Diagnostic accuracy of digital photography and image analysis for the measurement of foot conformation in the horse

QL Zoology;T Technology (General)

White, Jonathan M. ; Voute, Lance

University:University of Glasgow

Department:School of Veterinary Medicine

关键词: QL Zoology;    T Technology (General);   

Others  :  http://theses.gla.ac.uk/1559/1/2008whitemvm%28r%29.pdf

来源: University of Glasgow

【 摘 要 】

Experimental studies have shown that foot conformation influences the forces experiencedby the distal limb (Riemersma et al. 1996; Wilson et al. 1998; Willemen et al. 1999;Eliashar et al. 2004). Although some clinical studies have highlighted the importance offoot conformation as a risk factor for musculoskeletal injury (Kane et al. 1998; Andersonet al. 2004), this has not been a universal finding (Weller et al. 2006c). Clearly moreinformation from large, well designed, prospective studies is needed to elucidate furtherthe relationship between foot conformation and injury. This information would help toguide veterinary surgeons and farriers in the trimming and shoeing of horses’ feet, apractice carried out regularly (usually every 4-8 weeks) during the animal’s life. In order tofurther investigate this relationship, a tool capable of accurately, precisely and practicablycollecting suitable objective data, ideally at relatively low cost, is necessary.The variability introduced by image acquisition and subsequent analysis using digitalimage processing software of digital photographs of the foot has not been quantified.Similarly the level of agreement between objective measurements of foot conformationmade from digital photography and digital radiography is unknown.The objectives of this project were to assess the precision, accuracy and practicability ofthe entire process of obtaining measurements of horses’ feet using digital photography andthe digital image processing software (Metron-PX™).For the precision study (prospective in vivo randomised clinical measurement study) lateraldigital photographs of shod Thoroughbred racehorses were obtained twice by twoveterinary surgeons working independently (Image Acquisition - IAc). Each photographwas independently analysed by the two veterinary surgeons masked to the origin of theimages on two occasions using Metron-PX™ (Image Analysis - IAn). Measurementsgenerated by the software were compared within and between operators of the software forself and non-self acquired photographs. Intra- and inter-operator agreement indices (AIs)and 95% limits of agreement (LOA) were calculated for each measurement for the IAnprocess alone and for the combined IAc + IAn processes for self and non-self acquiredimages respectively. For the accuracy study (method comparison study) measurementsobtained from lateral digital photographs (index test), as in the precision study, werecompared with those obtained from lateromedial radiographs (reference standard).Agreement indices (AIs) and 95% limits of agreement were calculated for eachmeasurement.The results of the precision study identified excellent mean intra- and inter-operator AIsfor the IAn process alone ( 0.90 for all measurements). The mean intra- and inter-operatorAIs for the combined IAc + IAn processes were0.89 for all measurements with similarAIs obtained regardless of whether or not the individual whom acquired the images alsoanalysed them. The 95% limits of agreement for hoof angle, heel height/ toe height% andcoronary band angle for all comparisons were all within target values. The results of theaccuracy study identified mean AIs that were0.89 for all measurements. The 95% limitsof agreement for heel height/ toe height% and coronary band angle were within targetvalues.Overall, there was excellent precision both within and between operators of themeasurement process for both the image analysis process alone and the combined imageacquisition and analysis processes. When the described technique is used results arecomparable irrespective of whether the person whom acquired the images also analysesthem. Excellent accuracy was also identified between the photographic and radiographicmeasurements, especially for heel height/ toe height% and coronary band angle, suggestingthat these two methods may be used interchangeably for these measurements of footconformation in the horse.The clinical relevance of these findings is that the processes described for obtainingobjective measurements of foot conformation from digital photographs and digital imageprocessing software Metron-PX™ is practicable and produces highly precisemeasurements regardless of whether the same operator performs both image acquisitionand analysis. Photographic measurements of heel height/ toe height% and coronary bandangle closely approximate radiographic measurements such that the two techniques may beused interchangeably. Thus, digital photography and image analysis have applications inthe field of clinical telemedicine and would be particularly useful to a large prospectivemulti-centre study investigating the relationship between foot conformation andmusculoskeletal injury.