Preservation of motor function is critical and mapping of the corticospinal (pyramidal) tract (CST) and motor cortex is most commonly requested. There are multiple reports of its utility in preoperative planning [13, 16–20]. If pathology has not sufficiently altered the tract, basic single tensor DTI FT for the CST in general works well for trunk and in some cases hand and lower extremity; however, it is less successful for cortical regions like the face because intervening tracts like the SLF interfere with continuous tracking in critical voxels due to crossing fibers in these locations. Some detail in somatotopic mapping is possible [6, 13]. Yamada et al. [21] used cortical ROIs for lower extremity, trunk, hand, face, and tongue and found robust tracking only for the trunk with a single tensor approach while a HARDI-based multitensor approach was more successful in tracking for other cortical regions including face and tongue. Techniques such as HARDI will be required if detailed somatotopic mapping is desired [3, 21].

Multiple studies have addressed validation of cortical and subcortical motor mapping with DTI FT [3, 15, 22–24]. Berman et al. [22] used seed points identified during intraoperative stimulation mapping (ISM) to retrospectively map ipsilateral CST in 11 glioma patients using the preoperative imaging data after the surgery was complete. Though not used for surgery, they showed appropriate tracking to the central cerebral peduncle based on the seed points verified intraoperatively; however, there were some failures in tracking depending on the extent of pathologic alterations in or near the CST. Kamada et al. reported studies confirming CST mapping with subcortical [23] and cortical mapping [15]. The latter included 30 patients with lesions in proximity to the motor system, with central sulcus and sensorimotor cortex identified using DTI FT and verified with intraoperative cortical somatosensory evoked potential (SEP) measurements. Functional MRI or MEG identification of primary motor cortex failed in nine patients while DTI FT was successful in all.

Accuracy of CST localization with DTI FT compared with subcortical stimulation appears to be within 10 mm [3, 24, 25]. Berman et al. [3] reported a mean distance of 8.7 mm (±3.1 mm standard deviation) for 16 sites in 9 patients. Mikuni et al. [24] found consistent motor evoked potentials (MEP) at 7 mm or less and absent MEP over 13 mm when stimulating near the mapped CST fibers at the inferior aspect of the mass. Between 8 and 12 mm, stimulation near the mapped CST at the level of the corona radiata consistently elicited MEPs. Another study of 40 patients with masses within 2 cm of CST found MEPs elicited at less than 1 cm from mapped CST in 18 of 21, negative MEPs in 5 at stimulation points greater than 2 cm, and negative MEPs in 12 of 15 at 1–2 cm from mapped CST [26]. Motor function was ultimately preserved in all, improved in five cases with weakness prior to surgery, and two of the cases with negative MEP at 1–2 cm distance developed SMA syndrome , which subsequently resolved.

Other studies have also addressed impact on surgical plan and/or outcomes [16, 25, 27–31], in general with preserved or improved clinical function postoperatively. Bello et al. [27] reported high correlation of DTI FT for CST (and multiple language tracts) with intraoperative subcortical mapping (ISM). With low grade gliomas, fibers tracking through the neoplasm were often found. A combination of DTI FT and ISM decreased surgery time, patient fatigue and intraoperative seizure rate.

With increased extent of resection, postoperative deficits have been reported to increase transiently but with improvement over time [25, 32, 33]. The concept of a safety margin of about 5–10 mm has been introduced, but with careful intraoperative mapping this boundary may be minimized [32]. In a prospective randomized study by Wu et al. [30], 238 patients with suspected primary supratentorial glioma (214 ultimately proven gliomas) not previously treated and near the pyramidal tract with at least 1/5 motor strength were randomized to surgery with conventional neuronavigation versus neuronavigation with integrated DTI data. Only FA maps were used for intraoperative mapping, although in the later stages of the study tractography was added for preoperative planning and postoperative assessment (23 of the cases). Those in the neuronavigation plus DTI group were more likely to have complete resection of high grade tumors (74.4 % vs. 33.3 %), although low grade gliomas had similar extent of resection. Decline in motor function postoperatively was less likely overall (15.3 % versus 32.8 %) and 6 month Karnofsky scores were better (86 ± 20 vs. 74 ± 28). They also reported improved survival for high grade gliomas (21.2 months vs. 14.0 months).

The optic radiations (OR) and in particular Meyer’s loop pose difficulties for DTI FT. Problems include not only the anatomy with tight turns, but location along medial temporal lobe where susceptibility-related artifacts can affect image quality with DTI sequences using EPI [34, 35]. Crossing fibers can create problems here as elsewhere. There are early reports using DTI FT to map visual pathways with integrity of OR on DTI FT linked to deficits pre- and/or postoperatively [29, 35–38], and intraoperative correlation also described in at least one case [36]; however, results remain preliminary and further validation is required and caution warranted in this setting [35].

Validation of language tractography is also ongoing and though reports are less extensive than for CST mapping, experience is growing. Co-localization with intraoperative mapping seems to be similar to CST; that is, distance between DTI FT prediction and subcortical mapping is in the range of 5–10 mm. For example, Kamada et al. [39] studied 22 lesions near the superior longitudinal/arcuate fasciculus (SLF/AF) in the dominant hemisphere, using fMRI or MEG to choose seed and target ROIs. For two of these cases, functional mapping was integrated into neuronavigation and validated by cortical and subcortical mapping during awake surgery with stimulation near AF leading to paranomia without speech arrest at a distance of 6 mm. Leclercq et al. [40] studied multiple language tracts in ten patients with low grade gliomas or dysplasia, reporting positive subcortical stimulation responses 6 mm or less from DTI FT prediction in 81 %, but they also reported four false negative results (positive stimulations away from any mapped fiber tract). Of course, understanding of language functional cortical regions and connections remains incomplete, and important fiber tracts not anticipated and thus not mapped are considerations as well [41].

False positives for DTI FT based on subcortical mapping have also been described [27], with DTI FT mapping of SLF larger than identified on subcortical mapping in some cases such that ISM did not evoke a response in some tracked fibers, and there were also false negatives in this study which could be addressed in part by adjusting ROI placement. For low grade gliomas, tracts could be identified within the neoplasm and in some cases ISM evoked a response within the mass even when tracts appeared interrupted. IFOF and UF were also evaluated in this study. The majority of patients where motor or language tracts were identified with ISM had decline immediately after surgery, in general improving after 1 week (about 90 % normal at 1 month).

Zhao et al. [42] used preoperative as well as intraoperative DTI FT for the SLF/AF in 20 patients, 9 of which had no preoperative deficit and 11 with variable conduction aphasia. They monitored intraoperative shifts of the tract from −5 to 2 mm. After surgery, follow up within 1 month revealed one previously normal patient with new conduction aphasia while those with language deficit had improved. Many of the studies discussed here have integrated DTI FT data with neuronavigation systems , but not all scenarios necessarily require it. Review of preoperative data may be sufficient for confirming or altering an operative approach. Powell et al. [43] used a probabilistic fiber tracking method for hemisphere dominance prior to anterior temporal lobe resection, finding a significant correlation between frontotemporal tract lateralization to dominant hemisphere resections and postoperative naming deficits. Romano et al. [29] performed tractography for CST, OR, and AF in 25 patients, and for 35 of 75 tracts analyzed which were in proximity to a given neoplasm reported a change in surgical plan in 16 % and a change in extent of resection in 68 %.

The concepts for sparing eloquent structures in surgery extend also to radiosurgery and to an extent to other therapeutic radiation strategies. Koga et al. [44] for example integrated tractography for CST, OR, and AF with navigation for AVM Gamma Knife radiosurgery in 71 of 155 treatments for 144 patients, optimizing dose such that the maximum dose to CST, OR, and AF was 20 Gy, 8 Gy, and 8 Gy respectively. Of the 71/155 treatments planned with this approach, 60 % had less than 5 mm separation from these WM tracts. Maruyama [45] integrated tractography for OR with navigation in ten patients receiving Gamma Knife near OR, reporting 8 Gy or more single dose to be associated with neurological changes such as visual field deficit and new (or improved) migraine symptoms. Another study included 24 patients (9 retrospective and 15 prospective) [46], reporting maximum dose to CST of 23 Gy with 5 % complication risk and 20 Gy and 25 Gy volumes for 5 % complication risk were 58 mm³ and 21 mm³, respectively.

There is also an interest in using DTI metrics for microstructural assessment not only to estimate extent of infiltration [47–50] but also to improve accuracy in diagnosis, assess grade or prognosis, evaluate treatment response, and potentially predict recurrence patterns. Jena et al. [48] described the use of a DTI-based “high risk volume” or IHV_DTI which allowed a 35 % reduction in planning target volume and a dose escalation such that complications did not exceed conventional plan.

Conventional imaging is often inadequate for visualization of important functional anatomy; for example, the subthalamic nucleus (STN) is difficult to demonstrate with high image contrast and targeting is not trivial [51]. DTI data can be used to improve localization of functional anatomy relevant to deep brain stimulation (DBS) placement. Simple integration of color FA maps has been utilized to augment identification of STN and globus pallidus interna (GPi) targets [52].

Other studies have shown proof of concept for use of DTI FT in this setting. Pouratian et al. [53] retrospectively studied cases where DBS had been placed using traditional methods for tremor control. The procedure was efficacious in 11/12 electrodes placed (six patients), and postoperative CT or MR fused with preoperative probabilistic tractography for thalamic voxel connectivity to primary, premotor, and prefrontal cortex showed thalamic sites with connectivity to premotor and supplemental motor regions were most efficacious in controlling tremor. A postoperative study of white matter tracts in proximity to leads placed in GPi, STN, and ventral intermediate nucleus (VIM) was useful in understanding side effects related to stimulator setting [54]. Tracts of interest in proximity to these targets included motor, premotor, supplemental motor, and frontal eye fields. Coenen et al. [55] also used a retrospective approach for DTI FT, showing unilateral refractory tremor control in a patient with Parkinson disease with DBS placement involving the dentate-rubral-thalamic tract (DRT) , subsequently prospectively targeting the DRT for DBS implantation bilaterally with excellent control of head tremor in a patient with myoclonus dystonia [56]. Other fields of interest where tractography may be potentially useful in this setting include neuropsychiatric problems [57, 58] including OCD and depression [59, 60], headache and chronic pain [61, 62].