Comparison of video-based and sensor-based head impact exposure

Kuo, Calvin; Wu, Long; Loza, Jesus; Senif, Daniel; Anderson, Scott; Camarillo, David B.

doi:10.1371/journal.pone.0199238

Cited by 62 publications

(83 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Overall the frequency of head contact reported herein is lower than the rates for youth tackle football, even accounting for the longer duration of tackle football games 20 27–29. Differences in impact-recording methodology and definition of AE preclude direct comparison, particularly since sensor-based impact counts tend to overestimate impacts compared with video analysis 24 30 31. However, conservative estimates would suggest approximately two head impacts per player per 10 game-minutes20 29 32 compared with 0.1 head impacts per player per 10 game-minutes for non-tackle in the present study.…”

Section: Discussioncontrasting

confidence: 66%

Quantitative and qualitative analysis of head and body impacts in American 7v7 non-tackle football

Jadischke¹,

Zendler

Lovis³

et al. 2020

BMJ Open Sport Exerc Med

View full text Add to dashboard Cite

ObjectivesNon-tackle American football is growing in popularity, and it has been proposed as a safer alternative for young athletes interested in American football. Little is known about the nature of head contact in the sport, which is necessary to inform the extent to which protective headgear is warranted. The objective of this study was to identify the location, types and frequency of head and body contacts in competitive 7v7 non-tackle American football.MethodsVideo analysis was used to document the type, frequency and mechanism of contacts across a series of under 12, under 14 and high school non-tackle tournament games. A subset of impacts was quantitatively analysed via 3-D model-based image matching to calculate the preimpact and postimpact speed of players’ heads and the change in resultant translational and rotational velocities.ResultsThe incidence rate of head contact was found to be low (3.5 contacts per 1000 athlete-plays). Seventy-five per cent of head contacts were caused by a head-to-ground impact. No head-to-head contacts were identified. Most contacts occurred to the rear upper (occiput) or side upper (temporal/parietal) regions. Head-to-ground impact was associated with a maximum preimpact velocity of 5.9±2.2 m/s and a change in velocity of 3.0±1.1 m/s.ConclusionNon-tackle football appears to represent a lower contact alternative to tackle football. The distribution of head impact locations, mechanisms and energies found in the present study is different than what has been previously reported for tackle football. The existing tackle football standards are not appropriate to be applied to the sport of non-tackle football, and sport-specific head protection and headgear certification standards must be determined.

show abstract

Section: Discussioncontrasting

confidence: 66%

Quantitative and qualitative analysis of head and body impacts in American 7v7 non-tackle football

Jadischke¹,

Zendler

Lovis³

et al. 2020

BMJ Open Sport Exerc Med

View full text Add to dashboard Cite

show abstract

“…The ability to discriminate head impacts from spurious events poses a critical challenge for sensor systems. 24,31,39 Existing wearable systems have employed a range of strategies to discriminate spurious events from true head impacts. Given signal to noise issues at low magnitudes of impact, the simplest method of discrimination is the use of a sensor-level threshold (trigger) for data collection based on the magnitude of the linear acceleration time signal.…”

Section: Introductionmentioning

confidence: 99%

“…Recently, artificial intelligence methodology has been used to discriminate between head impacts and spurious events. 24,27,39 The application of artificial intelligence leverages features based on the time and frequency domains of the kinematic traces in machine learning (ML) models, and has the potential to be more effective than filters based on pulse magnitude, duration, or frequency content. In ML, features are attributes that describe observations.…”

Section: Introductionmentioning

confidence: 99%

On-Field Performance of an Instrumented Mouthguard for Detecting Head Impacts in American Football

Gabler¹,

Huddleston²,

Dau³

et al. 2020

Ann Biomed Eng

View full text Add to dashboard Cite

Wearable sensors that accurately record head impacts experienced by athletes during play can enable a wide range of potential applications including equipment improvements, player education, and rule changes. One challenge for wearable systems is their ability to discriminate head impacts from recorded spurious signals. This study describes the development and evaluation of a head impact detection system consisting of a mouthguard sensor and machine learning model for distinguishing head impacts from spurious events in football games. Twenty-one collegiate football athletes participating in 11 games during the 2018 and 2019 seasons wore a custom-fit mouthguard instrumented with linear and angular accelerometers to collect kinematic data. Video was reviewed to classify sensor events, collected from instrumented players that sustained head impacts, as head impacts or spurious events. Data from 2018 games were used to train the ML model to classify head impacts using kinematic data features (127 head impacts; 305 non-head impacts). Performance of the mouthguard sensor and ML model were evaluated using an independent test dataset of 3 games from 2019 (58 head impacts; 74 non-head impacts). Based on the test dataset results, the mouthguard sensor alone detected 81.6% of video-confirmed head impacts while the ML classifier provided 98.3% precision and 100% recall, resulting in an overall head impact detection system that achieved 98.3% precision and 81.6% recall.

show abstract

“…For 413 direct head impacts ≥20 g, the x-patch™ accurately recorded the location in 24.9% (n = 103) of impacts, see Table 3. Previously Kuo and colleagues, using a similar tri-axial linear accelerometer embedded into a mouthguard, reported similarly poor rates of agreement between sensor recorded and video identi ed impacts (37.3%), with impact locations that did not match the direction of motion [29]. Note.…”

Section: Discussionmentioning

confidence: 91%

“…A detailed overview of all impact events for a hit-up, tackle, and off-the-ball event is provided in Table 2. Chest 0 (0,0,0) 26 (10,16,0) 10 (2,8,0) 5 g (0,3,2) 0 (0,0,0) 0 (0,0,0) 18 (14,2,2) 59 (26,29,4) Arm 0 (0,0,0) 0 (0,0,0) 0 (0,0,0) 0 (0,0,0) 0 (0,0,0) 0 (0,0,0) 1 (1,0,0) 1 (1,0,0) Waist 0 (0,0,0) 3 (3,0,0) 0 (0,0,0) 0 (0,0,0) 0 (0,0,0) 0 (0,0,0) 4 (3,1,0) 7 (6,1,0) Insert Table 2 Table 3. Insert Table 3 About Here Tackles and Secondary Impacts Secondary impacts during a tackle (i.e., impacts after the initial contact) accounted for 53.5% (n = 221) of all direct head impacts and 46.1% (n = 260) of total impacts ≥20 g. For 260 secondary impacts, 16.2% (n = 42) were accompanied by a video veri ed primary impact ≥20 g, with 83.8% (n = 218) of all secondary impacts occurring after the primary impact was less than 20 g. There were 456 tackles that resulted in the 535 video veri ed impacts ≥20 g, excluding impacts that occurred "off the ball."…”

Section: Situational Characteristics Of Video Veri Ed and Sensor Recomentioning

confidence: 99%

Video Analysis and Verification of Direct Head Impacts Recorded by Wearable Sensors in Junior Rugby League Players

Carey

Terry

McIntosh

et al. 2020

Preprint

View full text Add to dashboard Cite

Background: Rugby League is a high-intensity collision sport that carries a risk of concussion. Youth athletes are considered to be more vulnerable and take longer to recover from concussion than adult athletes. Purpose: To review head impact events in elite level junior representative rugby league and to verify and analyze x-patchTM recorded impacts via video analysis.Study Design: Observational case series.Methods: The x-patchTM was used on twenty-one adolescent players (thirteen forwards and eight backs) during a 2017 junior representative rugby league competition. Game day footage, recorded by a trained videographer from a single camera, was synchronized with accelerometer timestamps. Impacts were double verified by video review. Impact rates, playing characteristics, and game play situations were described.Results: The x-patchTM recorded 624 impacts 20g between game start and finish, of which 564 (90.4%) were verified on video. Upon video review, 413 (73.2%) of all verified impacts 20g where determined to be direct head impacts. Direct head impacts 20g occurred at a rate of 5.2 impacts per game hour; 7.6 for forwards and 3.0 for backs (range=0-18.2). A defender’s arm directly impacting the head of the ball carrier was the most common event, accounting for 21.3% (n=120) of all impacts, and 46.7% of all “hit-up” impacts. There were no medically diagnosed concussions during the competition.Conclusion: The majority (90.4%) of impacts 20g recorded by the x-patchTM sensor were verified by video. Double verification of direct head impacts in addition to cross-verification of sensor recorded impacts using a secondary source such as synchronized video review can be used to ensure accuracy and validation of data.

show abstract

Comparison of video-based and sensor-based head impact exposure

Cited by 62 publications

References 42 publications

Quantitative and qualitative analysis of head and body impacts in American 7v7 non-tackle football

Quantitative and qualitative analysis of head and body impacts in American 7v7 non-tackle football

On-Field Performance of an Instrumented Mouthguard for Detecting Head Impacts in American Football

Video Analysis and Verification of Direct Head Impacts Recorded by Wearable Sensors in Junior Rugby League Players

Contact Info

Product

Resources

About