The present study was conducted following the Declaration of Helsinki (as revised in 2013) and approved by the Ethics Committee of our hospital (IRB No. 201902068). This is a secondary analysis of a prospective study, and some patients with chronic low back pain underwent MRI and/or CT scans at the recommendation of surgeons. With full communication and informed consent, some patients who underwent lumbar MRI and CT scanning were initially included in the study. Inclusion criteria were as follows: ① age should be ≥ 50 years; ② lumbar DECT and MRI scan within 48 h. Exclusion criteria included the following points: ① scoliosis; ② localized osteosclerosis in vertebral cancellous bone; ③ vertebral trauma and tumor; ④ postoperative state of lumbar vertebra. The flowchart displaying patient inclusion of this study is shown in Fig. 1.

The flowchart of the study. DECT Dual-energy computed tomography, QCT quantitative computed tomography, MRI magnetic resonance imaging, FF fat fraction, BMD bone mineral density

A total of 83 patients with chronic low back pain from April to November in 2021 were enrolled, including 30 males and 53 females. By placing a standard QCT corrected phantom (QCT Pro v5.0; Mindways, Tex) under the waist during DECT procedure, we obtained DECT parameters and QCT-based bone mineral density (BMD) concurrently. Thus, no additional radiation exposure was entailed.

DECT examinations relied on a second-generation 128-section dual-source unit operating in dual-energy mode (Somatom Definition Flash; Siemens Healthineers, Erlangen, Germany). Settings of both x-ray tubes were constant (tube A: 80 kV, 250 mAs; tube B: 140 kV with Sn filter, 97 mAs), pitch of 0.6, collimation width 32 × 0.6 mm, rotation time 500 ms/r, field of view (FOV) 500 mm × 500 mm. The scanning range extended from the 12th thoracic vertebra to the 1st sacral vertebra. Images were reconstructed using a kernel of I30f, 1-mm section thickness, and 0.75-mm increment. All radiation doses received by patients were recorded upon completion.

The quantitative measurement of the vertebral BMAT was carried out based on the liver virtual non-contrast function module of the dual-energy analytic software (Syngo. via VB10; Siemens Healthcare, Erlangen, Germany). According to the parameter settings in bone marrow analysis, the default value of soft tissue was modified to 55 HU and 51 HU, fat value to − 110 HU and − 87 HU, and iodine slope to 1.71 [16]. The FF values of 1st to 5th lumbar vertebrae were measured in the median sagittal view, and the mean FF value of each patient was recorded as DECT-FF (Fig. 2a).

Measurement of DECT-FF, MRI-FF, and BMD: A, B DECT-FF and MRI-FF derived from corresponding region of interest (ROI) in a standard median sagittal plane, respectively, and the ROI was delineated in 2/3 of the anterior vertebral body, avoiding the bone cortex, vertebral vein sulcus, and surrounding osteosclerosis; C BMD determined by ROI automatically drawn with QCT analytics system

MRI examinations relied on a 3.0 T superconducting MR scanner (Discovery 750, GE Healthcare, Milwaukee, WI, USA), with standard human body coil and sagittal scanning. Prior to IDEAL-IQ, T1 weighted image (T1WI) (repetition time [TR]/ time to echo [TE] = 400/13 ms), T2WI (TR/TE = 2500/102 ms), FOV 36 cm × 36 cm, matrix of 224 × 192, pixel size 1.6 mm × 1.9 mm, slice thickness of 3 mm, intersection gap of 0.4, number of excitations (NEX) of 1; IDEAL-IQ: TR of 7.4 ms, minimum TE of 1.3 ms, maximum TE of 5.3 ms, flip angle of 4°, echo train length of 5, bandwidth of 111.1 kHz, and other settings were the same as above.

Four group images were obtained once the IDEAL-IQ sequence was scanned: pure water image, pure fat image, fat fraction image, and R2* relaxation rate image. The FF value was measured on the viewer module of ADW4.7 workstation. Select the fat fraction image, draw a rectangular ROI on the first 2 / 3 of the vertebral body in the median sagittal diagram, then the FF values of 1st to 5th lumbar vertebrae were measured successively at one slice (Fig. 2b), the mean FF value of each patient was recorded as MRI-FF.

DECT equipment and QCT analytics (QCT Pro v5.0; Mindways, Tex) were calibrated in advance using a quality control phantom. The scan data of mixed ratio (0.5) were imported to the QCT software application, and the BMD values of 1st to 5th lumbar vertebrae were measured by drawing a region of interest automatically (Fig. 2c). Then, the mean BMD value of 5 vertebrae was taken as the patient's BMD. According to BMD, all patients were divided into normal group (BMD > 120 mg/cm3), osteopenia group (120 mg/cm3 ≥ BMD > 80 mg/cm3) and OP group (BMD ≤ 80 mg/cm3) [17].

All vertebral measurements were performed independently by 2 experienced radiologists, who had been engaged in musculoskeletal imaging research for more than 8 years, and take the mean value of 2 measurers as the final value. The consistency analyses for the measurements of 2 measurers were performed by intraclass correlation coefficient (ICC).

Statistical analysis was performed using Medcalc (version 19.0 MedCalc Software bvba, Belgium). Variables were tested for normality of distribution by Shapiro–Wilk test, and expressed as means ± SDs. The difference between normal, osteopenia and OP groups and between 1st and 5th vertebra were tested by one-way ANOVA, and the difference of FF values between adjacent vertebrae was tested by independent sample t test. The consistency between DECT-FF and MRI-FF was analyzed by ICC and Bland–Altman method (ICC of < 0.4 means poor consistency, ICC of 0.4–0.75 means general consistency, ICC of > 0.75 means good consistency); the Pearson correlation analysis was used to analyze the correlation between DECT-FF, MRI-FF, and BMD (|r| of < 0.5 means low linear relationship, |r| of 0.5–0.8 means a significant linear relationship, and |r| of > 0.8 means a highly linear relationship). Taking BMD as the gold standard, DECT-FF and MRI-FF diagnostic efficacy in different OP degrees were evaluated by receiver operating characteristic (ROC) curve, and the area under the curve (AUC) of DECT-FF and MRI-FF were compared by the DeLong test (AUC of 0.5–0.7 means low diagnostic value, AUC of 0.7–0.9 means moderate diagnostic value, and AUC of > 0.9 means high diagnostic value). A p value of < 0.05 was statistically significant.