For each space-occupying lesion, inflammatory or vascular disease should be considered before a real neoplasm is diagnosed. Especially subacute hemorrhage, venous infarction, arteriovenous malformations, large demyelinating diseases, or necrotic infections may look like a real neoplastic tumor. Patient’s history, neurological symptoms, and patient’s age are mandatory information to provide the correct diagnosis. Further, there are some imaging features which may specify the entity of a lesion. These features include diffusion restriction of T2-hyperintense lesions (Fig. 1), capsular and target-like structures of the margin, and the halo sign of the marginal zone (Fig. 2). Veins should be screened for thrombosis and arteries for any other pathology (e.g., aneurysm, malformations). Beside hemorrhage, T2*WI and especially thin-sliced susceptibility-weighted imaging (SWI) depict the “blooming” of any collection of blood either in hemorrhages and micro bleeds, cavernomas, or in vascular thromboses (Tong et al. 2008) (Fig. 3). The skull base, orbit, and viscerocranium should be screened for tumors and inflammatory diseases as they may secondarily infiltrate into the brain (Fig. 4). Especially in younger patients, the eyes and skin should also be inspected for malformations and hamartia indicating neurocutaneous diseases (Kandt 2003).

The most important question regarding the tumor entity is the differentiation between intracerebral (also named intra-axial) and extracerebral (extra-axial) localization. Extracerebral tumors are mostly benign neoplasms like meningiomas and schwannomas, whereas intracerebral tumors are malignant in the majority of cases, especially in older patients. Extracerebral tumors are rare in younger patients; thus neurocutaneous syndromes, previous radiation therapy, or metastases (e.g., neuroblastomas) should be considered.

Signs indicating an extracerebral tumor are illustrated in Fig. 5.

The next question addresses the number of tumors. Singular large intracerebral tumors are mostly glial tumors; otherwise metastases should be suspected.

Finally, the possibility of a cerebral lymphoma should be considered. Corticosteroids should be withheld if a lymphoma is a differential diagnosis based on the imaging findings until a stereotactic needle biopsy has been performed (Fig. 6). Lympholytic activity of corticosteroids makes the histopathological diagnosis more difficult or even impossible. Therefore, radiologists should be familiar with imaging features of this tumor entity (Bühring et al. 2001).

The description of tumor localization should consider two aspects: the precise anatomical localization and the spatial relation to functional representations of motor and language skills. Anatomical designation not only means the lobe and gyrus but also the relationship to white and gray matter structures. The infiltration of the cortical ribbon (cortical ribbon sign) and thickening of the corpus callosum are typical features of infiltrative glial tumors which help to differentiate them from vasogenic edema (Fig. 7).

The primary motor, sensory, auditory, and visual cortices are assigned at the brain surface. However, surface relief may be difficult to recognize in sectional images and anatomy might be distorted by the tumors. The identification of surface anatomy is easier with interactive observation of 3D objects. Further, planar reformatting of the brain surface helps to delineate the central sulcus which marks the perirolandic region with the primary sensorimotor cortex (Table 2) (Figs. 8 and 9).

However, this direct allocation of anatomy to function and vice versa is only true for primary cortical areas: the variance increases with the complexity of brain function. One of the most investigated functions is the language since neurosurgeons have the primary goal of preserving the language during tumor resection. Another critical issue is to protect important white matter tracts. To achieve these goals, functional MRI (fMRI) and tractography are well-established components of presurgical imaging (Fig. 10). MR tractography virtually dissects functionally critical white matter tracts, such as the corticospinal tract and the arcuate fascicle (Fig. 11), enabling the neurosurgeon to plan the surgical approach which best preserves the tract during resection. However, uncritical and inexpertly handling of these methods bears the imminent danger of misinterpretations (Jellison et al. 2004).

Therefore, the main indication and the undoubted strength of fMRI and DTI is the presurgical planning. Functional language MRI can localize language dominant hemisphere (Ruff et al. 2008; Roux et al. 2003; Kim et al. 2009; Spreer et al. 2002) (Fig. 11). However, surgical resection of tumors in the language areas of the dominant hemispheres still requires awake craniotomy and direct brain mapping to prevent postsurgical aphasia (Kim et al. 2009). Tractography may depict the relation between tumor and white matter tract, but it should not determine the extent of resection during surgery.

The imaging criteria of malignancy include blurring tumor margins, tumor edema, necroses, and contrast enhancement. However, these signs of malignancy are not as reliable as they allow the definite categorization into low- and high-grade tumors. Diffuse gliomas infiltrate the brain tissue by definition. Therefore, they mainly do not have sharp margins. In contrast, metastases and also highly malignant PNETs often have sharply delineated margins between tumor and edema (Fig. 12). Low-grade neuroepithelial glioneuronal tumors like gangliogliomas and DNETs may be associated with cortical dysplasia which may blur anatomical structures. Further, especially gangliogliomas may show areas of contrast enhancement (Fig. 13). Necroses are difficult to differentiate from tumor cysts; in both, the margins may enhance. However, cysts enhance linearly in contrast to the often partially nodular enhancement at the margins of tumor necroses.

Malignant gliomas generate tumor vessels with impaired blood-brain barrier. The higher vascular permeability causes extravasal accumulation of contrast agent with consecutive signal increase on T1WI. Therefore, contrast enhancement is one important hallmark of tumor malignancy. Considering that most of the WHO grade I astrocytomas and a larger amount of grade II oligodendroglioma (White et al. 2005) may also enhance due to higher vasculature, this sign of malignancy has a limited value (Fig. 14). On the other side, about 30 % of high-grade WHO III astrocytomas do not enhance contrast agent (Scott et al. 2002; Muragaki et al. 2008; Chaichana et al. 2009) (Fig. 15).

Although the histopathological WHO classification still is the gold standard to categorize tumor entity and their malignancy, the molecular genetic profile of a brain tumor becomes more and more important for tumor diagnosis. Up to date, there are only few studies investigating imaging features of brain tumors with different molecular profiles (Aghi et al. 2005; Eoli et al. 2007; Diehn et al. 2008). Considering that these profiles determine the metabolism of the tumor, MR spectroscopy and metabolic imaging with PET should be the methods of choice to recognize molecular pattern (“PET Imaging of Brain Tumors”).

Monitoring treatment effects on brain tumors becomes more and more complex. In conventional MR imaging, contrast enhancement is the hallmark of tumor growth. However, this imaging feature is more than ambiguous regarding tumor monitoring. On the one side, endothelial cells are sensitive to radiation and chemotherapy and—most often associated with temozolomide treatment—therapy-induced changes may yield a blood-brain barrier damage with contrast enhancement and mass effect of the treated brain tissue (Fig. 16). Neuro-oncologists introduced the term “pseudoprogression” defining an increased (>25 % in diameter) or new enhancement of irradiated tissue usually detected within 3 months of radiation that subsequently abates without further treatment. This phenomenon is observed in approximately 20–30 % of patients treated with today’s standard therapy for glioblastomas (Brandsma et al. 2008). On the other side, new antiangiogenic treatments normalize the blood-brain barrier damage and thus reduce the contrast enhancement independent from real antineoplastic effects, which is named “pseudoresponse” (Figs. 17 and 18). Being familiar with the imaging features of infiltrative brain tumors may however help to recognize nonenhancing tumors (Fig. 18). Neuro-oncologists are aware of these diagnostic challenges and addressed these problems in their new response criteria for malignant gliomas (Wen et al. 2010).

Figure 19 illustrates these response criteria (RANO) (Wen et al. 2010).

The measurement of contrast-enhancing lesions in irregularly shaped, necrotic, and inhomogeneous or ring-enhancing lesions is challenging and interobserver variability is high (Vos et al. 2003). The most important issue in monitoring glial tumors is to depict the tumor regardless whether or not it enhances. The imaging features of this nonenhancing tumor were more or less ignored longtime. Nonenhancing tumor, or rather brain tissue which is infiltrated by glioma cells, may look like vasogenic edema or gliosis since each of these entities increases T2 relaxation time and is thereby hyperintense on T2-weighted images. However, there are some imaging characteristics which may help to distinguish edema from brain tumor, bearing in mind that tumor cells are often found in both normal and edematous brain tissue (see also Figs. 7 and 17).

To guarantee the comparability of images, a standardized protocol should be mandatory, which has not yet been implemented. The slice thickness should not exceed 5 mm in order to minimize partial volume effects (Wen et al. 2010). Sagittal high-resolution 3D sequences might be advantageous to avoid effects from different slice angulations and partial volume effects, but they also have disadvantages considering movement artifacts and T1 contrast (see Figs. 23 and 24). Further, once there is an artifact (e.g., pulsation artifact), it might be reconstructed in three planes (MPR) simulating a real lesion

Further, each monitoring begins with an early postsurgical MRI in order to detect the residual tumor which has to be monitored thereafter. Postsurgical control should always include DWI to detect infarcts, which may show confounding enhancement in the follow-up (Fig. 20).