Correspondence between motion sensor kinematics and notational analysis of scanning in field-based football gameplay

In football, players frequently engage in scanning, which consists of clearly distinguishable head-orientation changes that support visual exploration and decision-making under time pressure. While previous studies have examined scanning using video-based notational analysis or laboratory-based motion sensors, it remains unclear how scan counts based on angular velocity thresholds align with those obtained from the notational analysis during actual gameplay. This study aimed to examine the correspondence between motion sensor-based scan counts obtained using different angular velocity thresholds and scan counts obtained from video-based notational analysis during field-based football gameplay. Twelve university-level football players participated in a 6-on-6 small-sided game while wearing head-mounted inertial measurement units. Head rotation data were analyzed using four angular velocity thresholds (125, 200, 250, and 300 deg/s). Detection results were compared with the notational analysis conducted by experienced evaluators, who counted clearly distinguishable bidirectional changes in head orientation. Agreement was assessed using ANOVA, Bland-Altman plots, and concordance correlation coefficients (CCC), while detection performance was evaluated using recall, precision, and F1-score. Results showed that the 250 deg/s threshold produced scan counts that did not significantly differ from the notational analysis, yielded the smallest mean difference and narrowest limits of agreement, and achieved the highest CCC (0.932) as well as acceptable detection performance (F1-score = 0.753). In contrast, the previously used 125 deg/s threshold significantly overestimated scan counts relative to the notational analysis and demonstrated weaker agreement and lower detection accuracy. These findings indicated that, among the thresholds examined, a 250 deg/s threshold provided the closest correspondence with notational analysis when quantifying scans during field-based football gameplay. The present results suggest the importance of calibrating kinematic detection thresholds to the dynamic demands of actual sports contexts and highlight the potential of motion sensor technology for analyzing scanning in situ.