Purpose: The goniometer, the gold standard for range of motion (ROM) assessment, has significant limitations including physical contact requirements and high interrater variability. While markerless motion capture using deep learning (e.g., Sports2D) offers a noncontact alternative, its clinical reliability remains unestablished. This study evaluated Sports2D's reliability for ROM measurement using smartphone cameras compared to conventional goniometry. Methods: Fifty healthy adults (25 males, 25 females; mean age, 22.42±2.58 years) participated. Ten joint ROM types (17 total measurements including left/right) were measured using both traditional goniometry (by an experienced clinician) and Sports2D (smartphone camera at 3 m distance). Reliability and agreement were analyzed using concordance correlation coefficient (CCC), Pearson correlation, and Bland-Altman plots. Results: Sports2D demonstrated very high reliability for elbow flexion (CCC, 0.906−0.955), neck extension (CCC, 0.928), and right hip flexion (CCC, 0.920). Good reliability (CCC, ≥ 0.8) was observed for neck flexion, left knee flexion, and trunk flexion. However, all ankle measurements showed clinically unacceptable reliability (CCC, ＜ 0.3), including dorsiflexion (CCC, 0.072−0.099) and plantarflexion (CCC, 0.174–0.211). Bland-Altman plots confirmed proportional bias for ankle measurements, particularly at smaller angles. Conclusion Sports2D is a reliable noncontact tool for measuring ROM in specific joints including elbow, neck, knee, hip, and trunk. However, it is unreliable for ankle joint assessment due to significant measurement discrepancies. These findings highlight the need for selective clinical application of markerless motion capture based on joint-specific validation.

PURPOSE: The purpose of this study was to examine the relationships between hamstring muscle thickness and knee flexion peak torque, and rate of torque development (RTD) calculated during 0–50 ms (RTD50) and 0–200 ms (RTD200).METHODS: Thirty-six active individuals' dominant side hamstring thickness were measured using portable ultrasound device. Participants performed maximal isometric voluntary contraction (MVIC) of knee flexion. Peak torque was identified as the maximum torque during MVIC testing. RTD was calculated initial 50 ms and 200 ms after the onset of joint torque. Pearson's correlation (r) coefficients were utilized to assess relationships between muscle thickness and knee flexion peak torque, RTD50 and RTD200. The significant level of hypothesis verification is set-up as α=0.05.RESULTS: Greater peak torque and RTD200 was associated with greater muscle thickness of semitendinosus and semimembranosus (p<0.05). Greater RTD50 was associated with greater muscle thickness of semitendinosus only. Biceps femoris thickness was not associated with knee flexion peak torque, RTD50, and RTD200.CONCLUSION: These results suggest that the training specific hamstring muscle (medial hamstrings) for improving muscle thickness would be effective for increasing knee flexion peak torque and RTD.
Joints ; Knee ; Torque ; Ultrasonography