A two-step neural network-based guiding system for obtaining reliable radiographs for critical shoulder angle measurement.

Background: The critical shoulder angle (CSA) has been reported to be highly associated with rotator cuff tears (RCTs) and an increased risk of RCT re-tears. However, the measurement of the CSA is greatly affected by the malpositioning of the shoulder. To address this issue, a two-step neural network-based guiding system was developed to obtain reliable CSA radiographs, and its feasibility and accuracy was evaluated. Methods: A total of 1,754 shoulder anteroposterior (AP) radiographs were retrospectively acquired to train and validate a two-step neural network-based guiding system to obtain reliable CSA radiographs. The study included patients aged 18 years or older who underwent X-rays and/or computed tomography (CT) scans of the shoulder. Patients who had undergone shoulder surgery, had a confirmed fracture, or were diagnosed with a musculoskeletal tumor or glenoid defect were excluded from the study. The system consisted of a two-step neural network that in the first step, localized the region of interest of the shoulder, and in the second step, classified the radiography according to type [i.e., 'forward' when the non-overlapping coracoid process is above the glenoid rim, 'backward' when the non-overlapping coracoid process is below or aligned with the glenoid rim, a ratio of the transverse to longitudinal diameter of the glenoid projection (RTL) ≤0.25, or a RTL >0.25]. The performance of the model was assessed in an offline, prospective manner, focusing on the sensitivity and specificity for the forward, backward, RTL ≤0.25, or RTL >0.25 types (denoted as SensF, B, -, + and SpecF, B, -, +, respectively), and Cohen's kappa was also reported. Results: Of 273 cases in the offline prospective test, the SensF, SensB, Sens-, and Sens+ were 88.88% [95% confidence interval (CI): 50.67-99.41%], 94.11% (95% CI: 82.77-98.47%), 96.96% (95% CI: 91.94-99.02%), and 95.06% (95% CI: 87.15-98.40%), respectively. The SpecF, SpecB, Spec-, and Spec+ were 98.48% (95% CI: 95.90-99.51%), 99.55% (95% CI: 97.12-99.97%), 95.04% (95% CI: 89.65-97.81%), and 97.39% (93.69-99.03%), respectively. A high classification rate (93.41%; 95% CI: 89.14-96.24%) and almost perfect agreement (Cohen's kappa: 0.903, 95% CI: 0.86-0.95) were achieved. Conclusions: The guiding system can rapidly and accurately classify the types of AP shoulder radiography, thereby guiding the adjustment of patient positioning. This will facilitate the rapid obtainment of reliable CSA radiography to measure the CSA on proper AP radiographs.

Enhancing prediction of supraspinatus/infraspinatus tendon complex injuries through integration of deep visual features and clinical information: a multicenter two-round assessment study.

OBJECTIVE: Develop and evaluate an ensemble clinical machine learning-deep learning (CML-DL) model integrating deep visual features and clinical data to improve the prediction of supraspinatus/infraspinatus tendon complex (SITC) injuries. METHODS: Patients with suspected SITC injuries were retrospectively recruited from two hospitals, with clinical data and shoulder x-ray radiographs collected. An ensemble CML-DL model was developed for diagnosing normal or insignificant rotator cuff abnormality (NIRCA) and significant rotator cuff tear (SRCT). All patients suspected with SRCT were confirmed by arthroscopy examination. The model's performance was evaluated using sensitivity, specificity, accuracy, and area under the curve (AUC) metrics, and a two-round assessment was conducted to authenticate its clinical applicability. RESULTS: A total of 974 patients were divided into three cohorts: the training cohort (n = 828), the internal validation cohort (n = 89), and the external validation cohort (n = 57). The CML-DL model, which integrates clinical and deep visual features, demonstrated superior performance compared to individual models of either type. The model's sensitivity, specificity, accuracy, and area under curve (95% confidence interval) were 0.880, 0.812, 0.836, and 0.902 (0.858-0.947), respectively. The CML-DL model exhibited higher sensitivity and specificity compared to or on par with the physicians in all validation cohorts. Furthermore, the assistance of the ensemble CML-DL model resulted in a significant improvement in sensitivity for junior physicians in all validation cohorts, without any reduction in specificity. CONCLUSIONS: The ensembled CML-DL model provides a solution to help physicians improve the diagnosis performance of SITC injury, especially for junior physicians with limited expertise. CRITICAL RELEVANCE STATEMENT: The ensembled clinical machine learning-deep learning (CML-DL) model integrating deep visual features and clinical data provides a superior performance in the diagnosis of supraspinatus/infraspinatus tendon complex (SITC) injuries, particularly for junior physicians with limited expertise. KEY POINTS: 1. Integrating clinical and deep visual features improves diagnosing SITC injuries. 2. Ensemble CML-DL model validated for clinical use in two-round assessment. 3. Ensemble model boosts sensitivity in SITC injury diagnosis for junior physicians.

Deep learning for automated measurement of CSA related acromion morphological parameters on anteroposterior radiographs.

BACKGROUND: The Critical Shoulder Angle Related Acromion Morphological Parameter (CSA- RAMP) is a valuable tool in the analyzing the etiology of the rotator cuff tears (RCTs). However, its clinical application has been limited by the time-consuming and prone to inter- and intra-user variability of the measurement process. OBJECTIVES: To develop and validate a deep learning algorithm for fully automated assessment of shoulder anteroposterior radiographs associated with RCTs and calculation of CSA-RAMP. METHODS: Retrospective analysis was conducted on radiographs obtained from computed tomography (CT) scans and X-rays performed between 2018 and 2020 at our institution. The development of the system involved the utilization of digitally reconstructed radiographs (DRRs) generated from each CT scan. The system's performance was evaluated by comparing it with manual and semiautomated measurements on two separate test datasets: dataset I (DRRs) and dataset II (X-rays). Standard metrics, including mean average precision (AP), were utilized to assess the segmentation performance. Additionally, the consistency among fully automated, semiautomated, and manual measurements was comprehensively evaluated using the Pearson correlation coefficient and Bland-Altman analysis. RESULTS: A total of 1080 DRRs generated from 120 consecutive CT scans and 159 X-ray films were included in the study. The algorithm demonstrated excellent segmentation performance, with a mean AP of 57.67 and an AP50 of 94.31. Strong inter-group correlations were observed for all CSA-RAMP measurements in both test datasets I (automated versus manual, automated versus semiautomated, and semiautomated versus manual; r = [0.990---0.997], P < 0.001) and dataset II (r = [0.984---0.995], P < 0.001). Bland-Altman analysis revealed low bias for all CSA-RAMP measurements in both test datasets I and II, except for CD (with a maximum bias of 2.49%). CONCLUSIONS: We have successfully developed a fully automated algorithm capable of rapidly and accurately measuring CSA-RAMP on shoulder anteroposterior radiographs. A consistent automated CSA- RAMP measurement system may accelerate powerful and precise studies of disease biology in future large cohorts of RCTs patients.

Single NV centers array preparation and static magnetic field detection.

To solve the problem of static magnetic field detection accuracy and consistency, we prepared an array of single NV centers for static magnetic field vector and gradient detection using the femtosecond laser direct writing method. The prepared single NV centers are characterized by fewer impurity defects and good stress uniformity, with an average spatial positioning error of only 0.2 µm. This array of single NV centers can achieve high accuracy magnetic field vector and gradient measurement with GBZ≈-0.047 µT/µm in the Z-axis. This result provides a new idea for large-range, high-precision magnetic field vector and gradient measurements.

Phase customization in photonic integrated circuits with trimmed waveguides.

Accurate photon phase control on a chip is essential to improve the expandability and stability of photonic integrated circuits (PICs). Here, we propose a novel, to the best of our knowledge, on-chip static phase control method in which a modified line is added close to the normal waveguide with a lower-energy laser. By controlling the laser energy and the position and length of the modified line, the optical phase can be precisely controlled with low loss and a three-dimensional (3D) path. Customizable phase modulation ranging from 0 to 2π is performed with a precision of λ/70 in a Mach-Zehnder interferometer. The proposed method can customize high-precision control phases without changing the waveguide's original spatial path, which is expected to control the phase and solve the phase error correction problem during processing of large-scale 3D-path PICs.

Evanescent Field Controllable MZ Sensor via Femtosecond Laser Processing and Mechanic Polishing.

Recently, optical sensors interacting with evanescent fields and the external environment around waveguides have attracted extensive attention. In the process of light propagation in the waveguide, the depth of the evanescent field is closely related to the accuracy of the optical sensor, and adjusting the depth of the evanescent field to obtain higher accuracy has become the primary challenge in fabricating on-chip optical sensors. In this study, the waveguide structure of a Mach-Zehnder interferometer was written directly in Corning Eagle 2000 borosilicate glass by a femtosecond laser, and the sensing window was exposed out of the bulk material by mechanical polishing. The refractive index detection device based on the proposed on-chip Mach-Zehnder interferometer has the advantages of small volume, light weight, and good stability. Its sensitivity can reach 206 nm/RIU or 337 dB/RIU, and the theoretical maximum measurement range is 1-1.508. Therefore, it can measure the refractive index quickly and accurately in extreme or complex environments, and has excellent application prospects.

Necessary and sufficient condition for absolute stability of normal neural networks.

Globally convergent dynamics of a class of neural networks with normal connection matrices is studied by using the Lyapunov function method and spectral analysis of the connection matrices. It is shown that the networks are absolutely stable if and only if all the real parts of the eigenvalues of the connection matrices are nonpositive. This extends an existing result on symmetric neural networks to a larger class including certain asymmetric networks. Further extension of the present result to certain non-normal case leads naturally to a quasi-normal matrix condition, which may be interpreted as a generalization of the so-called principle of detailed balance for the connection weights or the quasi-symmetry condition that was previously proposed in the literature in association with symmetric neural networks. These results are of particular interest in neural optimization and classification problems.