The role of scapular morphological structure in degenerative rotator cuff injury: a digital reconstruction radiograph study

Abstract

Objective

To investigate the relationships among the acromial tilt (AT), the glenoid inclination angle (GIA), the critical shoulder angle (CSA), the acromioglenoid angle (AGA), and degenerative rotator cuff injury via digital reconstruction radiographs (DRRs).

Methods

A retrospective cohort of 63 patients who met the inclusion criteria were enrolled from the Affiliated Hospital of Chengdu University of Traditional Chinese Medicine between April 2023 and April 2024, 33 patients with rotator cuff injury and 30 controls were included. Clinical data and scapular computed tomography (CT) scans were collected, and three-dimensional reconstructions were generated via Amira software (Thermo Fisher Scientific, USA). Digital reconstruction radiograph (DRR) images were simulated at the Grashey and Y positions of the scapula, and the parameters were measured via Digimizer software (MedCalc Software, Belgium). The diagnostic and predictive validity of each parameter for DRR for rotator cuff injury was assessed through statistical analysis of the measured data, including inter-observer consistency, correlation analysis among parameters, and receiver operating characteristic (ROC) curve analysis.

Results

(1) Inter-observer reliability was excellent for all the parameters (intra-class correlation coefficient (ICC) > 0.8). (2) CSA and AGA were significantly greater in the rotator cuff injury group (P_CSA= 0.003;P_AGA < 0.001), whereas GIA was significantly lower (P < 0.001); no difference was detected in AT (P > 0.05). (3) Binary logistic regression analysis revealed that GIA, CSA, and AGA were risk factors for rotator cuff injury (P_CSA=0.008; P_AGA = 0.001; P_GIA = 0.001). (4) ROC analysis yielded area under the curve (AUC) values of 0.755 (GIA), 0.700 (CSA), and 0.796 (AGA), with cutoff thresholds of 83.98°, 30.67°, and 42.04°, respectively. (5) Strong correlations were observed between the CSA and AGA (r = 0.939), whereas the GIA was inversely correlated with both the CSA (r = -0.423) and the AGA (r = -0.471).

Conclusion

Measurements of GIA, CSA, and AGA derived from DRRs demonstrate high diagnostic value for degenerative rotator cuff injury. DRR represents a reliable alternative to conventional radiography in clinical practice. In the future, DRR holds promise as a tool for assessing the impact of degenerative rotator cuff injury. However, due to radiation exposure concerns, the current application of DRR is recommended primarily for retrospective studies utilizing existing computed tomography (CT) data.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12891-025-09086-w.

Introduction

Rotator cuff injury is a leading cause of shoulder pain and functional impairment. With the aging global population, the annual incidence of rotator cuff injury has risen significantly [1], particularly among individuals aged ≥ 60 years, where the prevalence exceeds 25% [2], and among those over 75 years of age, the incidence exceeds 40% [3]. While trauma contributes to acute cases, degenerative changes associated with scapular morphology are increasingly implicated in chronic pathology [4]. In 1972, Charles S. Neer et al. [5] proposed the concept of subacromial impingement for the first time and proposed that mechanical compression of the rotator cuff by the coracoacromial ligament and the anterolateral edge of the acromion underlies injury pathogenesis. Subsequent research has expanded this framework, identifying morphological parameters such as acromial type [6], AT [7], GIA [8], CSA [9], and AGA [10] as potential diagnostic markers. However, accurate measurement of these parameters traditionally relies on standardized anterior‒posterior view (Grashey-view) radiographs [11], which require precise patient positioning (the X-ray projection direction is approximately 35‒45° to the chest, the anterior‒posterior edge of the glenoid completely overlaps to form a straight line, and the glenoid is exposed without teardrop morphology). The accurate measurement of bony landmarks of the shoulder joint depends on the spatial relationship between the scapula and the humerus relative to X-rays [12]. Clinically, obtaining standardized Grashey-view radiographs is often hindered by technical and patient-related limitations, and standard-position X-ray films are extremely rare [13]. One study showed that DRRs can be used as a substitute for clinical radiography [14]. DRR derived from CT-based 3D models. By generating a 3D model and simulating X-ray irradiation on the 3D model, a radiological imaging projection is formed to obtain radiographs with any desired projection angle. It has no overly high requirements for the professional level of radiographic technicians and avoids multiple radiation exposures of patients to obtain X-ray films at the ideal position, increasing the cooperation and acceptability of patients. The reconstructed images can be standard anteroposterior (AP) positions to correct errors or nonstandard AP positions with a specific deflection angle to study the influence of these deflection position errors on the efficiency of the measurement results [14]– [15]. DRR technology provides a solution to the difficulty in obtaining standard AP position X-ray films in clinical practice. This study aimed to validate DRRs as diagnostic tool for assessing changes in the morphological and structural parameters of the scapula (AT, GIA, CSA, and AGA) and its association with degenerative rotator cuff injury.

Methods

Patient selection

This retrospective study analyzed scapular CT images and clinical data from 63 patients treated at the Affiliated Hospital of Chengdu University of Traditional Chinese Medicine between April 2023 and April 2024. Among the patients, 33 had rotator cuff injuries (average age 56.93 ± 7.91 years, male/female: 12/21), and 30 were controls (average age 58.89 ± 9.85 years, male/female: 14/16).

The inclusion criteria for the rotator cuff injury group were as follows: (1) age ≥ 18 years, (2) no history of trauma, (3) magnetic resonance imaging (MRI)-confirmed rotator cuff injury [16], and (4) availability of scapular CT scans with sufficient resolution.

The exclusion criteria for patients in the rotator cuff injury group were as follows: (1) had traumatic rotator cuff injury, (2) had other shoulder diseases other than rotator cuff injuries in the past or during treatment, or (3) had CT and MRI examinations that lasted more than 3 months.

The inclusion criteria for the control group were as follows: (1) age ≥ 18 years, (2) no history of shoulder trauma or surgery, (3) no shoulder pain symptoms, (4) availability of scapular CT scans with sufficient resolution, (5) come to our hospital for other conditions unrelated to shoulder joint disease (such as cold, pneumonia, etc.).

The exclusion criteria for the control group were as follows: (1) patients with shoulder pain, (2) patients who had shoulder diseases in the past or during treatment.

Sampling method: (1) Firstly, 33 patients with rotator cuff injuries who met the inclusion criteria and visited our hospital from April 2023 to April 2024 were screened according to the inclusion criteria. (2) Then, according to the inclusion criteria of the control group, patients with non-shoulder joint diseases who visited our hospital from April 2023 to April 2024 were screened out and randomly numbered. (3) The two groups of patients were stratified by age (divided into five layers: 18–30 years old, 31–40 years old, 41–50 years old, 51–60 years old, and over 60 years old). According to the approximate proportion of the rotator cuff injury group (0: 3: 7: 13: 10), stratified random sampling was conducted on the control group at the corresponding proportion. Finally, 30 control patients (0: 3: 4: 14: 9) who met the criteria were selected.

The study protocol (Ethics No. 2024KL-026, Ethics Committee of the Affiliated Hospital of Chengdu University of Traditional Chinese Medicine) adhered to institutional guidelines for retrospective research. This was a retrospective study on the digital reconstruction of scapular CT images taken in the past, and informed consent could not be obtained in advance. Moreover, in this study, only reconstructed CT images were used, which does not harm patients and does not involve the disclosure of patient privacy. Therefore, with the approval of the Ethics Committee of the Affiliated Hospital of Chengdu University of Traditional Chinese Medicine, patients’ informed consent can be exempt.

Imaging protocols

CT imaging was performed via a GE Discovery CT750 HD scanner (GE Healthcare, USA) at the Department of Radiology, Affiliated Hospital of Chengdu University of Traditional Chinese Medicine, with parameters set to 130 kV, 220 mA, and 5 mm slice thickness. MRI examinations utilized a GE Discovery 3.0T system (GE Healthcare, USA) with T1-weighted imaging (T1WI), T2-weighted imaging (T2WI), and proton density-weighted imaging (PDWI) sequences.

DRR reconstruction and measurement

Digital Imaging and Communications in Medicine (DICOM)-formatted CT data were imported into Amira 2019 (Thermo Fisher Scientific, USA) for 3D scapular reconstruction. DRR images simulating the Grashey and Y positions were generated, and osseous parameters (AT, GIA, CSA, AGA) were measured by two blinded observers (two orthopedic doctors) via Digimizer software (MedCalc Software, Belgium). A third investigator (orthopedic doctors) collected and recorded the data and evaluated it.

When there were objections, discrepancies were resolved by a senior radiologist. The measurement methods for all the parameters are illustrated in Figs. 1, 2 and 3, and 4.

Fig. 1.

AT: Y-view radiograph of the scapula showing the angle formed by the intersection of the line connecting the most anterior-lateral and posterior-lateral edges of the acromion with the horizontal line of the glenoid

Fig. 2.

GIA: Angle formed when the line connecting the upper and lower glenoid border intersects the line where the bottom margin of the supraspinatus fossa is located

Fig. 3.

CSA: Angle between the line connecting the upper and lower margins of the glenoid cavity and the line connecting the bottom margin of the glenoid cavity to the most lateral margin of the acromion

Fig. 4.

AGA: Angle formed by the line between the upper and lower glenoid margins and the most lateral point of the acromion passing through the midpoint of the line between the upper and lower glenoid margins

Statistical methods

Statistical analyses were conducted via SPSS 27.0 (IBM Corp., USA). Continuous variables are expressed as the mean ± standard deviation (SD). Inter-observer reliability was assessed via the ICC, and strong consistency among surveyors was demonstrated when the ICC was ≥ 0.80. For group comparisons, independent t tests or Mann‒Whitney U tests were used. Binary logistic regression and ROC curve analysis were used to evaluate diagnostic performance; when AUC > 0.7, the parameter had good diagnostic performance. Pearson correlation coefficients were used to quantify parameter associations. Significance was set at P < 0.05.

Results

Basic characteristics

Patient images were screened strictly according to the inclusion criteria. A total of 63 patients whose scapular CT images were available were included. There were 33 patients with rotator cuff injuries, including 12 males and 21 females. All the patients were Chinese (mainly from the southwest region of China); that is, they were all of East Asian ethnicity. The age distribution was normal (P > 0.05), with an average age of 56.93 ± 7.91 years. In the control group, there were 30 patients without rotator cuff injury, including 14 males and 16 females, with an average age of 58.89 ± 9.85 years, and the age distribution conformed to a normal distribution (P > 0.05). There was no significant difference in age or sex between the two groups (P > 0.05) (Table 1).

Table 1.

Basic characteristics of the two groups

	Rotator cuff injury group(n = 33)	Control group(n = 30)	t/χ2	P
Age (Mean ± SD, years old)	56.93 ± 7.91	58.89 ± 9.85	−0.877	0.384
Sex (Male/Female, n)	12/21	14/16	0.688	0.407

SD standard deviation

Inter-observer reliability

The ICC indicated excellent inter-observer agreement for all the parameters (ICC > 0.8; 95% confidence interval (CI) range: 0.932–0.979), confirming the reproducibility of the DRR-based measurements (Table 2).

Table 2.

Intra-observer reliability

	CSA	AGA	AT	GIA
ICC (95% CI)	0.961 (0.936, 0.976)	0.960(0.932, 0.976)	0.965(0.943, 0.979)	0.962(0.939, 0.977)

AT acromial tilt, CSA critical shoulder angle, AGA acromioglenoid angle, GIA glenoid inclination angle

Group comparisons

Compared with those of the controls, the rotator cuff injury groups’ CSA (31.65 ± 3.71° vs. 29.50 ± 1.35°, P = 0.003) and AGA (43.52 ± 4.83° vs. 39.20 ± 2.05°, P < 0.001) significantly increased, whereas the GIA significantly decreased (79.24 ± 5.67° vs. 84.36 ± 4.53°, P < 0.001). No intergroup difference was detected in AT (40.18 ± 5.76° vs. 42.32 ± 4.78°, P = 0.115) (Table 3). Patients with degenerative rotator cuff injury were stratified into three subgroups according to injury severity: early injury (n = 14), partial tear (n = 15), and full-thickness tear or rupture (n = 4). All measured parameters demonstrated statistically significant differences across these subgroups (P < 0.001; see Supplementary Table 1).

Table 3.

Comparison of the measured values of each parameter between the two groups

	Rotator cuff injury group(n = 33)	Control group(n = 30)	t	P
CSA	31.65 ± 3.71	29.50 ± 1.35	3.113	0.003*
AGA	43.52 ± 4.83	39.20 ± 2.05	4.690	< 0.001*
AT	40.18 ± 5.76	42.32 ± 4.78	−1.597	0.115
GIA	79.24 ± 5.67	84.36 ± 4.53	−3.933	< 0.001*

*:significant difference between groups (P < 0.05)

AT acromial tilt, CSA critical shoulder angle, AGA acromioglenoid angle, GIA glenoid inclination angle

Risk factor analysis

Binary logistic regression identified GIA (odds ratio (OR) = 0.804, 95% CI (0.705, 0.917)), CSA (OR = 1.332, 95% CI (1.077, 1.648)), and AGA (OR = 1.404, 95% CI (1.155, 1.708)) as risk factors for degenerative rotator cuff injury (P < 0.01) (Table 4). According to the Beta and OR, the probability of rotator cuff injury increased by 0.332 and 0.404 times for every 1° increase in the CSA and AGA measurements, respectively, and 0.196 times for every 1° decrease in the GIA measurements.

Table 4.

Binary logistic regression analysis of the relationships between each parameter and rotator cuff injury

	Beta	P	OR	95% CI
CSA	0.287	0.008*	1.332	(1.077, 1.648)
AGA	0.340	0.001*	1.404	(1.155, 1.708)
GIA	−0.218	0.001*	0.804	(0.705, 0.917)

*: significant difference between groups (P < 0.05)

CSA critical shoulder angle, AGA acromioglenoid angle, GIA glenoid inclination angle.

Receiver operating characteristic (ROC) curve analysis

ROC curve analysis demonstrated moderate to high diagnostic accuracy for GIA (AUC = 0.755, cutoff = 83.98°), CSA (AUC = 0.700, cutoff = 30.67°), and AGA (AUC = 0.796, cutoff = 42.04°) measured on DRR, with the AGA exhibiting the highest predictive performance (Table 5 and Fig. 5). According to the analysis results, patients with a CSA > 30.67°, an AGA > 42.04° or a GIA < 83.98° measured on the DRR are at high risk for rotator cuff injury.

Table 5.

Analysis of the ROC curves of the CSA, AGA and GIA

	AUC	Cutoff	Youden index	Sensitivity	Specificity‌	95% CI
CSA	0.700	30.67°	0.536	0.636	0.9	(0.560, 0.840)
AGA	0.796	42.04°	0.634	0.667	0.967	(0.676, 0.916)
GIA	0.755	83.98°	0.418	0.6	0.818	(0.637, 0.872)

CSA critical shoulder angle, AGA acromioglenoid angle, GIA glenoid inclination angle

Fig. 5.

ROC curves of the CSA, AGA and GIA

Parameter correlations

After averaging the measured values of 2 surveyors for each parameter, correlation analysis was included to evaluate the correlation between different parameters. Pearson correlation analysis revealed a strong positive association between CSA and AGA (r = 0.939, P < 0.001), whereas GIA was inversely correlated with both CSA (r = −0.423, P = 0.001) and AGA (r = −0.471, P < 0.001), which was consistent with previous relevant literature (Table 6).

Table 6.

Correlation analysis between parameters

		CSA	AGA	GIA	AT	Tol	VIF
CSA	r	1.000	0.939	−0.423	0.118	1.115	8.701
	P	-	< 0.001*	0.001*	0.357
AGA	r	0.939	1	−0.471	0.061	1.111	9.012
	P	< 0.001*	-	< 0.001*	0.636
GIA	r	−0.423	−0.471	1	0.164	0.740	1.351
	P	0.001*	< 0.001*	-	0.200
AT	r	0.118	0.061	0.164	1	0.922	1.084
	P	0.357	0.636	0.200	-

*: significant difference between groups (P < 0.05)

AT acromial tilt, CSA critical shoulder angle, AGA acromioglenoid angle, GIA glenoid inclination angle

Discussion

Key findings and clinical implications

Degenerative rotator cuff injury arises from multifactorial etiologies, encompassing intrinsic tendon degeneration and extrinsic biomechanical factors. Our study used DRR to verify the critical role of scapular morphology in injury pathogenesis, with DRR technology emerging as a validated alternative to conventional radiography. DRR may serve as a reference standard for validating the measurement accuracy of non-Grashey view projections in clinical imaging [17]. By enabling standardized measurements of the CSA, AGA, and GIA without positioning constraints, DRR overcomes a key limitation of traditional Grashey-view imaging.

Relative advantages and prospects of DRR

Imaging flexibility and radiation safety

DRR is a two-dimensional projected image generated by simulating the X-ray penetration path on the basis of three-dimensional image data such as CT or MRI through computer algorithms [18]. Its core advantage lies in “customizability”. By adjusting parameters such as the position, angle, and energy of the virtual X-ray source, images from any perspective can be generated, avoiding the radiation dose from repeated scans. It complies with the “reasonable minimum dose” principle proposed by the International Commission on Radiation Protection (ICRP) [19]. For example, for adolescent patients with scoliosis who require long-term follow-up, DRR supports three-dimensional reconstruction of the whole body bone and weight-bearing position imaging (such as the EOS system) [20], which can more accurately reflect the functional state of the skeletal system, and its radiation dose is only 10%−12.5% of that of traditional X-ray films [21].

Traditional X-ray films rely on X-rays passing through the patient and being projected onto a photosensitive plate or digital detector that receives the radiation signal to form an image. The viewing angle is fixed. To obtain the target image or multiple angles of images, the angle needs to be adjusted, and multiple shots need to be taken, resulting in a relatively larger radiation dose, a more complex process, and a limited number of images obtained.

Multimodal integration and post-processing capabilities

DRR supports the fusion of multimodal image data such as CT and MRI data, provides more precise positioning, enhances the contrast of specific tissues (such as tumor boundaries), and optimizes the display effect through window width/window position adjustment. The traditional X-ray post-processing capability is limited and relies on the original exposure parameters and detector performance.

Limitations and development prospects of DRR

Despite its significant advantages, DRR is fundamentally constrained by the quality of source three-dimensional imaging data. Suboptimal resolution of the original CT scans directly compromises DRR output quality. In the current study, CT slice thickness was limited to 5 mm due to institutional equipment constraints. This exceeds the 1 mm slice thickness established as optimal in prior literature [17], thereby reducing DRR measurement accuracy and introducing potential measurement error. Future studies utilizing submillimeter slice thickness can mitigate this limitation.

In terms of development prospects, the application and development of DRR need to be combined with clinical needs and technical limitations. In the future, with the introduction of AI algorithms (such as deep learning to optimize DRR generation), the potential of DRR in dynamic imaging, real-time navigation and multimodal fusion will be further explored.

Parameter-Specific analysis

AT and diagnostic ambiguity

In contrast to prior reports associating AT with impingement risk [22]–[23], our study revealed no significant association between AT and degenerative injury (40.18 ± 5.76° vs. 42.32 ± 4.78°, P = 0.115). This discrepancy may stem from substantial anatomical variability in the Y-position scapular morphology, potentially confounding standardized AT measurements despite uniform positioning protocols.

Current biomechanical models posit that a reduced acromiohumeral distance (AHD) secondary to supraspinatus muscle pathoanatomy predisposes patients to subacromial impingement and subsequent rotator cuff degeneration [24–26]. However, isolated Y-position acromial tilt measurements may inadequately predict AHD narrowing, as dynamic humeral head positioning and acromial morphology collectively determine subacromial space dimensions. These findings challenge the diagnostic utility of AT as an independent parameter, highlighting the need for multimodal assessment incorporating dynamic joint kinematics.

GIA and glenoid biomechanics

The GIA, which serves as a quantitative measure of glenoid upward tilt, demonstrated significant diagnostic potential in this study. Compared with the control group, the rotator cuff injury group presented a significantly lower GIA (79.24 ± 5.67° vs. 84.36 ± 4.53°, P < 0.01), with an AUC of 0.755, suggesting moderate diagnostic utility for detecting degenerative injury. These findings align with Hughes’s hypothesis [8] that reduced glenoid inclination promotes humeral head migration, increasing rotator cuff strain. The moderate accuracy of the GIA underscores the need for multimodal assessment.

Prior investigations highlight three key aspects of GIA: (1) 2D CT-based GIA measurements exhibit superior accuracy compared with AP position radiographs [27], likely due to reduced projection-related artifacts; (2) MRI-derived GIA measurements demonstrate high reliability [28], although their clinical relevance remains debated, as inclination angles may contribute secondarily to injury pathogenesis through altered glenohumeral kinematics; and (3) GIA shows a linear association with CSA [8], reflecting their shared dependence on glenoid orientation.

While the independent diagnostic performance of the GIA was moderate (AUC = 0.755), logistic regression revealed its synergistic relationship with the CSA (OR = 1.244, 95% CI (1.090, 1.429)), suggesting complementary biomechanical roles. This biphasic relationship—where elevated GIA correlates with osteoarthritis risk, whereas reduced GIA predicts rotator cuff injury—warrants further exploration to delineate threshold effects across pathologies.

CSA as a biomechanical marker

The CSA combines measurements of acromial lateral extension and glenoid tilt in the coronal plane. MOOR et al. [9] found that higher levels of CSA (> 35°) are often associated with degenerative rotator cuff injury, whereas lower levels of CSA are associated with shoulder osteoarthritis [29]. The reason is that the larger the CSA is, the more inclined the force vector of the deltoid intermediate fibers is to pull the humeral head upward. As a result, the supraspinatus muscle needs to exert more horizontal force to stabilize the center of rotation during active abduction, resulting in rotator cuff degeneration and tear. In the case of a small CSA, the force vector generated by the deltoid muscle is unbalanced with the glenoid, resulting in increased load on the shoulder surface, which leads to osteoarthritis due to wear. Therefore, the CSA is a reliable radiological parameter. In this study, we found that the CSA in the rotator cuff injury group was significantly greater than that in the control group (31.65 ± 3.71° vs. 29.50 ± 1.35°, P < 0.01). Although it is lower than the CSA > 35° reported in previous studies [30], the size of the CSA is affected by race and region [31], and a CSA > 35° is more common in Europeans [30], whereas the threshold is relatively small in Asian races [32]. All the patients involved in this study were Chinese, so the results are in line with this view. On the other hand, this study did not conduct specific qualitative or quantitative comparisons of the severity of rotator cuff injuries, and a considerable proportion of the patients with rotator cuff injuries included in this study had only grade 1 rotator cuff injuries; that is, only high signals were shown in the tendon area during MRI examination, but the tendon morphology was normal, the continuity was intact, and no tears occurred [16]. On the basis of previous research and data analysis, we believe that the more severe the rotator cuff injury is, such as partial tears to full-thickness tears, the greater the CSA may be. Therefore, the CSA values of the patients included in this study were to some extent lower than those reported in previous related studies, which did not affect the reliability of the conclusions of this study. In addition, there might be another reason for the relatively lower CSA values in this study; that is, there might be differences in imaging methods between traditional X-ray films and DRRs, which may cause a certain degree of difference in measurement values. However, according to the comparison results between the two groups in this study, this does not seem to affect the diagnostic efficacy of using the CSA to determine the risk of rotator cuff injury. The remaining indicators are the same (AGA, GIA).

AGA: a novel diagnostic parameter

Miswan MF et al. [10] proposed the AGA parameter, which is based on the mechanical theory that the rotator cuff muscle has the most stable effect on the humeral head when the shoulder joint reaction force is concentrated in the middle of the shoulder joint during arm lift, and found that a larger AGA is related to rotator cuff injury. Our findings corroborate this hypothesis, with the AGA significantly elevated in the injury group (43.52 ± 4.83° vs. 39.20 ± 2.05°, P < 0.01) and demonstrating a relatively high diagnostic accuracy (AUC = 0.796).

CSA–AGA synergy and clinical implications

The strong positive correlation between the CSA and AGA (r = 0.939, P < 0.01) suggests that these parameters share biomechanical origins. While the CSA primarily reflects static osseous geometry, the AGA may contribute to dynamically capturing the effect of joint instability on the rotator cuff. Notably, both parameters exhibit biphasic relationships with shoulder pathologies: increased CSA/AGA predicts rotator cuff injury, whereas decreased CSA/AGA is associated with osteoarthritis [9]–[10]. This dual predictive ability highlights their potential for differential diagnosis, although threshold validation across ethnic groups remains imperative.

Methodological advancements

Our application of DRR addresses a critical gap in clinical imaging. Unlike conventional X-ray, DRR allows retrospective reconstruction of ideal projections from existing CT data, increasing accessibility for population-specific studies. For example, the lower CSA threshold observed here (cutoff = 30.67° in our study vs. 35° in European cohorts [30]) highlights the importance of ethnic considerations [31], as previously noted by Shinagawa et al. [32]. However, clinical implementation of DRR remains constrained by radiation exposure concerns. Consequently, the current study exclusively utilized DRR derived from pre-existing CT scans. Building on recent advances in low-dose imaging [20, 21], future research should prioritize developing DRR generation techniques based on ultra-low-dose EOS systems or radiation-free MRI protocols.

Clinical implications for early intervention and prevention

Aberrant increases in CSA/AGA and decreased GIA serve as indicators of elevated rotator cuff injury risk. At this preclinical stage, targeted preventive interventions—including rotator cuff and scapular stabilizer strengthening, optimization of scapulohumeral rhythm, and movement pattern retraining—can mitigate risk through three synergistic mechanisms: (1) enhanced rotator cuff tensile strength, (2) reduced mechanical wear through improved glenohumeral kinematics, and (3) subacromial space maximization with consequent impingement reduction. Implementing such biomechanically informed strategies may effectively preserve shoulder function, improve quality of life, and reduce surgical indications by interrupting the degenerative cascade.

Limitations and future directions

This study has several limitations: (1) Relatively small sample size and the retrospective, monocentric design constrain causal inference. (2) Retrospective CT data analysis indicates a potential risk of selection bias. (3) DRR resolution may introduce minor measurement variability. (4) Due to the limitations of equipment conditions, the CT layer thickness in this study is relatively high, which to some extent affects the measurement accuracy. (5) Exclusively Chinese cohort limits generalizability across races. (6) Due to radiation factors, DRR is currently only recommended for patients in retrospective studies where CT has already been present.

Future Research Priorities: (1) External validation in multi-race cohorts to establish generalizability. (2) Conduct a more in-depth subgroup study on the degree of rotator cuff injury. (3) Large-scale prospective multicenter studies for definitive verification. (4) Explore the application of DRR in clinical screening to balance diagnostic benefits and radiation risks. (5) Exploration of real-time DRR integration into clinical workflows through AI-accelerated reconstruction.

Conclusion

This study established statistically significant correlations between scapular morphometric parameters (GIA, CSA, AGA) and degenerative rotator cuff injury. DRR-based measurements provide a reliable, clinically actionable diagnostic approach. Notably, AGA emerged as a novel predictor of rotator cuff injury (AUC = 0.796), demonstrating potential to complement CSA in early-stage diagnostic algorithms. By integrating biomechanical principles with clinical data, these findings underscore the clinical utility of quantitative scapular assessment in personalized orthopedic management.

Abbreviations

AT

acromial tilt

CSA

critical shoulder angle

AGA

acromioglenoid angle

GIA

glenoid inclination angle

DRRs

digital reconstruction radiographs

CT

computed tomography

ICC

intra-class correlation coefficient

ROC

receiver operating characteristic

AUC

area under curve

MRI

magnetic resonance imaging

T1WI

T1-weighted imaging

T2WI

T2-weighted imaging

PDWI

proton density-weighted imaging

DICOM

digital imaging and communications in medicine

OR

odds ratio

AHD

acromiohumeral distance

AP

anteroposterior

SD

standard deviation

Authors’ contributions

Conceptualization: WB-Z and FL-L.Measurements: WB-Z, FL-L and YL-M.Statistical analysis: WB-Z.Writing: WB-Z, FL-L and YL-M.Supervision and modifications: YM-H.Revising the manuscript: WB-Z, H-S, ZX-Y and XM-D.WB-Z, FL-L, and YL-M contributed equally to the study.All the authors reviewed the manuscript.

Funding

This research was funded by the National Natural Science Foundation of China (project number 82174262), the Sichuan Provincial Administration of Traditional Chinese Medicine (project number 2024MS559), and the Scientific Research Promotion Plan of Xinglin Scholars Discipline Talents of Chengdu University of TCM (project number 030058047).

Data availability

All the data generated or analyzed during this study are included in this published article, where specific details are reasonably available by contacting the first author (Wen-bin Zhang) if necessary.

Declarations

Ethics approval and consent to participate

This study was carried out in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.