Additional Methodologies 1 Study 4.1 – Validity

Validity Instrumentation

Bowling kinematics were recorded using a 14 camera Vicon Motion Analysis System (Oxford, UK) operating at 200Hz. Ground reaction forces at BFI and FFI were also recorded using two Kistler force plates (900x600mm) sampling at 1000Hz.

39 14mm retroflective markers were attached to each participant, positioned on landmarks dictated by the full body plug-in-gait model (Figure 3.2.1)(Vicon Nexus 2.7). Inertial sensors were also attached as in the main methods section of this thesis.

Validity Procedure

Each participant completed a self-selected warm up and was then allowed as many bowls as was needed for them to familiarise themselves with the experimental set-up. The lab allowed for a full-length run up. The ball was then bowled into a net 5m away from the point of ball release. Following familiarisation one over (6 balls) were bowled maximally and recorded for analysis. If clean contact with force plates at BFI and FFI were not achieved the trial was repeated.

Figure 3.2.1. Marker placement for full body plug-in-gait model (Vicon Nexus 2.7). Validity Data Processing

For the purpose of this study only orientation of the thorax and pelvis segments were needed from the plug-in gait kinematic model. The thorax segment was constructed using clavicle (CLAV), sternum (STRN), C7 and T10 markers (as labelled in figure 3.2.1). The pelvis segment was constructed using left and right anterior and posterior iliac spines (LASI, RASI, LPSI, RPSI) (figure 3.2.1).

Back and front foot impact were defined as the point at which a force greater than a 5 Newton threshold was observed on the corresponding force plate. Raw force data were

exported and processed in Matlab (R2012a). All force data were filtered using a bidirectional second-order, low-pass Butterworth filter with a cut-off frequency of 50Hz. Peak vertical, anterior-posterior, mediolateral and resultant GRF at BFI and FFI were recorded. Resultant GRF was defined as the square root of the sum of squares of vertical GRF, anterior-posterior GRF and mediolateral GRF (same method as resultant accelerations in the main methodology). Time to peak vertical and resultant GRF were defined as the time between initial contact with the force plate and peak vertical and resultant GRF. These values were also recorded at both BFI and FFI.

Shoulder counter-rotation was defined as the orientation of the thorax at BFI subtracted from max rotation away from the direction of delivery (same as with T1 inertial sensor in main method). Three-dimensional lumbar kinematics were obtained via the relative orientations of the thorax segment relative to the pelvis. Lumbar flexion, lateral flexion and rotation were recorded at BFI and FFI to enable comparisons to inertial sensor data. Processing of inertial sensor data was identical to that described in the main method section in this thesis.

3.2.2 Study 4.2 - Playing surfaces and lower limb impacts Impact Testing Procedure

These methods of data collection were adapted in order to quantify surface firmness as was the aim of study 4.2. In order to quantify the surface properties of different cricket playing surfaces, a custom-built impactor was developed. The same ±200 g tri-axial accelerometer (THETAmetrix, Waterlooville, UK, ADXL377), sampling at 750 Hz was utilised, aligned vertically with the centre of mass of an impact weight 63 mm in diameter and 2.5 kg in weight. The impact weight was suspended in a guidance tube to standardise drop height to 200mm (Figure 3.2.2). This testing rig system was based on similar impactor devices (Baker et al. 2001). An additional 20 mm of Adiprene polyurethane foam and 3 mm of rubber (taken from the heel of a typical sports training shoe) was attached to the bottom of the impact weight to more accurately simulate impact conditions during cricket bowling.

Acceleration data were collected across four different cricket playing surfaces: Grass wicket, artificial outdoor wicket, indoor wood and indoor rubber composite (Uniturf) (as seen in figure 3.2.4a, b, c, d,). Impact data were sampled at 12 locations of the

planning frame, consisting of twelve 400x440mm squares was used to identify sampling locations (seen in figure 3.2.3). Each square was tested 6 times in random locations within the square.

Figure 3.2.3. Segmentation of popping crease for impactor testing locations

Data Processing

All data were collected in Sensor Suite (v504) and transferred to Matlab (Ed. R2012a) where peak and time-to-peak acceleration data were identified for the initial impact of the weight with the playing surface. Acceleration data were filtered using a bidirectional second-order, low-pass Butterworth filter with a cut-off frequency of 50Hz. Peak acceleration was identified manually, and time-to-peak acceleration defined as the time taken for acceleration to reach its peak from the point of initial increase on the impact peak.

For surface impactor values a mean of the 6 tests at each square was taken for all variables, thus giving 12 values used to describe the characteristics of each surface. Average loading rate was calculated by dividing peak tibial acceleration by time-to- peak acceleration (Stiles and Dixon, 2007).

3.2.3 Study 5.2 – Fast bowling biomechanics and lower back pain risk Injury Surveillance

Before the start of the 2015 season, history of low back pain or injury was explored using a specifically created questionnaire, which included playing history (See Appendix 3). The questionnaire was administered with the guidance of the researcher and verified by the club physiotherapist where possible. The questionnaire sought to determine if a previous history of low back pain or injury was present, enabling a sub- grouping of bowlers based on LBP history. In addition to previous history of back pain, pain experienced in the following season was also explored. Bowlers were instructed to keep a record of any LBP or injury during the 2015 season if a physiotherapist was not able to do this for them. This study defined LBP as any pain affecting the area of the back inferior to the lower ribs, superior to the inferior gluteal folds and medial to the mid-axillary line that impacted on their ability to bowl for a minimum of 3 days. Junior and senior fast bowlers were grouped separately in order to avoid age becoming a confounding variable. Therefore, bowlers were able to be sub-classified based on whether they had a history of LBP as well as whether they went on to develop LBP in the following season.

3.2.4 Study 6.1

Ball Release Speed Analysis

In addition to the fast bowling technique kinematic and kinetic variables highlighted above, study 6.1 aimed to provide a comparison with ball release speed.

Instrumentation

One high-speed video camera (Sony FX1000) sampling at 200Hz was used to record ball release speed. This camera was positioned as shown in figure 3.2.5.

Figure 3.2.5. Camera position and example of the digitising process used to calculate ball release speed.

Figure 3.2.6. Target area (red box) used to define a ‘successful’ bowl.

5m

2m 4m

Bowler’s End Batsmen’s End

X = x position, Y= y position, t = time

The camera was positioned to record in the sagittal plane 5m from the middle stump and aligned with the popping crease. Markers were placed on the camera-facing stump 30cm apart to allow distance calibration. A target area was also placed on the pitch using cones. This area denotes a ‘good’ line and length and as such only trial landing inside this area were analysed. Figure 3.2.5 provides this study’s definition of a ‘good’ line and length.

Data Processing

All video data were processed in Kinovea (v0.8.15). As shown in figure 3.2.5, the ball was tracked manually for the first 5 frames following ball release and x(horizontal) and y (vertical) coordinates were recorded. Ball speed was calculated between each of these points (as detailed in the Equation 3.2.1) with the average value recorded as the participant’s ball release speed.

Equation 3.2.1. Calculation of ball release speed from digitised two-dimensional co-ordinates of ball position of two consecutive frames at 200Hz sampling frequency following ball release.