, PHAL) or transynaptic viral agents can not be used in humans. Thus, the investigation of the complex connectivity within the human brainstem has suffered from the limitations of these invasive techniques. Despite the artifacts that appear in fixed material, Nauta and Gygax (1951) techniques represent the gold standard method for the study of human brain connections. In addition, although relatively few studies have been conducted with silver staining methods of axonal degeneration, they have provided important information about the course and termination of the main ascending (Marani and Schoen, 2005) and descending (Schoen, 1969; see also for review Kuypers, 1981) tracts that cross the brainstem, as well as the connections of the cerebellum and the precerebellar nuclei (Voogd, 2004). In this line, the use of Diffusion Tensor Imaging (DTI) techniques (Basser et al., 1994), based on the direction of water diffusion, is opening new perspectives in the analysis of the human brainstem connectivity. Although the resolution obtained is still limited, its application has allowed to characterize the topographic organization of the corticospinal tract (Wakana et al., 2004; Chen et al., 2007; Habas and Cabanis, 2007; Hong et al., 2010), the corticopontine fibers, transverse fibers of the pons and cerebellar peduncles (Wakana et al., 2004; Habas and Cabanis, 2007), the medial lemniscus (Wakana et al., 2004; Chen et al., 2007; Habas and Cabanis, 2007), the trigeminal system (Upadhyay et al., 2008), and tentatively, the medial longitudinal fasciculus (Habas and Cabanis, 2007).

The increase in availability of higher magnetic fields has led to several high-resolution DTI studies performed in the ex vivo human brain (Lanyon et al., 2009; McNab et al., 2009; Naidich et al., 2009). Interestingly, studies that investigate neuronal diseases in living patients by using ultra-high magnetic fields, such as 7 T magnets, are increasing their frequency (Hammond et al., 2008a, b; Marqués et al., 2010; Metcalf et al., 2010; Zhang et al., 2010). With this in view, high-resolution MRI and DTI techniques might become a standard method for detecting pathological changes in the brain at the earliest possible time (Bridge and Clare, 2006; Vaillancourt et al., 2009), and 7 T magnets will most likely be the next generation of MRI systems, not only for human research purposes, but for early diagnosis of CNS diseases in the clinical practice.

Hence, an exhaustive depiction of the structural neuroanatomy of the brainstem that can be described at 7 T MRI is necessary to develop new in vivo acquisition techniques reaching similar histological resolution in a suitable acquisition time. For multimodal imaging purposes it is important to compare the appearance of the same brainstem structures at both DTI and T1 weighted. Thus, the aim of this work was to provide an accessible and detailed topographic map of the main gray matter (GM) nuclei and white matter (WM) tracts of the human brainstem with postmortem high-resolution MRI and DTI, simultaneously acquired in the same specimen at 7 T magnetic field strength.

MATERIALS AND METHODS

Tissue Processing

Brainstems were obtained from donor dead people (N = 3) with no history of neurological diseases. In all cases, informed consent was obtained for using the tissues with research purposes at the Donor Service, Unit of Human Anatomy and Embryology, Facultat de Medi-cina, Universitat de Barcelona. Brain fixation was performed by intracarotid and vertebral injection of 10% neutral buffered formalin by using a perfusion pump. On the following day, the brain was dissected under operating microscoping control (Zeiss OPMI 16 and Contraves, Oberkochen, Germany). Cranial nerves were sectioned at the points of perforation of the dura mater. Next, the brainstem was obtained as one whole piece including tissue from the level of the optic chiasm to the first cervical segment of the spinal cord. The cerebellum was removed after sectioning the cerebellar peduncles. The specimens were then immersed in the same fixing solution during two additional months. For the MRI acquisition, samples were introduced in a plastic tube, filled with the same fixing solution, which had the following dimensions: Ø 35 mm, length 125 mm.

MRI

MRI experiments were conducted on a 7.0 T BioSpec 70/30 horizontal scanner (Bruker BioSpin, Ettlingen, Germany), equipped with a 12 cm inner diameter actively shielded gradient system (400 mT/m). A transmit/receive quadrature volume coil was used for image acquisition. Specimens were placed horizontally in a Plexiglas holder. Tripilot scans were used for accurate positioning of the specimens inside the magnet. The sample was positioned to have the center of the brainstem approximately located at the magnet's isocenter. In general, the parameters were modified to adapt the usual protocol used in clinical settings to this high-resolution study performed in a Bruker system dedicated to small animal imaging. In particular, three-dimensional (3D) T1 anatomical images were acquired using the Modified Driven Equilibrium Fourier Transformation (MDEFT) sequence with inverse preparation (also known as magnetization prepared rapid gradient echo or MPRAGE). This sequence with mixed T1 and FLAIR characteristics was chosen since it has been demonstrated to achieve images with high signal to noise ratio, enhanced GM/WM contrast, and high resolution in the shortest acquisition time (Kochunov et al., 2010). The acquisition was performed with the following parameters: repetition time (TR) = 4,000 msec, echo time (TE) = 3.5 msec, inversion time = 1,100 msec, eight segments, two averages, field of view (FOV) = 38.4 × 38.4 × 67.2 mm3, matrix size = 256 × 256 × 112, with a resulting voxel resolution of 0.150 × 0.150 × 0.600 mm3. The acquisition time was 1 hr and 59 min.

For DTI images, taking into account the well known artifacts associated to Echo Planar Imaging (susceptibility-induced geometric distortions, eddy current effects, etc.), we decided to use the Spin Echo DTI sequence to gain image quality to the detriment of short acquisition time (Ardekani and Sinha, 2005). Thus DTI acquisition was performed using a Spin Echo DTI sequence with TR = 250 sec, TE = 26 sec. Diffusion sensitizing gradients were applied along 30 directions with b value = 1,000 sec/mm2. FOV = 50 × 40 × 67.2 mm3, matrix size = 125 × 100 × 112, with a resulting spatial resolution of 0.4 × 0.4 × 0.6 mm, with no gaps between slices. The acquisition time was 8 hr and 40 min. Overall, MRI studies were performed overnight with a total acquisition time of about 10 hr for specimen.

MRI Data Analysis and Processing

T1 weighted images were directly reconstructed at the scan workstation by using Paravision 5.0 (Bruker Biospin, Ettlingen, Germany) and subsequently imported to Amira 4.2 software (Visage Imaging, CA) for visualization and processing. A surface rendering of the brainstem specimen was performed to show the selected levels of anatomical description (Fig. 1). Except for some bright/dark contrast, no smoothing or other image postprocessing techniques were performed in this image modality. For DTI images, brainstem diffusion weighted images acquired under 30 different gradient directions, plus one b0 image, were reconstructed at the scan workstation by using Paravision 5.0 software (Bruker Biospin, Ettlingen, Germany) and subsequently imported to DTIWeb software (Prados et al., 2007; available at http://trueta.udg.edu/DTI/) to calculate fractional anisotropy maps and the corresponding directional color coded red-green-blue (RGB) maps. After that, the RGB-maps were visualized and processed by using Amira 4.2 software (Visage Imaging, CA). The midline of the brainstem was used to separate left and right hemispheres and for presentation in this work the left side of both T1 weighted and DTI images were selected and are shown in Figs. 211. Full images are available in Supporting Information. To identify the anatomical structures, two observers with high experience in the human anatomy analyzed and compared the T1 and DTI images with published histological and MRI atlases (Tamraz and Comair, 2000; Haines, 2004; Mori et al., 2005; Naidich et al., 2009).

Figure 1.

Surface reconstruction from a representative brainstem showing the levels selected for the posterior description.

Figure 2.

Pyramidal decussation. The left part of the brainstem is presented in this image where the left side corresponds to T1 weighted images and the right side corresponds to DTI RGB maps (In red: medio-lateral oriented fibers, in green: antero-posterior fibers, and in blue: superior-inferior fibers). CuN: cuneatus nucleus, CuT: cuneatus tract, GrN: gracile nucleus, GrT: gracile tract, spXIN: spinal nucleus of the accessory nerve, spVN: spinal trigeminal nucleus, spVT: spinal trigeminal tract, STT: spinothalamic tract, vSCT: ventral spinocerebellar tract, xPy: pyramidal decussation.

RESULTS

Brainstem structures usually not observed at low field in vivo MRI T1 and DTI images were now topographically identified at different midbrain, pons, and medullar levels, as shown in Fig. 1. T1 weighted images acquired with the MDEFT sequence showed nuclei appearing as gray structures brighter than tracts. The RGB-maps show directionality of fibers by applying red color to medio-lateral oriented fibers, green color to antero-posterior fibers, and blue to superior-inferior fibers, following conventional coding. Both T1 weighted and DTI images were derived from exactly the same specimen. For better correlation and identification of structures both types of images are represented in each figure, where the two halves correspond to the left side of the brainstem. A few imaging artifacts were evident at some of the shown levels. In T1 weighted images, intensity inhomogeneities were observed in Figs. 6–8 in the posterior part of the middle cerebellar peduncle (MCP). In DTI images, there was a bigger partial volume effect which might be a consequence of the lower in-plane resolution used in comparison to the T1 weighted images. These effects can be well noticed along the midline borders, where both images do not perfectly match. Moreover, the partial volume effect is higher in areas containing population of WM fibers with different orientations like the middle cerebellar peduncle (MCP). This might account to explain the inhomogeneities observed in the green intensity of this structure in Fig. 7, as well as in the boundaries between GM and WM such as the transverse fibers of the pons (Fig. 9).

The different nuclei and tracts have been identified in this section following, first, a functional criterion and, second, a caudocranial order. An exception was made with the descending motor systems which, due to their functional distribution, were described craniocaudally. To simplify the recognition of the structures in the figures and the comprehension of the description that follows, both the complete anatomical names and their corresponding abbreviations have been used. Complete T1 weighted images and RGB-maps are available as Supporting Information.

Motor and Sensory Structures Associated to Cranial Nerves

At caudal levels of the medulla, the spinal nucleus of the accessory nerve (spXIN), that innervates neck muscles, aligns ventrolateral to the pyramidal decussation (xPy) (Fig. 2). More dorsally, the hypoglossal nucleus (XIIN), the main tongue muscle supplying center through the XII cranial nerve (CN) (XIIN), was identified. It has an oval appearance, and is located in the central GM surrounding the central canal at caudal levels of the medulla (Fig. 3) whereas at cranial levels remains in contact with the floor of the fourth ventricle (Fig. 4). The vagal complex, comprised by the parasympathetic dorsal motor nucleus of the vagus (dMNX) and the sensory visceral solitary nucleus (SolN), is located dorsal to the hypoglossal nucleus within the central GM whereas at cranial levels of the medulla they dispose laterally to it (Fig. 4). The solitary nucleus (SolN) runs parallel to the solitary tract (SolT) which is clearly identifiable by its lower signal intensity and its oval shape (Fig. 4). The spinal trigeminal nucleus (spVN) (Figs. 2–5) occupies the dorsolateral region of the medulla, receiving descending sensory fibers from the V CN, which constitutes the spinal trigeminal tract (spVT) always lying closer to the surface of the medulla. The vestibular complex (VeN), associated to the VIII CN, is situated between the spinal trigeminal nucleus and the vagal complex at the cranial portion of the medulla (Fig. 5). The first relay of the auditory pathway is located in the ventral and dorsal cochlear nuclei (vCoN and dCoN). The ventral co-chlear nucleus is identified laterally at the pontomedullary junction level together with the entry of the VIII CN, whereas the dorsal cochlear nucleus is located in the most dorsolateral parts of the medulla (Fig. 5).

Figure 3.

Inferior medulla. The left part of the brainstem is presented in this image where the left side corresponds to T1 weighted images and the right side corresponds to DTI RGB maps (In red: medio-lateral oriented fibers, in green: antero-posterior fibers, and in blue: superior-inferior fibers). Arcf: arcuate fibers, CST: corticospinal tract, CuN: cuneatus nucleus, CuT: cuneatus tract, GrN: gracile nucleus, GrT: gracile tract, ML: medial lemniscus, MLF: medial longitudinal fasciculus, spVN: spinal trigeminal nucleus, spVT: spinal trigeminal tract, STT: spinothalamic tract, vSCT: ventral spinocerebellar tract, XIIN: hypoglossal nucleus.

Figure 4.

Medium medulla. The left part of the brainstem is presented in this image where the left side corresponds to T1 weighted images and the right side corresponds to DTI RGB maps (In red: medio-lateral oriented fibers, in green: antero-posterior fibers, and in blue: superior-inferior fibers). AmO: amiculum of olive, CST: corticospinal tract, dAON: dorsal accessory olivary nucleus, dMNX: dorsal motor nucleus of the vagus, hION: hillium of ION, ICP: inferior cerebellar peduncle, ION: inferior olivary nucleus, mAON: medial accessory olivary nucleus, ML: medial lemniscus, MLF: medial longitudinal fasciculus, SolN&T: solitary nucleus and tract, spVN: spinal trigeminal nucleus, spVT: spinal trigeminal tract, STT: spinothalamic tract, VeN: vestibular nuclei, vSCT: ventral spinocerebellar tract, XIIN: hypoglossal nucleus.

Figure 5.

Superior medulla. The left part of the brainstem is presented in this image where the left side corresponds to T1 weighted images and the right side corresponds to DTI RGB maps (In red: medio-lateral oriented fibers, in green: antero-posterior fibers, and in blue: superior-inferior fibers). AmO: amiculum of olive, CST: corticospinal tract, dCoN: dorsal cochlear nucleus, hION: hillium of ION, ICP: inferior cerebellar peduncle, ION: inferior olivary nucleus, ML: medial lemniscus, MLF: medial longitudinal fasciculus, OCf: olivocerebellar fibers, spVN: spinal trigeminal nucleus, spVT: spinal trigeminal tract, STT: spinothalamic tract, vCoN: ventral cochlear nucleus, VeN: vestibular nuclei.

The abducens nucleus (VIN), the motor nucleus of the VI CN, is located at the caudal pons level (Fig. 6), near the floor of the fourth ventricle, with its intramedullary roots coursing ventrocaudally. This nucleus is surrounded by the genu of the facial nerve (gVII) identified by its round shape and hypointensity near the midline (Fig. 6). The descending root of the facial nerve (drVII) comes from the genu and courses ventrolaterally, medial to the spinal trigeminal nucleus (spVN) and lateral to the facial nucleus (VIIN) where it originates. The spinal trigeminal nucleus (spVN), at more cranial sections, is substituted by the principal sensory and the medially located motor trigeminal nuclei (p&mVN) (Fig. 7). Both are partially separated by the root of the trigeminal nerve. Moreover, in its ascending path toward the midbrain, the mesencephalic tract of the trigeminal nerve (mesVT), carrying propiocetive information from the jaw, is initially identified in a dorsolateral position (Fig. 7), although it is progressively more visible at rostral levels in the limit of the pontine GM (Fig. 8).

Figure 6.

Inferior pons. The left part of the brainstem is presented in this image where the left side corresponds to T1 weighted images and the right side corresponds to DTI RGB maps (In red: medio-lateral oriented fibers, in green: antero-posterior fibers, and in blue: superior-inferior fibers). CST: corticospinal tract, CTT: central tegmental tract, drVII: descending root of the facial nerve (VII CN), gVII: genu of the facial nerve (VII CN), LL: lateral lemniscus, MCP: middle cerebellar peduncle, ML: medial lemniscus, MLF: medial longitudinal fasciculus, SON: superior olivary nucleus, spVN: spinal trigeminal nucleus, spVT: spinal trigeminal tract, STT: spinothalamic tract, TPf: transverse pontine fibers, VI: abducens nerve, VIN: abducens nucleus (VI CN), VIIN: facial nucleus (VII CN).

Figure 7.

Medium pons. The left part of the brainstem is presented in this image where the left side corresponds to T1 weighted images and the right side corresponds to DTI RGB maps (In red: medio-lateral oriented fibers, in green: antero-posterior fibers, and in blue: superior-inferior fibers). CST: corticospinal tract, CTT: central tegmental tract, FPT: frontopontine tract, LL: lateral lemniscus, MCP: middle cerebellar peduncle, mesVT: mesencephalic tract of the trigeminal nerve, ML: medial lemniscus, MLF: medial longitudinal fasciculus, OPTPT: occipito-parieto-temporopontine tract, p&mVN: principal sensory and motor trigeminal nuclei, PN: pontine nuclei, SCP: superior cerebellar peduncle, STT: spinothalamic tract, TPf: transverse pontine fibers, V: trigeminal nerve (V CN).

Figure 8.

Superior pons. The left part of the brainstem is presented in this image where the left side corresponds to T1 weighted images and the right side corresponds to DTI RGB maps (In red: medio-lateral oriented fibers, in green: antero-posterior fibers, and in blue: superior-inferior fibers). CST: corticospinal tract, CTT: central tegmental tract, FPT: frontopontine tract, LL: lateral lemniscus, MCP: middle cerebellar peduncle, mesVT: mesencephalic tract of the trigeminal nerve, ML: medial lemniscus, MLF: medial longitudinal fasciculus, OPTPT: occipito-parieto-temporopontine tract, PN: pontine nuclei, SCP: superior cerebellar peduncle, STT: spinothalamic tract, TPf: transverse pontine fibers.

In the midbrain (Figs. 10, 11), the trochlear nucleus (IV CN), seen as a brighter structure, lies ventral to the cerebral aqueduct and is surrounded by the medial longitudinal fasciculus (MLF), which is clearly distinguishable in the pontomesencephalic junction (Fig. 9). This fasciculus contains mainly ascending and descending second order vestibular fibers. Ventral to the periaqueductal GM and medially located, we found the oculomotor nucleus (III CN) (Fig. 11). Indeed, the medial longitudinal fasciculus (MFL) runs dorsally throughout the whole brainstem, at both sides of the midline, interconnecting the III, IV, and VI CN, being still noticeable at the level of the hypoglossal nucleus (Fig. 3).

Figure 9.

Pontomesencephalic junction. The left part of the brainstem is presented in this image where the left side corresponds to T1 weighted images and the right side corresponds to DTI RGB maps (In red: medio-lateral oriented fibers, in green: antero-posterior fibers, and in blue: superior-inferior fibers). CST: corticospinal tract, CTT: central tegmental tract, FPT: frontopontine tract, LL: lateral lemniscus, MCP: middle cerebellar peduncle, mesVT: mesencephalic tract of the trigeminal nerve, ML: medial lemniscus, MLF: medial longitudinal fasciculus, OPTPT: occipito-parieto-temporopontine tract, PN: pontine nuclei, SCP: superior cerebellar peduncle, STT: spinothalamic tract, TPf: transverse pontine fibers.

Figure 10.

Inferior midbrain. The left part of the brainstem is presented in this image where the left side corresponds to T1 weighted images and the right side corresponds to DTI RGB maps (In red: medio-lateral oriented fibers, in green: antero-posterior fibers, and in blue: superior-inferior fibers). CST: corticospinal tract, CTT: central tegmental tract, FPT: frontopontine tract, IC: inferior colliculus, IVN: trochlear nucleus, LL: lateral lemniscus, ML: medial lemniscus, MLF: medial longitudinal fasciculus, OPTPT: occipito-parieto-temporopontine tract, PAG: periaqueductal gray matter, SNc: substantia nigra pars compacta, SNr: substantia nigra pars reticulata, STT: spinothalamic tract, xSCP: decussation of the superior cerebellar peduncle.

Figure 11.

Superior midbrain. The left part of the brainstem is presented in this image where the left side corresponds to T1 weighted images and the right side corresponds to DTI RGB maps (In red: medio-lateral oriented fibers, in green: antero-posterior fibers, and in blue: superior-inferior fibers). CST: corticospinal tract, CTT: central tegmental tract, FPT: frontopontine tract, IC: inferior colliculus, IVN: trochlear nucleus, LL: lateral lemniscus, ML: medial lemniscus, MLF: medial longitudinal fasciculus, OPTPT: occipito-parieto-temporopontine tract, PAG: periaqueductal gray matter, RET: rubrospinal tract, RN: red nucleus, SNc: substantia nigra pars compacta, SNr: substantia nigra pars reticulata, STT: spinothalamic tract, xSCP: decussation of the superior cerebellar peduncle.

Primary Sensory Relay Structures

The main sensory relay structures of the brainstem at the caudal sections of the medulla are the cuneatus (CuN) and gracile (GrN) nuclei (Figs. 2, 3). The gracile nucleus is medially located and is surrounded by the gracile tract (GrT), which carries ascending information from the ipsilateral lower limb. Lateral to the gracile nucleus and medial to the spinal trigeminal nucleus (spVN), appears the cuneatus tract (CuT) characterized by a dark gray color with a wedge shape. The cuneatus tract carries sensory information from the ipsilateral upper limb to the homologous nucleus.

Ascending Sensory Systems

The gracile and cuneatus nuclei (GrN and CuN) are the centers of origin of the lemniscal system. From these nuclei emerge the arcuate fibers (Arcf) (Fig. 2) which, laterally surrounding the central GM, cross the midline to compose the sensory decussation, and ascend generating the medial lemniscus (ML) in the contralateral side (Figs. 4–6). At cranial sections of the medulla, the medial lemniscus is located at both sides of the ventral midline, with an anteroposterior orientation (Figs. 3–5). In its ascending course through the pontomedullary junction, the medial lemniscus (ML) changes its orientation to progressively become horizontal (Figs. 6, 7). At the pons and midbrain levels, the medial lemniscus occupies a more lateral position neighboring the decussation of the superior cerebellar peduncle (SCP) and the red nucleus (RN) (Fig. 10).

The spinothalamic tract (STT) transmits pain and temperature information from the spinal cord to the thalamus; therefore, it ascends along the whole brainstem. From caudal to cranial, this tract locates laterally near the surface of the medulla (Figs. 2, 3). However, it is better identified at more cranial sections of the medulla (Figs. 4, 5), dorsal to the inferior olive (ION). Once the medial lemniscus (ML) has become horizontally orientated, the spinothalamic tract (STT) can be found adjacent and lateral to this structure (Figs. 6–9). Moreover, at the level of the inferior colliculus (IC) the spinothalamic tract is dorsolaterally located and in close proximity to the brainstem surface (Figs. 10, 11).

The lateral lemniscus (LL), the main auditory ascending pathway carrying fibers from the cochlear nuclei, is first identified at medium levels of the pons (Figs. 6, 7), lateral to the spinothalamic tract (STT). Separating these two structures, the superior olive (SON) appears characterized by a small bright area (Fig. 6). On its ascending course, similar to the medial lemniscus (ML), the lateral lemniscus (LL) lies laterally and close to the surface, ending in the inferior colliculus (IC) (Figs. 8–10).

Descending Motor Systems

The corticospinal tract (CST), the main descending motor tract from the cortex, enters the brainstem through the central portion of the cerebral crus (Fig. 11). In addition to the CST, through the cerebral peduncle descend the frontopontine tract (FPT), medially located, and laterally the occipito-parieto-temporopontine tract (OPTPT) (Figs. 10, 11). These three tracts compacted at the superior sections, descend, and disperse at pons levels compounding different fiber bundles. In fact, the corticospinal tract (CST) occupies the central region of the basilar part of the pons whereas the frontopontine tract (FPT) and the occipito-parieto-temporopontine tract (OPTPT) tend to occupy the ventromedial and dorsolateral regions, respectively (Figs. 7–9). The last two tracts end at cranial pons levels synapsing with the neurons of the pontine nuclei. However, the corticospinal tract (CST) continues its descending course to the medulla, being located at the most ventral pyramids. Finally, at the most caudal level of the medulla (Fig. 2), a great portion of the corticospinal tract (CST) decussates dorsolaterally (xPy). This decussated portion of the CST will form the lateral corticospinal tract in the lateral funiculus of the spinal cord.

Cerebellar System

Somatosensory information from the lower limbs and trunk travels through the dorsal and ventral spinocerebellar tracts (dSCT and vSCT), which are located in the lateral parts of the medulla (Figs. 2–4). The dSCT locates posterior to the spinal trigeminal nucleus (spVN) very close to spinal trigeminal tract (spVT). The ventral portion of the spinocerebellar tract is situated between the dorsal portion and the spinothalamic tract (STT), from which is hard to distinguish. Mainly, the dorsal spinocerebellar tract (dSCT) becomes progressively more dorsal until it converges with the inferior cerebellar peduncle (ICP) (Figs. 4, 5).

The inferior olivary complex is clearly identified at mediocranial levels of the medulla (Figs. 4, 5). It is composed by the principal olivary nucleus (ION), the medial accessory and the dorsal accessory olivary nuclei (mAON and dAON) (Fig. 4). The principal olivary nucleus (ION) is characterized by its undulating shape, with the hil-lium (hION) that opens medially (Figs. 4, 5). From the hillium emerge the olivocerebellar fibers. The principal olivary nucleus (ION) is surrounded by the capsula or amiculum (AmO), with which the central tegmental tract (CTT) converges in its dorsal surface (Fig. 6). From this caudal pons level to the cranial sections of the midbrain (Figs. 6–11), this tract is located in the central part of the tegmentum always located dorsal to the medial lemniscus (ML).

At the pons level, the brighter areas surrounding the corticospinal (CST), the frontopontine (FPT), and the occipito-parieto-temporopontine (OPTPT) tracts correspond to the pontine nuclei (PN) (Figs. 7–9). The characteristic transverse pontine fibers (TPf) originate in these GM regions and course laterally to form the contralateral middle cerebellar peduncle (MCP) (Figs. 6–9).

The superior cerebellar peduncle (SCP), dorsally located at the level of the ponto-mesencephalic junction (Figs. 7–9), changes progressively its orientation to a more medioventral position until its decussation (xSCP) in the caudal section of the midbrain (Fig. 10).

Other Integrative Centers of the Brainstem

Different integrative nuclei of the brainstem are distinguished in the midbrain. The substantia nigra (SN) is ventrally limited by the cerebral peduncle and is clearly identified by its laminated and irregular shape (Figs. 10, 11). The substantia nigra pars reticulata (SNr) lies ventrally behind the cerebral peduncle whereas the pars compacta (SNc) is found dorsally, with a characteristic clumped aspect. Medially, the substantia nigra extends until it confounds with the ventral tegmental area (VTA).

The red nucleus (RN) is located dorsal to substantia nigra and cranial to the decussation of the superior cerebellar peduncles (Fig. 11). It is clearly identified by its round shape and hypointensity. Through the red nucleus (RN) the oculomotor fascicles (IIIN) can be observed. Then, the rubrospinal tract (RET), coursing medial and ventral to the red nucleus (RN) is located both sides of the midline with a characteristic dark gray appearance and with an almond shape (RET) (Fig. 11).

The periaqueductal GM (PAG), very bright, surrounds the cerebral aqueduct at the midbrain level. It limits directly with the III and IV nerve motor nuclei (Figs. 10, 11). The laminated structure identified in the most cranial section of the midbrain (Fig. 11) corresponds to the superior colliculus (SC), dorsally located to the periaqueductal GM.

DISCUSSION

According to our study, images of high-resolution structural MRI and DTI of the human brainstem acquired postmortem at 7 T magnetic field strength revealed the basic cyto- and myeloarchitectonic organization of this region, currently visualized only by histological techniques or higher magnetic fields. To the best of our knowledge, this is the first work showing a multimodal study of the human brainstem architecture correlating the high-resolution anatomical image of the brainstem with the directionality of its fibers obtained from the DTI, derived exactly from the same specimens.

From a general point of view, our study is in agreement with previous works (Fatterpekar et al., 2002; Naidich et al., 2009; van Rooden et al., 2009) demonstrating that, in contrast to the images acquired in living brains, WM in T1 weighted images of formalin-fixed specimens appears darker than GM. This was important to depict the most clinically relevant neuronal structures (Table 1) and their topographic organization within the brainstem. Such a description was made based on several histological and MRI atlases (Olszewski and Baxter, 1982; Tamraz and Comair, 2000; Haines, 2004; Mori et al., 2005; Naidich et al., 2009). In most cases, T1 weighted images provided better structural information than the RGB map. This is the case of the solitary tract (Fig. 4), a pathway carrying descending afferent visceral fibers. Another example of the detail level gained with this technique is the layered organization observed in the superior colliculus (Fig. 11). In addition, under our conditions the brighter area at the caudal spinal trigeminal nucleus was identified as the superficial marginal and gelatinous layers, previously described with histological techniques (Usunoff et al., 1997). In the same way, the small dark aggregations observed in the pars compacta of the substantia nigra (Fig. 11), might correspond to the neuromelanine deposits in dopaminergic neurons (van Domburg and ten Donkelaar, 1991). Conversely, although not labeled in the figures, on DTI images we observed structures which could not be identified in T1 weighted images. We suggest that at cranial levels of the medulla the green band with anteroposterior orientation along the midline might represent some of the raphe nuclei, including raphe pallidus, raphe obscurus, and raphe magnus nuclei, which have not been described previously in MRI atlases. At pontine and midbrain levels, the medial green formations could be more related to midsaggital radial glial system, preserved in the brainstem from developmental stages, as suggested in experimental studies (Mori et al., 1990).

Table 1. List of structures described and their abbreviations
AbbrevStructureAbbrevStructure
  1. Numbers correspond to the figures where the structures are contained.

AmOamiculum of olive 4, 5OCfolivocerebellar fibers 5
Arcfarcuate fibers 3OPTPToccipito-parieto-temporopontine tracts 7, 8, 9, 10, 11
  PAGperiaqueductal gray matter 10, 11
CSTcorticospinal tract 3, 4, 5, 6, 7, 8, 9, 10, 11PNpontine nuclei 7, 8, 9
CTTcentral tegmental tract 6, 7, 8, 9, 10, 11p&mVNprincipal sensory and motor trigeminal nucleus 7
CuNcuneatus nucleus 2, 3RETrubrospinal tract 11
CuTcuneatus tract 2, 3RNred nucleus 11
dAONdorsal accessory olivary nucleus 4SCsuperior colliculus 11
dCoNdorsal cochlear nucleus 5SCPsuperior cerebellar peduncle 7, 8, 9
dMNXdorsal motor nucleus of the vagus 4SNcsubstantia nigra pars compacta 10, 11
drVIIdescending root of the facial nerve (VII CN) 6SNrsubstantia nigra pars reticulata 10, 11
dSCTdorsal spinocerebellar tract 2, 3SolNsolitary nucleus 4
FPTfrontopontine tract 7, 8, 9, 10, 11SolTsolitary tract 4
GrNgracile nucleus 2, 3SONsuperior olivary nucleus 6
GrTgracile tract 2, 3spVNspinal trigeminal nucleus 2, 3, 4, 5, 6
gVIIgenu of the facial nerve (VII CN) 6spVTspinal trigeminal tract 2, 3, 4, 5, 6
hIONhillium of inferior olivary nucleus 4, 5spXINspinal nucleus of the accessory nerve 2
ICinferior colliculus 10STTspinothalamic tract 2, 3, 4, 5, 6, 7, 8, 9, 10, 11
ICPinferior cerebellar peduncle 4, 5TPftransverse pontine fibers 6, 7, 8, 9
IIIoculomotor nerve (III CN) 11vCoNventral cochlear nucleus 5
IIINoculomotor nucleus 11VeNvestibular nuclei 5
IONinferior olivary nucleus 4, 5vSCTventral spinocerebellar tract 2, 3, 4
IVNtrochlear nucleus 10Vtrigeminal nerve (V CN) 7
LLlateral lemniscus 6, 7, 8, 9, 10VIabducens nerve (VI CN) 6
mAONmedial accessory olivary nucleus 4VINabducens nucleus 6
MCPmiddle cerebellar peduncle 6, 7, 8, 9VIINfacial nucleus (VII CN) 6
mesVN&Tmesencephalic tract and nucleus of the trigeminal nerve 7, 8, 9XIINhypoglossal nucleus (XII CN) 3, 4
MLmedial lemniscus 3, 4, 5, 6, 7, 8, 9, 10, 11xPypyramidal decussation 2
MLFmedial longitudinal fasciculus 3, 4, 5, 6, 7, 8, 9, 10, 11xSCPdecussation of the superior cerebellar peduncle 10

Several limitations were found in describing some GM and WM structures. For instance, it was hard to identify the tectospinal tract classically located ventral to the medial longitudinal fasciculus. This can be explained since relatively a small number of tectospinal neurons (in a range of 200) have been demonstrated in primates further suggesting that this tract might be quite small in humans (Nudo and Masterton, 1989). Similarly, the reticular formation territory, the spinotectal and spinoreticular fibers (contained in the anterolateral system) and the corticobulbar fibers (enclosed in the corticospinal tract) were not identified. Similar to what happens with conventional histological techniques, it is highlighting that some tracts were undoubtedly recognized at some levels of the brainstem whereas at other levels it was not easy to distinguish them. For instance, the corticospinal (CST), the occipito-parieto-temporopontine (OPTPT), and frontopontine (FPT) tracts were difficult to separate at the level of the cerebral peduncle, due to their confluence and homogeneous appearance (Stieltjes et al., 2001). However, we were able to identify the centrally located CST from the dorsolateral OPTPT and the ventromedial FPT at the midcranial levels of the pons (Figs. 8, 9). In this regard, it has been experimentally demonstrated that the corticopontine projections have a topographical organization in the pons (Leergaard and Bjaalie, 2007). Fibers originating in the frontal cortex course ventromedially whereas fibers originating progressively in posterior parts of the cortex course more dorsolaterally. In agreement with our findings, it has been shown in primates that associative descending corticopontine projections mainly end at the cranial levels of the pons whereas the motor corticopontine and CST descending projections continue to more caudal levels (Schmahmann et al., 2004). This might explain why the CST, OPTPT, and FPT were discriminated only at the mediocranial levels of the pons (Figs. 8–10).

High-resolution MRI acquired in living patients allows the possibility of acquiring narrow brain radiological windows, near to histological resolution, which in combination with neuronavegation systems could help in the accurate localization of brainstem therapeutical targets. In this line, the electrical stimulation of the pedunculopontine nucleus (PPN) represents an emergent therapy for Parkinsonian patients with intractable akinesia. The localization of this nucleus is based on its topographical relation to different brainstem structures (Aravamuthan et al., 2007). Thus, due to the higher contrast obtained between these structures, high-resolution MRI might be applied for optimizing the surgical targeting of the PPN nucleus. In addition, our study is of great interest for neuroradiologists to better identify structures potentially affected by circumscribed lesions such as arteriovenous malformations, tumors, demyelinating diseases, or posterior circulation infarcts. Hence, the main goal of this work was to generate an exhaustive and accessible guide of the brainstem main gray nuclei and WM tracts, most of them omitted in the current MRI and DTI living human studies. Indeed, the MRI protocol was design to obtain the highest resolution and the maximal GM/WM contrast, regardless of the long acquisition time. To acquire the same image quality in living patients, future studies will be necessary to develop new MRI sequences that reach such image quality in less acquisition time.

Although an exhaustive brainstem atlas has been recently published by using a 9.4 T scanner (Naidich et al., 2009), our work significantly complements it; first, because the description of the structures was performed in parallel between anatomical and DTI images derived from exactly the same specimens; and second, because the present atlas was performed at 7 T magnetic field strength. As it happened with the 3 T systems, which represent now a standard diagnostic tool, 7 T magnets might be the next generation of MRI scanners. Although higher magnetic fields such as 9.4 T or 11.7 T are employed for experimental research its use for clinical research purposes is far to be frequent. Therefore, a precise description of the CNS architecture assessed by 7 T MRI is needed to be used as a gold standard for what these high field magnets are capable to image in patients.

Acknowledgements

The authors thank O. Fuentes, A. Benet, and C. Justicia for their technical and scientific collaboration. G.S. is supported by CSIC (JaeDoc). They acknowledge the Experimental MRI 7 T Unit (IDIBAPS). CIBER-BBN is an initiative funded by the VI National R&D&i Plan 2008–2011, Iniciativa Ingenio 2010, Consolider Program, CIBER Actions and financed by the Instituto de Salud Carlos III with assistance from the European Regional Development Fund.

Ancillary