Diagnostic performance of conventional fat-suppression T1WI versus fat-suppression 3D-T1WI-SPACE sequences in detecting glenoid labral injuries via direct shoulder MR arthrography

Abstract

Objective

This study aimed to compare the diagnostic performance of conventional fat-suppression T1-weighted imaging (T1WI) and fat-suppression three-dimensional T1-weighted imaging with sampling perfection with application-optimized contrasts using different flip angle evolution (3D-T1WI-SPACE) sequences in detecting glenoid labral injuries during direct shoulder magnetic resonance arthrography (MRA).

Methods

A retrospective analysis was conducted on imaging data from 60 individuals presenting with shoulder symptoms. Both imaging sequences were performed following intra-articular contrast injection. The intra-articular contrast-to-noise ratio (CNR) of both sequences was calculated to evaluate image quality. The images were analyzed to identify and classify glenoid labral injuries. Diagnositc sensitivity, specificity, and accuracy of the two sequences were calculated using arthroscopy as the reference standard. Agreement between MRA findings (from the two sequences) and arthroscopic results was assessed using the Kappa consistency test.

Results

The fat-suppression 3D-T1WI-SPACE sequence reduced scanning time by 52%. However, its intra-articular CNR was significantly lower than that of the conventional fat-suppression T1WI sequence (p < 0.001). When using arthroscopy as the reference standard, the sensitivity, specificity, and accuracy for diagnosing glenoid labral injury were 97.5%, 95%, and 96.7%, respectively, surpassing those of the conventional fat-suppression T1WI sequence (90.0%, 90.0%, and 90.0%). Despite these differences, the diagnostic performance of both sequences was comparable. The Kappa values indicated high consistency between MRA findings and arthroscopy, with values of 0.925 for the 3D-T1WI-SPACE sequence and 0.78 for the conventional T1WI sequence.

Conclusion

While the fat-suppression 3D-T1WI-SPACE sequence demonstrated inferior image quality compared to the conventional fat-suppression T1WI sequence, its diagnostic accuracy was similar. Its substantially reduced scanning time supports its utility as a supplementary option for evaluating glenoid labral injuries during shoulder joint MRA.

Introduction

Shoulder glenoid labral injury is a common cause of shoulder pain and instability, with direct shoulder magnetic resonance arthrography (MRA) being the most accurate imaging test for diagnosing shoulder glenoid labrum lesions compared to conventional magnetic resonance (MR) imaging [1]. In clinical practice, the T1WI fat suppression fast spin-echo sequence is commonly used for shoulder MRA; however, false-negative results can occur due to partial volume effects from larger scanning layer thickness. The 3D sampling perfection with application optimized contrasts using different flip angle evolution (3D-SPACE) technique is an advanced magnetic resonance imaging (MRI) method that offers isotropic millimeter-level high-resolution 3D fast spin-echo contrast imaging. This technique provides a high signal-to-noise ratio and enhanced post-processing capability for 3D reconstruction, which reduces scanning time while accurately displaying subtle anatomical structures [2–6]. Despite its potential, few studies have compared the diagnostic performance of conventional fat suppression T1WI and fat suppression 3D-T1WI-SPACE sequences in direct shoulder MRA. Thus, this study aims to compare and analyze the signal-to-noise ratios of these two sequences in direct shoulder MRA images and evaluate their diagnostic value for glenoid labral injuries based on shoulder arthroscopy. To avoid diluting the primary end-point, we deliberately confined the reference standard to arthroscopically proven labral lesions and did not score additional shoulder pathologies; rotator-cuff or cartilage assessment was therefore outside the scope of the present protocol.

Data and methods

Study participants

A total of 60 patients who underwent 3 T MRI direct shoulder arthrography between December 2020 and June 2024 were retrospectively analyzed. All patients presented with symptoms of shoulder instability, suspected superior labrum anterior and posterior (SLAP) injuries, or a history of recurrent shoulder dislocation. The cohort included 50 males and 10 females, with an age range of 15 to 66 years and a mean age of 30.1 ± 10.8 years. Among them, 18 patients had left shoulder injuries, while 42 had right shoulder injuries.

All arthroscopies were performed by board-certified orthopaedic surgeons specialised in sports medicine (≥ 10 years of experience); intra-operative labral findings were prospectively recorded on a standardised form and reviewed in consensus by two surgeons before being used as the reference standard.

Arthrography technique

Under fluoroscopic guidance, the radiologist identified a puncture site in the upper space of the glenoid joint and punctured the joint cavity. After confirming that the needle tip was correctly positioned in the joint cavity, approximately 12 to 15 ml of diluted contrast agent (Gadopentetate dimeglumine solution 1:300) was injected into the shoulder joint cavity until the joint capsule was fully dilated. The affected limb was then fully mobilized, and an MRI scan was performed immediately thereafter.

Contrast preparation and concentration considerations

A diluted gadopentetate dimeglumine solution was prepared immediately before each examination by mixing 0.5 ml of the commercially available 469 mg/ml formulation with 149.5 ml of sterile saline, yielding a final concentration of 1:300 (1.56 mmol/L), which falls within the previously reported optimal range of 1:100–1:500 for shoulder MR arthrography. This concentration interval has been shown to maximise T1 signal while avoiding T2*-related signal loss. Although the same dilution protocol was used throughout the study, small batch-to-batch variations in syringe draw-up accuracy, saline volume measurement or incomplete mixing cannot be excluded; such deviations could theoretically alter intra-articular signal intensity and consequently influence the measured contrast-to-noise ratio. Future work should record the exact concentration of each injected dose or use pre-filled standardised vials to further reduce this potential confounder.

MRI image acquisition

A Siemens Skyra 3.0 T MRI scanner and a 16-channel shoulder coil were used to acquire both conventional fat suppression T1WI and fat suppression 3D-T1WI-SPACE images. The detailed scanning parameters for the examination sequences are provided in Table 1. After scanning, multiplanar reconstruction of the fat suppression 3D-T1WI-SPACE images was performed in the transverse and oblique sagittal planes on the post-processing workstation, with a reconstruction layer thickness of 1 mm and no interval.

Table 1.

Scanning parameters for conventional fat suppression T1WI and fat suppression 3D-T1WI-SPACE sequences in shoulder MR arthrography

Parameter	Fat suppression 3D-T1WI-SPACE	Conventional fat suppression T1WI
	Coronal	Axial	Sagittal	Coronal
TR(ms)	700	936	844	687
TE(ms)	11	11	10	10
FA(°)	150	90	90	90
Matrix	256 × 256	256 × 256	256 × 256	256 × 256
FOV (mm)	160	160	160	160
Slice thickness (mm)	0.9	3	3	3
Slice spacing (mm)	0	0	0	0
Bandwidth (HZ)	580	150	140	140
Number of acquisitions	1	2	2	2
Acquisition time (min)	4:39	3:35	3:12	3:00

Quantitative image analysis

A blinded physician performed the procedure on the PACS system. A region of interest (ROI) with an area of 5 to 10 mm² was placed at the same location on both the fat suppression 3D-T1WI-SPACE and conventional fat suppression T1WI coronal images. Signal intensity was assessed by measuring the ROI at the glenohumeral joint space, deltoid muscle, and external background air (Fig. 1) [2]. To avoid volumetric artifacts, the ROI was measured three times at these locations, and the mean value was used. The intra-articular contrast-to-noise ratio (CNR) was calculated using the following formula: [2].

$${\text{CNR}}_{\text{intra}-\text{articular}}=({\text{SI}}_{\text{glenohumeral joint}}-{\text{SI}}_{\text{deltoid muscle}})/{\text{SI}}_{\text{background air}}$$

Fig. 1.

Contrast-to-noise ratio analysis of MRA of the right shoulder in a 33-year-old male patient with shoulder instability. On the left, a conventional fat suppression T1WI sequence coronal image at the level of the axillary fossa recess of the shoulder, with the intra-articular joint fluid, deltoid muscle, and background air set as regions of interest (ROIs) 1, 2, and 3, respectively. On the right, a fat suppression 3D-T1WI-SPACE coronal image at the same level, with the intra-articular joint fluid, deltoid muscle, and background air set as ROIs 4, 5, and 6, respectively

Diagnostic analysis

Radiologic diagnosis was performed independently by two experienced radiologists. If the results were inconsistent, they discussed and agreed upon the findings before comparing them with the arthroscopic results. The main observations included the contour, morphology, and signal of the glenoid labrum, which were used to assess whether the labrum was injured or torn. Tears were categorized as: anterior glenoid labrum tear, posterior glenoid labrum tear, and superior glenoid labrum tear. The criteria for tear injury were as follows [7]: 1. Detection of contrast agent extending into the torn fissure of the glenoid labrum; 2. Significant deformation of the glenoid labrum; 3. Absence of the glenoid labrum; 4. High signal intensity within the glenoid labrum reaching the articular surface; 5. Displacement of the glenoid labrum from its normal anatomical position.

Statistical analysis

SPSS statistical software was used for data processing. Sensitivity, specificity, and accuracy of the conventional fat suppression T1WI sequence and the fat suppression 3D-T1WI-SPACE sequence for diagnosing glenoid labral injury were calculated, using the results of shoulder arthroscopy as the gold standard. A t-test was performed for measurement data comparison, with statistical significance defined as p < 0.05. The chi-squared (X²) test was used for comparison of enumeration data, expressed as cases (%), with significance set at p < 0.05. A Kappa test was performed to assess the consistency of diagnostic results between the conventional fat suppression T1WI and fat suppression 3D-T1WI-SPACE sequences. Consistency was categorized as poor (Kappa < 0.4), moderate (0.4 ≤ Kappa ≤ 0.7), and good (Kappa > 0.7).

Results

Shoulder arthroscopy revealed that 30 of the 60 patients who underwent direct shoulder MRI had isolated superior labral injuries (Fig. 2), 6 had superior labral injuries combined with anterior labral injuries (Fig. 3), 3 had superior labral injuries associated with posterior labral injuries (Fig. 4), and 1 had multiple injuries involving the superior, anterior, and posterior labra.

Fig. 2.

MRA images of the right shoulder in a 41-year-old male patient with shoulder instability. Conventional fat suppression T1WI coronal (a) and sagittal (b) images, along with fat suppression 3D-T1WI-SPACE coronal (c) and reconstructed sagittal (d) images, revealing tears in the superior glenoid labrum (indicated by arrows). Arthroscopy (e) confirmed the presence of superior labrum tears (indicated by arrows)

Fig. 3.

MRA images of the right shoulder in a 36-year-old female patient with recurrent shoulder dislocation. Conventional fat suppression T1WI axial (a) and sagittal (b) images, along with fat suppression 3D-T1WI-SPACE reconstructed axial (c) and sagittal (d) images, demonstrated tears of the anterior glenoid labrum and bony defect (indicated by arrows). Arthroscopy (e) confirmed the presence of the anterior glenoid labrum tear and bony defect (indicated by arrows)

Fig. 4.

MRA images of the right shoulder in a 29-year-old male patient with shoulder instability. Conventional fat suppression T1WI axial (a) and sagittal (b) images, along with fat suppression 3D-T1WI-SPACE reconstructed axial (c) and sagittal (d) images, revealing a posterior glenoid labral tear injury (indicated by arrows). Arthroscopy (e) confirmed the presence of the posterior glenoid labral tear injury (indicated by arrows)

The total imaging time for the fat suppression 3D-T1WI-SPACE sequence was 4 min and 39 s, representing a 52% reduction compared to the 9 min and 47 s required for the conventional fat suppression T1WI sequences.

The mean intra-articular CNR for fat suppression 3D-T1WI-SPACE images was 86.3 (range: 50.5–134.8), while the mean intra-articular CNR for conventional fat suppression T1WI images was 156.5 (range: 105.5–237.5). The mean CNR of conventional fat suppression T1WI images was significantly higher than that of fat suppression 3D-T1WI-SPACE images (p < 0.001; 95% confidence interval).

The diagnosis of glenoid labral injuries using fat suppression 3D-T1WI-SPACE and conventional fat suppression T1WI images, with arthroscopic examination results as the reference, is presented in Table 2. The sensitivity, specificity, and accuracy for diagnosing glenoid labral injuries using both image groups are presented in Table 3. The fat suppression 3D-T1WI-SPACE images demonstrated higher sensitivity, specificity, and accuracy for detecting glenoid labral injuries compared to conventional fat suppression T1WI images; however, the differences were not statistically significant.

Table 2.

Analysis of glenoid labral injury using fat suppression 3D-T1WI-SPACE images versus conventional fat suppression T1WI images (cases)

Examination method		Arthroscopy
		Positive	Negative	Total
Conventional fat suppression T1WI images	Positive	36	2	38
	Negative	4	18	22
	Total	40	20	60
Fat suppression 3D-T1WI-SPACE images	Positive	39	1	40
	Negative	1	19	20
	Total	40	20	60

Table 3.

Sensitivity, specificity, and accuracy (%) of fat suppression 3D-T1WI-SPACE images versus conventional fat suppression T1WI images for the diagnosis of glenoid labral injuries

Examination method	Sensitivity	Specificity	Accuracy
Conventional fat suppression T1WI images	90.0%	90.0%	90.0%
Fat suppression 3D-T1WI-SPACE images	97.5%	95%	96.7%
X2	2.047	0.367	2.236
P	0.152	0.545	0.135

The consistency of diagnostic results for glenoid labral injuries between fat suppression 3D-T1WI-SPACE images, conventional fat suppression T1WI images, and arthroscopy was found to be good after the consistency test (Kappa values were 0.925 and 0.78, respectively), with fat suppression 3D-T1WI-SPACE images demonstrating a higher degree of consistency.

Discussion

Principal findings

This study demonstrated that the image quality of conventional fat suppression T1WI images was slightly superior to that of fat suppression 3D-T1WI-SPACE images. However, no significant difference was observed between the two sequences in diagnosing glenoid labral injuries of the shoulder, and both exhibited good consistency with arthroscopic results. Furthermore, the total scanning time of the fat suppression 3D-T1WI-SPACE sequence was shorter than that of the conventional fat suppression T1WI sequence.

Technical considerations and clinical relevance

Shoulder MRI is prone to motion artifacts, which can lead to false-negative or false-positive diagnoses, highlighting the importance of a short acquisition time in clinical practice [8, 9]. Conventional fat suppression T1WI sequences have long imaging times, and patients who cannot tolerate prolonged procedures, particularly the older adults, are at higher risk of image artifacts that can compromise the diagnosis of glenoid labral injuries. In this study, the fat suppression 3D-T1WI-SPACE sequence was employed, which reduces the SAR value by lowering the flip angle of the reconstruction pulse, minimizes the echo gap through non-planar selective radiofrequency pulses, increases the echo chain length to enhance acquisition efficiency, and optimizes K-space filling, significantly shortening the scanning time [10]. Furthermore, the 3D imaging technique minimizes the partial volume effect, facilitating the visualization of small structures such as the glenoid labrum and enhancing the diagnosis of glenoid labral injury [11]. Although the quality of fat suppression 3D-T1WI-SPACE images was slightly lower than that of conventional fat suppression T1WI images, the scanning time was reduced by nearly 50%, while still maintaining adequate image quality, making it suitable for MRA examination in patients who cannot tolerate long scanning durations.

The 3D-T1WI-SPACE sequence provides isotropic data, ensuring that image resolution remains unaffected by processing in any plane, whether standard or non-standard. A recent study demonstrated that 3D isotropic sequences offered high diagnostic performance for glenoid labral tears in the shoulder joint, with minimal difference between conventional sequence MRA and 3D isotropic sequences [12]. Previous studies compared the diagnostic performance of 3D isotropic imaging sequences with conventional sequences in patients with glenoid labral injuries, mainly focusing on SLAP and Bankart injuries, while the rarer posterior glenoid labral injuries were not examined [13–15]. In this study, glenoid labral injuries were categorized into anterior, posterior, and superior labral tears, with three cases of posterior glenoid labral injury recorded. Furthermore, the diagnostic efficacy of fat suppression 3D-T1WI-SPACE images for glenoid labral injuries was found to be similar to that of conventional fat suppression T1WI images, with almost identical diagnostic results. When compared to arthroscopic diagnoses, the 3D isotropic sequence demonstrated higher consistency than the conventional sequence MRA.

Limitations and future directions

This study has several limitations. First, the total sample size was insufficient, and the study was conducted at a single institution, which may limit the generalizability of the findings. Second, the cohort was demographically skewed, with a marked male predominance (50 males vs. 10 females), potentially reducing the applicability of the results to broader or more diverse populations. Third, an objective analysis of the image quality of individual anatomical structures, such as the glenoid labrum, ligaments, and cartilage, was not performed. Fourth, although arthroscopy was conducted by experienced sports-medicine surgeons in consensus, the lack of inter-observer statistics for the surgical findings and the absence of video recording for external review may still leave reference-standard uncertainty. Recording surgical videos and calculating surgeon-vs-surgeon agreement would further strengthen the gold standard. Additionally, the quantitative CNR was affected by potential variations in the manually prepared contrast dilution, as discussed in the Methods section. Moreover, the diagnostic efficacy of the fat suppression 3D-T1WI-SPACE sequence and conventional fat suppression T1WI sequence for rotator cuff tendon tears was not assessed. Furthermore, complete blinding for the sequences was not achieved. Finally, the potential advantage of 3D isotropy for glenoid labral injuries, in comparison with the conventional fat suppression T1WI sequence, was not evaluated.

Future multicenter studies with larger and more diverse populations are warranted to validate the diagnostic performance of the 3D-T1WI-SPACE sequence across different clinical setting.

Spectrum and prevalence bias

Our cohort consisted of patients with clinically suspected labral lesions (shoulder instability or SLAP suspicion); consequently, the prevalence of labral tears was high (≈80%). This enriched spectrum may inflate sensitivity and mask specificity compared with an unselected shoulder-pain population. Future prospective studies should enrol consecutive patients with broader indications (e.g., non-specific shoulder pain, rotator-cuff suspicion) to estimate test performance under more realistic prevalence conditions.s and demographic groups.

Exclusion of other shoulder structures

This trial was designed as a head-to-head comparison of two T1-weighted sequences for labral tear detection; rotator-cuff tendons, cartilage, and ligaments were not scored. Consequently, we cannot infer whether the shorter 3D-T1WI-SPACE acquisition could also serve as a “one-stop” replacement for the conventional combination of T1-, T2- and intermediate-weighted images when the clinical question is broader. Future work should therefore test the 3D sequence in two specific scenarios: a “pan-shoulder” screening setting in patients with ambiguous pain, where labrum, cuff, and cartilage must be evaluated simultaneously; and a “concomitant-lesion” setting where labral tears are already suspected and co-existing partial-thickness cuff or ligamentous injuries need to be ruled-in or ruled-out before arthroscopic planning.

Observer bias and future blinding strategies

Because ROI placement and tear grading were performed by readers who were aware of imaging sequence and, in many cases, clinical history, subconscious expectations may have influenced CNR calculation and the higher sensitivity/specificity observed for 3D-T1WI-SPACE. For example, readers might have selected more homogeneous fluid voxels when measuring 3D images if they presumed lower CNR, or they may have over-called subtle labral changes when they believed the 3D sequence was “technically superior”. Such biases could artificially elevate the Kappa agreement with arthroscopy and mask true inter-sequence performance differences.

Future work should: Strip DICOM headers of sequence names, flip angles and voxel size before analysis; employ independent technologists (not involved in diagnosis) to place ROIs after standardized window/level adjustment; use a multi-reader, multi-stage design in which radiologists first interpret images without clinical data, then re-interpret with data, allowing quantification of history-related bias; randomise case and sequence order, incorporate wash-out intervals, and invite external readers to reduce institutional preference; report bias risks with tools such as QUADAS-2. These measures will separate genuine sequence advantages from reader-dependent effects and produce more generalisable accuracy estimates.

To more robustly establish the clinical utility and diagnostic equivalency of 3D-T1WI-SPACE, prospective longitudinal cohorts that serially enrol patients with varying ages, sexes, and primary shoulder complaints are warranted; such studies should record long-term arthroscopic or conservative outcomes to determine whether early 3D-T1WI-SPACE diagnosis truly alters management and improves prognosis across different indications.

Conclusion

While the quantitative image quality of the fat suppression 3D-T1WI-SPACE sequence was inferior to that of the conventional fat suppression T1WI sequence, the two sequences exhibited comparable diagnostic performance in identifying glenoid labral tear lesions. Notably, the fat suppression 3D-T1WI-SPACE sequence significantly reduced scanning time, decreasing the total duration by nearly 50% without compromising diagnostic accuracy. Consequently, this sequence offers a practical option for MRA of the shoulder joint, particularly for individuals who may have difficulty tolerating extended scan durations.

Author contributions

Conception and design of the research: Binbin Teng Acquisition of data: Hang Wu, He Lin, Binbin Teng Analysis and interpretation of the data: Binbin Teng, He Lin Statistical analysis: Binbin Teng, Hang Wu Obtaining financing: Binbin Teng Writing of the manuscript: Binbin Teng, Hang Wu Critical revision of the manuscript for intellectual content: Binbin Teng, Hang Wu All authors read and approved the final draft.

Funding

This work was supported by Fuzhou Science and Technology Plan Projects (No.2022-S-021) and Funding for the Construction Project of Key Clinical Specialties in Fuzhou Municipality (No.20220302).

Data availability

The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted with approval from the Ethics Committee of Fuzhou Second General Hospital (No.2022085). This study was conducted in accordance with the declaration of Helsinki. Written informed consent was obtained from all participants.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.