sDTI provides practical information in the perioperative surgical evaluation of spinal pathology. Nearby tract fibers can be located, and quantitative data allow clinical disease to be evaluated [50]. Fractional anisotropy (FA) provides a scalar value from 0 to 1 denoting the degree to which observed molecules preferentially diffuse across a single axis [50]. In healthy myelinated axons, FA values within a specific area of interest, or voxel, approach 1 as molecules move in a relatively linear fashion within an axon, but injured axons return FA values approaching 0 as molecules diffuse randomly in all directions in the absence of a functional cellular barrier [50]. Similarly, apparent diffusion coefficients (ADC) quantify the degree to which water diffusion across tissues is impeded, presumably because of intact cellular membranes [50]. Both values, along with additional parameters, have been heavily investigated to assess the extent of or changes in tissue injury.

Degenerative spinal disease represents a broad grouping of related conditions and accounts for upwards of 30.1% of public health insurance claims in 2021 in the USA with increasing prevalence [51]. Given the complex multifactorial etiology and the challenge of managing a spectrum of interrelated pathologies, studies have attempted to elucidate disease prognosis and treatment efficacy in spinal spondylosis using objective grading schemes, radiographic evidence, and patient-reported outcome measures (PROMs) [52,53,54]. The literature reviewed showed a significant proportion of reports describing sDTI in prognosis prediction, severity grading, and surgical outcome evaluation in patients with spondylotic myelopathy or radiculopathy [2, 4, 5, 7,8,9,10,11,12,13, 15,16,17,18, 21,22,23,24, 28, 30,31,32, 35,36,37,38,39,40, 42]. A total of 12 cases evaluated sDTI correlations with the severity of cervical spondylytic myelopathy or spinal canal stenosis (Table 1). Among these, 58% (7/12) reported significant associations between FA and the measures investigated, 29%(2/7) of them including multivariate analysis [22, 23]. FA along with ADC correlated significantly with severity of myelopathy as illustrated by Nurick grade, suggesting that clinically significant injury to white matter tracts was detected accurately using sDTI [22, 23]. In the univariate analyses including 338 patients, FA was significantly associated with disease severity, discrimination between myelopathic and normal subjects, and modified or classic Japanese orthopedic association scales(mJOA/JOA) [2, 16, 35, 36, 38]. Additionally, FA, along with axial(AD), radial(RD), and mean diffusion(MD), was associated in all white matter columns at affected levels [5]. One univariate analysis across 129 patients reported no significant relationship between FA or ADC [35] and recovery rate [35], while another including 40 patients [16] reported no association of FA with mJOA. As both of those reports reported associations between said parameters in either JOA [35] or pathology versus normal [16], it is reasonable to conclude that FA and ADC are useful for identifying myelopathy and determining severity, but do not necessarily allow the slope of meaningful recovery to be predicted. Additionally, one univariate [35] and one multivariate [36], provided conflicting results about the usefulness of fiber tract density (FTD), the univariate study showing meaningful correlations of FTD with recovery rate and JOA and the multivariate study reporting no correlation between FTD and mJOA, though FTD was significantly associated with pathology versus normal (Table 1). The usefulness of FTD for identifying the severity or prognosis of myelopathy therefore remains uncertain and is a recommended focus for further investigation.

A total of 14 reports discussed the use of sDTI for assessing operative outcomes in cervical myelopathy [4, 7,8,9,10,11,12, 15, 18, 21, 24, 28, 30, 39, 42]; 11 of those 14 discussed results from posterior decompression with or without fusion (Table 1), seven (64%) including multivariate analyses [4, 9,10,11, 18, 28, 30]. Multivariate studies revealed significant associations between FA and mJOA in 42% of reports (5/12), FA with visual analog scales (VAS) and neck disability index (NDI) in one report and poor outcome as determined by a JOA recovery rate less than 50% in another (Table 1). An additional multivariate study found FTD to be negatively associated with mJOA at the most compressed level across 27 patients [10]. Also, two studies found FA to be correlated with differences in mJOA measured pre- and postoperatively [4, 24]. One univariate study across 15 patients found no associations of FA or MD with pre- versus postoperative mJOA values despite a noted significant improvement in mJOA status post-decompression [15]. In addition to FA and FTD, one univariate study investigated ADC, RD, AD, and MD before and after surgery, with significant findings for all parameters relating to changes in mJOA [42]. Another two multivariate reports queried FA or FA and FTD before and after cervical laminoplasty, with conflicting results [12, 21]. The first study, a prospective cohort across 75 patients, found FA to be an independent predictor of mJOA recovery rate at three and 6 months following surgery [12]. The second study, a case series across 20 patients, found significant associations between poor recovery and FA and reported a decreased FTD as an independent predictor [21]. The remaining study, focusing on DTI and surgically-treated cervical myelopathy, queried parameters relative to ACDF across 30 patients in a univariate fashion using multi-shot high-resolution DTI (msDTI) [39]. The authors reported a significant correlation between pre- and postoperative mJOA and FA values [39]. Considering the foregoing, it can be concluded that FA is likely at minimum to have significant clinical value for evaluating operative outcomes from laminectomy, laminoplasty, or ACDF, which is intuitively valid as those procedures are known to provide clinical benefit,[55, 56] and observed changes would be expected to correlate with radiographic parameters associated with white matter viability.

Of the seven reports that focused on degenerative lumbosacral pathology, three queried the severity grading of radiculopathy [8, 32, 40], two surgical posterior column decompression [13, 37], and two PTED [17, 31], one of which was the only lumbosacral report to include multivariate analysis [31]. All three reports that commented on severity grading described positive associations between FA and ADC and the extent of disease according to PROMs or observed severity of spinal canal or foraminal stenosis (Table 1). Two of the reports [31, 37] discussed significant associations of FA or FA and ADC with improvement in surgical outcome measures and found FA to be an independent predictor of mJOA [31]. The third, a univariate analysis, found FA to correlate significantly with severity of compression at three and 30 days post-intervention, but significance was lost at 90 days [17]. The final study was a case report describing increases of multiple outcome measures along with improvement in FA values following posterior column decompression at one, two and three months post-surgery [13]. Although they were less numerous, studies investigating degenerative lumbosacral disease therefore paralleled those focusing on cervical pathology. One question arising is whether sDTI is clinically necessary in pre- and postoperative assessment, given that all associations are relative to PROM scales as a surrogate for operative efficacy. It is likely that the added costs of performing interval MRI in such patients provides only academic value, which could pave the way for further developments.

Of the seven reports discussing the use of sDTI in treating spinal cord neoplasia or vascular malformations (limited to cavernous malformations), three studies were case series [7, 41, 47] with univariate analyses while the remainder comprised reports of 1–4 cases [1, 3, 4, 6, 25, 46]. The level of evidence for such pathologies in these reports was therefore no greater than IV (Table 2). The studies overwhelmingly discussed the use of sDTI for preoperative assessment of lesion resectability (77%,7/9), all such lesions being intramedullary and including all spinal cord segments (Table 2). The three case series with statistical analyses found the method useful for identifying discrete operative plans, one finding a significant association between FA and tensor trace (TRACE) [45]. The remaining two reports described sDTI and pre-defined resection types, grouping the lesions based on the proportion of fibers identified on diffuse tensor tractography (DTT) as piercing or circumventing them [7, 41]. One such study found an association between DTT category and FA and ADC values, suggesting a more objective bridge to operators or radiologists, individually identifying the locations of tracts by image appearance [41]. The other study found sDTI category to be significantly positively and negatively associated, respectively, with distinct intraoperative planes in type 1 (no fibers passing through lesion) and type III (large volume of fibers passing through lesion). Accordingly, most type 1 lesions were amenable to gross total resection while type III lesions were more often managed by subtotal resection or biopsy, although these findings were not statistically significant [27]. Likewise, no significant differences were found between sDTI type and postoperative McCormick grade, neurological improvement, or good functional outcome [27].

Two case reports described preoperative sDTI for intraoperative neuronavigation in the resection of IDIM and IDEM tumors [3, 25]. One reported the successful use of sDTI integrated into an intraoperative virtual reality display for various degrees of resection for an IDIM cervical ependymoma and glioblastoma [3], while the other reported the use of preoperative sDTI combined with intraoperative computed tomography in resecting an IDIM recurrent cervicomedullary subependymoma and an IDEM thoracic spinal neurofibroma [