Three-dimensional static-fluid MR urography with gradient- and spin-echo (GRASE) at 3.0T: comparison of image quality and diagnostic performance with respiratory-triggered fast spin-echo (FSE)

Patients

This prospective, single-institution study was approved by the ethics committee of our hospital. Written informed consent was obtained from all patients. We continuously recruited patients referred to our MR department for diagnosing, evaluating, or monitoring urinary tract dilation by MRU from January to May 2021. Inclusion criteria were as follows: (1) no contraindications for MRI examination, (2) completion of the two types of MRU techniques described below, (3) no history of urinary tract surgery, (4) visible urinary tract dilation.

Magnetic resonance imaging sequences and parameters

All MRU examinations were performed on a 3T MR scanner (uMR 790, United Imaging Healthcare, Shanghai, China) using two pieces of 12-channel body matrix coil combined with a 32-channel spine matrix coil. Patients were asked to fast for 6 h and hold urine for 1 h to distend the bladder before the examination. Before the MR examination, routine respiratory training was conducted for every patient including regular breathing and breath-holding.

Imaging protocols included axial T1-weighted sequence, axial and coronal T2-weighted sequences, and axial diffusion-weighted imaging sequence. The 3D MRU sequences were acquired in the coronal plane and have two protocols: (1) BH-GRASE and (2) RT-FSE. The order of the two 3D MRU protocols occurred in a random order, covering the same volume.

3D BH-GRASE MRU was obtained using the GRASE sequence. Acquisition parameters were as follows: repetition time (TR),1300 ms; echo time (TE), 267 ms; flip angle (FA), 180°; parallel acquisition factor, 3; FSE factor, 35; EPI factor, 7; field of view (FOV), 400 × 400 mm; matrix, 213 × 304; resolution (reconstruction) 1.88 × 1.32 × 3 (1.25 × 0.88 × 1.5) mm; slice number, 28; acquisition time, 14.8 s.

Conventional 3D RT-FSE MRU was performed with a FSE sequence. TR/TE, variable depending on respiratory/697 ms; FA, 110°; parallel acquisition factor, 2.7; FSE factor, 205; FOV, 400 × 400 mm; matrix, 285 × 352; resolution (reconstruction) 1.42 × 1.14 × 2 (0.95 × 0.76 × 1) mm; slice number, 42; acquisition time, variable (range 132–392 s).

If the patients could not cooperate with the breath instruction well, we did not repeat the sequence. Both were reconstructed using the maximum intensity projection (MIP) algorithm on the satellite console of the MR unit. The actual acquisition time of both 3D MRU protocols was recorded.

Image analysis

Three radiologists [Reader A (**), B (**), and C (**), with 9, 11, and 14 years of clinical experience in genitourinary MR imaging, respectively] independently reviewed the FSE MRU and GRASE MRU images on the picture archiving and communication system (PACS) with a 2-week interval to minimize recall bias. The source and MIP images were anonymized and arranged randomly without information on acquisition methods. The readers were free to adjust the window settings according to the reader's experience. The flowchart of the study is demonstrated in Fig. 2.

Fig. 2figure 2

The flowchart of the study. BH-GRASE breath-hold gradient- and spin-echo, RT-FSE respiratory-triggered fast spin-echo

Qualitative image analysis

Reader A and B were asked to grade the overall image quality, artifacts, and urinary tract visualization (renal calyces, renal pelvis, ureter on each side as well as bladder) on a 5-point grading scale. The overall image quality was graded as follows: 1, non-diagnostic; 2, major artifact and main duct only partially visible; 3, moderate artifact with blurring or partial lack of duct visualization; 4, slight artifact without loss of diagnostic value; 5, no detectable artifact and excellent duct visualization. The image artifact assessment was divided into two parts: (1) image blur due to respiratory artifacts; and (2) image distortions or local signal change caused by susceptibility artifacts. The two types of artifacts were separately graded as follows: 1, severe artifacts, image not diagnostic; 2, major artifacts, significantly decreased diagnostic ability; 3, moderate artifacts, partially affecting diagnosis; 4, minor artifacts, without affecting diagnosis; 5, no artifacts. The urinary tract visualization was graded as follows: 1, non-visualization; 2, poor visualization with limited diagnostic value; 3, partially visualized; 4, near-complete visualization; 5, excellent clear visualization.

With three readers being blinded to the previous results of qualitative analysis, the two MRU sets (BH-GRASE and RT-FSE) with source and MIP images were reviewed for side-by-side comparison in blind randomized order. Three readers were asked to rank their preference based on diagnostic image quality and confidence: equally prefer both, or prefer set 1, or 2.

Quantitative image analysis

Two readers (A and C) performed quantitative analysis of the source images of two MRU sequences, with interobserver agreement assessed afterward. To measure the signal intensities (SI), representative slices that depicted the largest area of the dilated renal pelvis, dilated ureter, and bladder were selected, and their SI was measured by applying regions of interest (ROIs) in these slices. ROIs were placed in homogeneous, artifact-free areas of the renal pelvis, ureter, and bladder as well as homogeneous, artifact-free areas adjacent to the pelvis, ureter, and bladder at the same slice, respectively. For the measurement of SIs of the adjacent area, the ROI was placed avoiding other fluid-containing structures. The relative contrast ratio (CR) was selected as a quantitative index to reflect the contrast between the urinary tract (U) and adjacent area (A) [18]. The CR was selected instead of the common CNR calculation based on image noise, as the heterogeneous signal intensity of the background was unreliable and difficult to define [19]. The CR value was calculated using the following formula:

$$} = (}_}} - }_}} ) \, / \, \left( }_}} + }_}} } \right)$$

where SIU and SIA stand for the signal intensities of the urinary tract and adjacent area, respectively. If bilateral dilation of the renal pelvis or ureter was observed, the CR values of both sides were measured and calculated. The final CR values of the renal pelvis or ureter were the mean CR values of both dilated sides.

Performance evaluation

Two readers (B and C) separately assessed the degree of dilation, obstructive side (unilateral or bilateral), obstructive level (ureteropelvic junction, abdominal ureter, pelvic ureter, intravesical ureter) and defined the imaging features of the obstructive site as benign (smooth, gradual tapered) or malignant (abrupt, irregular cut off) on both MRU protocols. The degree of dilation was defined as follows: mild (pelvis dilation alone), moderate (with mild calyceal dilation), or severe (with severe calyceal dilation) [20].

The reference standard was confirmed by the pathological results of the endoscope biopsy or surgery. For the patients suspicious of benign etiologies whose pathological results were unavailable, a diagnosis of benign obstruction was confirmed by clinical data or laboratory examinations.

Statistical analysis

All descriptive data are described as the means ± standard deviation (SD). Differences in the acquisition time and CR value between BH-GRASE and RT-FSE were analyzed using paired t-test or Wilcoxon signed-rank test after the normality test. A Wilcoxon signed-rank test was used to evaluate the differences in the qualitative scores of overall imaging quality, artifacts, and structure visualization between the two MRU techniques. A two one-sided test of equivalence (TOST) based on the Wilcoxon signed-rank test was performed to assess the equivalence of qualitative scores of the two MRU images, using a ± 0.5 equivalence region [21, 22]. Kappa statistics or intraclass correlation coefficient (ICC) were calculated (1) to measure the interobserver agreement between two readers; and (2) to evaluate the inter-sequence consistency for performance ability between BH-GRASE and RT-FSE. A value of less than 0.20 was considered as disagreement; 0.21–0.40, poor agreement; 0.41–0.60, moderate agreement; 0.61–0.80, good agreement; and over 0.80, excellent agreement. Receiver operating characteristic (ROC) analysis was used to explore the diagnostic performance of the two MRU sequences in differentiating malignant from benign dilation. Sensitivity, specificity, and accuracy were calculated by 2 × 2 contingency tables. Statistical analyses were performed using SPSS software (version 26; IBM) or NCSS software (version 12). p < 0.05 was considered to indicate statistical significance.

留言 (0)

沒有登入
gif