Introduction

Lumbar degenerative disease increases with age and is common in the elderly population. The symptoms caused by the osseous degeneration affect the quality of life and the risk of mortality from surgical treatment increases in the elderly population [1]. Radiographic manifestations serve as a link to symptoms and clinical findings for degeneration diagnosis. Previous studies have confirmed that there is relationship between osseous structures and symptoms. For instance, facet orientations and facet joint asymmetry have been associated with lumbar facet joint osteoarthritis (OA) [2], and the patients without lumbar symptoms have wider foramina from L3 to L5 than those with lumbar symptoms [3]. To quantify and qualify lumbar morphometry is essential, especially for reproducible anatomical assessment and differentiation between normal and pathological conditions, which could offer the essential information for treatment and the surgery planning.

Several imaging modalities, including X-ray, multi-slice computed tomography (MSCT), and magnetic resonance imaging (MRI), have been utilized for evaluation of lumbar degeneration. In the lumbar bone morphology radiologic assessment, radiography is the modality with the screening purposes, while MSCT is used for surgical planning and more detailed assessment of osseous morphology with properties of high resolution and contrast although it is associated with radiation exposure. Conventional MRI has excellent depiction and better soft tissue contrast in evaluation of intervertebral discs and other soft tissue pathologies in spine based on the proton content in the tissues and acquisition parameters such as repetition time and echo time [4]. However, cortical bone shows no signal in conventional MRI images for its extremely short transverse relaxation time caused by highly organized structure with low-proton content [5]. Therefore, MSCT and MRI examinations are both required for spine degenerative assessment for the osseous as well as soft tissue information in the clinical practice before the treatment [6].

An advanced MRI technique, zero echo time (ZTE), is a relatively silent 3D sequence with rapid acquisition with immediately center-out k-space fill-in and steadily increases gradient instead of on–off switching after data acquisition and less sensitive to gradient fidelity [7,8,9]. The sequence could capture and reveal short-T2 materials, allowing to obtain osseous information without ionizing radiation. ZTE has been used for the visualization of bone with high spatial resolution and CT-like contrast in several structures including cranium, extremities and cervical foraminal stenosis [10,11,12,13,14,15], which has shown strong agreement in numerical measurements and ordinal variables of the bone. Additional CT examination may be reduced if the ZTE could depict the lumbar osseous structure for degeneration assessment, thus avoiding the ionizing radiation exposure.

The purpose of this study is to determine and compare the performance of the ZTE and conventional MRIs with MSCT as image standard in evaluation of the lumbar osseous changes at different degeneration levels.

Materials and methods

This study was conducted under the approval of the institutional review board in our hospital and written informed consent was obtained.

22 subjects (14 females, 8 males; age: 52.68 ± 9.90 years) were recruited from daily lumbar MRI with degenerative concern in our institution. Inclusion criterion was a patient with a scheduled or prior MSCT examination. Exclusion criteria were lumbar traumatic or surgery between MSCT and MRI examinations, idiopathic scoliosis, ankylosing spondylitis and general contraindications to MRI. Conventional MRI sequences listed in the clinical protocol and ZTE images were acquired in order with imaging parameters shown in Table 1 on a 3.0 T scanner (Pioneer, GE Healthcare) with a 32-channel body array coil. Participants were positioned supine, feet-first with feet placed neutral. MSCT scans were acquired as follows: FOV: 16 × 16 to 34.5 × 34.5mm2 varying with the patient sizes, in-plane pixel range: 0.312 × 0.312– 0.791 × 0.791mm2, slice thickness: 0.8–1.25 mm. ZTE and MSCT acquisitions with approximately isotropic resolutions enabled multi-planar reformation. ZTE-MRI showed greater bone conspicuity and appeared CT-like contrast after post-processed with correction for signal intensity irregularities, logarithmically transformation and intensity inversion, background (air) removement [15, 16].

Table 1 Overview of conventional MRI and ZTE-MRI parameters
Full size table

Image analysis

In this study, degenerative osseous changes within vertebral body and corresponding facet joints in MSCT, ZTE and conventional MRI images were evaluated. Osseous morphology parameters included were as follows: axial facet joint orientations, the lateral recess, anteroposterior diameter of the foramina, interarticular distance (IAD), pedicle width, pedicle height and the facet joints OA grade, and osteophytes grade.

All images were anonymized, randomized and independently analyzed by three fellowship-trained, full-time musculoskeletal radiologists with 3, 4 and 4 years of experience in interpreting musculoskeletal system, respectively. The readers were provided with MSCT, post-processed ZTE and conventional MRI images for evaluation of facet joints OA, osteophyte grade, relatively. All quantitative evaluation was performed after the co-registration between each two modalities to assure the consistent view and measure at the similar slices (CT-MRI, CT-ZTE). There was one week interval between different modalities with the same reader to avoid the bias of qualification assessment.

Interarticular distance (IAD) was measured at the level of the disc slice joining the internal borders of the facet articulations [3]. The lateral recess was the distance measured bilaterally at the level of the upper vertebrae platform between the posterior edge of the vertebral body and the anterior part of the articular facets [3]. The anteroposterior diameter of the foramina was determined at dorsal ganglia level as the distance between the posterolateral edge of the vertebral body and the anterior part of the facets [3]. At the axial plane bisected the intervertebral disc level, axial facet joint angles were measured via a line drawn between the anteromedial and posterolateral of each facet joint and the other line drawn in the midsagittal plane of the vertebrae [17, 18], and facet joint asymmetry is calculated each level of the lumbar. Pedicle width measurement was conducted on an axial plane parallel to the upper vertebral endplate at the middle height of the pedicle [19,20,21]. Pedicle height was measured on a sagittal plane parallel to the anteroposterior midline of a vertebral body [19,20,21]. The illustration is provided in Fig. 1.

Fig. 1
figure 1

Measurement illustration for osseous structure parameters of the vertebrae and facet joints on CT images in the axial plane: a axial orientations, b foramina diameter, c lateral recess parameters, d pedicle width, e pedicle height, and f IAD

Full size image

The grade for each facet joint OA was defined as grade I (Normal), grade II (Mild), grade III (Moderate), and grade IV (Severe) [2, 22, 23]. The osteophyte formation was assessed in term of the number and length of osteophytes growing at the edges of the vertebral body with four-point, grade I (Normal), grade II (Mild), grade III (Moderate), and grade IV (Severe) [24].

The overall image quality of different modalities was graded on a five-point Likert scale () (inadequate: 1, poor: 2, moderate: 3, good: 4, excellent: 5) [26]. The detail is displayed in Supplement Table 1.

Statistical analysis

Statistical analyses were performed using SPSS software (version 24.0). Intraclass correlation coefficient (ICC) [27, 28] was applied for the agreement of numerical measurements between the readers and modalities. The agreements for the grading of facet joint OA, osteophyte, respectively, between readers and modalities were determined with Weighed Cohen’s κ. For statistical analysis, 95% confidence intervals (CI) were calculated. Bland–Altman plot was used to illustrate bias and consistency. The Kruskal–Wallis test was employed to determine the difference of the diagnosis and image quality with different modalities. The ICC and Kappa values for agreement level were categorized into poor (κ ≤ 0), slight (0 < κ ≤ 0.20), fair (0.20 < κ ≤ 0.40), moderate (0.40 < κ ≤ 0.60), substantial (0.60 < κ ≤ 0.80) and almost perfect (0.80 < κ ≤ 1.00) [29]. P value < 0.05 was considered statistically significant.

Results

Quantitative assessment

The overview of the quantitative measurement is provided in Table 2. There was no significant difference for the measurements of axial orientation (p = 0.444), IAD (p = 0.381), lateral recess (p = 0.370), pedicle width (p = 0.067) and pedicle height (p = 0.056) on conventional MRI, ZTE-MRI and MSCT. The measurement of the foramina diameter was statistically different between conventional MRI and MSCT (p < 0.05) but not between the MSCT and ZTE (p = 0.660).

Table 2 Overview of osseous structure parameters assessment
Full size table

The agreement between the modalities is provided in Table 3. The substantial to almost perfect level (0.613–0.977) agreement between the MSCT and conventional MRI was relatively lower than the almost perfect level (0.903–0.978) agreement between the MSCT and ZTE images.

Table 3 Overview of the inter-modality agreement for quantitative and qualitative assessment
Full size table

The details of the inter-reader assessment are provided in Table 4. The agreement between readers for MSCT, ZTE-MRI and conventional MRI were almost perfect level (0.811–0.974), almost perfect level (0.820–0.980) and from the moderate to almost perfect level (0.504–0.984). Bland–Altman plots illustrated inter-modality difference of quantitative measurements in Fig. 2.

Table 4 Overview of the inter-reader agreement for quantitative and qualitative assessment
Full size table
Fig. 2
figure 2

Bland–Altman plots for lumbar osseous measurements show the bias and 95% limit of the agreement between the MSCT and conventional MRI (thick dash line) and the agreement between the ZTE and MSCT (fine dotted line)

Full size image

Qualitative assessment

Qualitative evaluation including the facet joint OA, osteophyte grade in the patients with degenerative lumbar on MSCT, ZTE-MRI and conventional MRI is presented in Table 2. The facet joints OA grades were dominantly grade II (MSCT: 48.2%, ZTE: 47.7%, MRI: 62.7%); those of the L4-5 and L5-S1 levels were mostly grade IV; most of joints in L1-2 levels were grade I. Most osteophytes were grade II and III (MSCT: 75.5%, ZTE: 68.15%, MRI: 72.8%); those in L4 and L5 vertebral bodies were mainly grade III and IV. The degree of facet joint OA was statistically significant between the MSCT and MRI images (p = 0.000) but not between the MSCT and ZTE images (p = 0.293). There was no significant difference for the assessment of osteophyte grade for the MSCT, ZTE and MRI images (p = 0.052).

The inter-modality and inter-reader agreements are shown in Tables 3 and 4. The weighed kappa for the agreement of ordinal scaled variables indicated substantial to almost perfect level between the MSCT and ZTE (0.720–0.824) and substantial level between the MSCT and MRI (0.633–0.710), relatively. The agreements for qualitative assessment between readers were substantial level for ZTE images (0.647–0.720), moderate to substantial level for MSCT images (0.550–0.775), and moderate to substantial level for MRI images (0.448–0.630).

Image quality assessment

The quality score was higher on CT images (mean = 5) than ZTE (mean = 4.18) and conventional MRI (mean = 4.86) based on a 5-point Likert scale (p = 0.166). ZTE and conventional MRI were vulnerable to motion artifacts due to relative longer scan time compared to MSCT. The osseous morphology was clearly depicted on ZTE-MRI in Figs. 3, 4 and 5.

Fig. 3
figure 3

An exemplary case of normal facet joint on (a) MSCT images, (b) ZTE-MRI and (c) conventional MRI. The cortical bone in ZTE-MRI is depicted well with better tissue contrast than conventional MRI images

Full size image
Fig. 4
figure 4

The osteophyte and sclerosis in L1 and L2 vertebras of a 52-year-old male with back pain are depicted well on the (a) MSCT images, (b) corresponding ZTE-MRI and (c) conventional MRI

Full size image
Fig. 5
figure 5

Comparison of the (a) MSCT images, (b) corresponding ZTE-MRI and (c) conventional MRI on the same patient. The relatively small osteophyte in L3 vertebra as well as the sclerosis in L5 vertebra of a 57-year-old female is clear in the MSCT and ZTE-MRI images

Full size image

Discussion

In this study, the performance of the ZTE-MRI and conventional MRI on the lumbar osseous at different degeneration levels were evaluated using MSCT as image standard. Higher consistency of the measurements and no statistical difference of the lumbar osseous morphometry were found between MSCT and ZTE-MRI. The measurement of the foramina between the MSCT and conventional MRI was statistically different due to dark signal of cortical bone in the conventional MRI images, and the measurement mainly relied on the description of the nerve root and perineural fat.

The assessment of the lumbar osseous morphology with MSCT, ZTE and conventional MRI images was consistent with the previous study [10,11,12,13,14, 30], no difference of qualitative assessment between the MSCT and ZTE images but significant difference of the facet joint OA grade between the MSCT and conventional MRI was found. Compared with CT images, cortical bone appeared blurred on ZTE images especially in severe cases in this study, which may exaggerate the severity assessment of higher-degeneration facet joints OA. Irregularly blurred cortical bone on ZTE images might be caused by the changes of fat content and surrounding soft tissues and thus possibly made the facet joint OA degree overestimated especially at higher degeneration level. The abnormalities diagnosis mainly depended on bone marrow, joint space and effusion on conventional T2WI. The cortical bone in the conventional MRI displayed no signal, leading to missing information of small cortical bone changes at early stage of degeneration. The osteophytes were displayed well in the ZTE images, and conventional MRI was not sensitive to the small osteophytes but displayed the severe osteophytes well.

Osseous morphology is always affected by the degeneration diseases. The evaluation of lumbar osseous morphology could afford essential information for the treatment and prognosis. The reduced capacity of the foramina and lateral canal with nerve compressed would produce leg pain symptom [31]. The axial orientation of facet joints maybe an effective factor of facet joint OA [2]. The osteophyte formation is one of radiographic signs of the vertebral body degeneration [24]. The detection and assessment of osteoarthritis involving the facet joints may aid in the treatment of facet block therapy to patients [22]. Evaluation of the pedicle morphology before the operation is essential for appropriate screw size selection and the accurate insertion of pedicle screws [20].

With the consideration for less medical resource burden and rapid demand for non-invasive diagnosis of the lumbar conditions, no-radiation examination such as advanced MRI techniques for bone assessment and visualization is indispensable. 3D ZTE-MRI imaging sequence arouses interests for the depiction of the osseous structures. With the advantage of less signal loss attributed to instant data acquisition at minimum delay via step-by-step minimal gradient increment, minimal reactive power from gradient amplifiers and easier post-processed than ultra-short echo time (UTE) images [9], 3D ZTE-MRI allowed to view images in arbitrary planes at the maximal flexibility.

Some limitations exist in this study. MRI and MSCT imaging were not carried out on the same day but at a relatively short time interval (the maximal: 11 days). It is unlikely that degenerative osseous changes would present huge progression at a short time. A single ZTE sequence takes longer scanning time (around 5 min) than MSCT and conventional MRI sequence, so the patients in emergent situations were not enrolled in this study. The in-plane resolution of MSCT images, conventional MRI and ZTE images were different to some extent but the resolution was approximately 1 × 1mm2 and sufficient to reveal most lumbar osseous pathologies. Lastly, except for osseous-related observations of lumbar spine that were assessed in the present study, manifestations such as fracture or ligament ossification need to further explore in the future.

In conclusion, ZTE-MRI could be a promising tool to depict the bone morphology compared to conventional MRIs and be an alternative bone diagnosis tool without additional ionizing radiation exposure.