==== Front PLoS One PLoS One plos PLOS ONE 1932-6203 Public Library of Science San Francisco, CA USA 10.1371/journal.pone.0287381 PONE-D-22-28433 Research Article Biology and Life Sciences Anatomy Musculoskeletal System Skeleton Femur Medicine and Health Sciences Anatomy Musculoskeletal System Skeleton Femur Biology and Life Sciences Anatomy Musculoskeletal System Skeleton Tibia Medicine and Health Sciences Anatomy Musculoskeletal System Skeleton Tibia Biology and Life Sciences Organisms Eukaryota Animals Vertebrates Amniotes Mammals Equines Biology and Life Sciences Zoology Animals Vertebrates Amniotes Mammals Equines Biology and Life Sciences Anatomy Musculoskeletal System Skeleton Skeletal Joints Medicine and Health Sciences Anatomy Musculoskeletal System Skeleton Skeletal Joints Research and Analysis Methods Computational Techniques Biometrics Biology and Life Sciences Organisms Eukaryota Animals Vertebrates Amniotes Mammals Equines Horses Biology and Life Sciences Zoology Animals Vertebrates Amniotes Mammals Equines Horses Medicine and Health Sciences Diagnostic Medicine Diagnostic Radiology Bone Imaging Research and Analysis Methods Imaging Techniques Diagnostic Radiology Bone Imaging Medicine and Health Sciences Radiology and Imaging Diagnostic Radiology Bone Imaging Medicine and Health Sciences Rheumatology Arthritis Osteoarthritis Anatomical variations of the equine femur and tibia using statistical shape modeling Statistical shape models of the equine femur and tibia https://orcid.org/0000-0002-4743-1452 He Hongjia Conceptualization Data curation Formal analysis Investigation Methodology Resources Software Visualization Writing – original draft Writing – review & editing 1 Banks Scott A. Conceptualization Data curation Investigation Methodology Supervision Visualization Writing – review & editing 2 https://orcid.org/0000-0001-7750-3640 Biedrzycki Adam H. Conceptualization Data curation Funding acquisition Methodology Project administration Resources Software Supervision Visualization Writing – review & editing 1 * 1 Department of Large Animal Clinical Science, College of Veterinary Science, University of Florida, Gainesville, Florida, United States of America 2 Department of Mechanical & Aerospace Engineering, Herbert Wertheim College of Engineering, University of Florida, Gainesville, Florida, United Stated of America Shaheen Aliah Faisal Editor Brunel University London, UNITED KINGDOM Competing Interests: The authors have declared that no competing interests exist. * E-mail: dradam@ufl.edu 30 6 2023 2023 18 6 e028738114 10 2022 5 6 2023 © 2023 He et al 2023 He et al https://creativecommons.org/licenses/by/4.0/ This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. The objective of this study was to provide an overarching description of the inter-subject variability of the equine femur and tibia morphology using statistical shape modeling. Fifteen femora and fourteen tibiae were used for building the femur and tibia statistical shape models, respectively. Geometric variations in each mode were explained by biometrics measured on ±3 standard deviation instances generated by the shape models. Approximately 95% of shape variations within the population were described by 6 and 3 modes in the femur and tibia shape models, respectively. In the femur shape model, the first mode of variation was scaling, followed by notable variation in the femoral mechanical-anatomical angle and femoral neck angle in mode 2. Orientation of the femoral trochlear tubercle and femoral version angle were described in mode 3 and mode 4, respectively. In the tibia shape model, the main mode of variation was also scaling. In mode 2 and mode 3, the angles of the coronal tibial plateau and the medial and lateral caudal tibial slope were described, showing the lateral caudal tibial slope angle being significantly larger than the medial. The presented femur and tibia shape models with quantified biometrics, such as femoral version angle and posterior tibial slope, could serve as a baseline for future investigations on correlation between the equine stifle morphology and joint disorders due to altered biomechanics, as well as facilitate the development of novel surgical treatment and implant design. By generating instances matching patient-specific femorotibial joint anatomy with radiographs, the shape model could assist virtual surgical planning and provide clinicians with opportunities to practice on 3D printed models. http://dx.doi.org/10.13039/100012651 College of Veterinary Medicine, University of Florida UF Graduate Student Fellowship Award https://orcid.org/0000-0002-4743-1452 He Hongjia HH The project was partially funded by the University of Florida Graduate Student Fellowship Award. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Data AvailabilityHere is the link for the files used in this study: https://www.ebi.ac.uk/biostudies/studies/S-BSST1116. Data Availability Here is the link for the files used in this study: https://www.ebi.ac.uk/biostudies/studies/S-BSST1116. ==== Body pmcIntroduction The equine stifle has been a comparatively overlooked joint from an advanced biomechanical perspective due to difficulties acquiring volumetric images in comparison to the distal limb joints. Equine stifle osteoarthritis (OA) is a common source of lameness and can be spontaneous or secondary to trauma [1–4]. In humans, numerous studies have investigated the role of joint morphology in the onset and progression of OA in order to understand the relationship between shape and the development of OA [5–7]. Joint geometry plays an essential role in biomechanical behaviors and contributes to the progression of human knee OA [8]. However, the relationship between equine stifle morphology and OA has yet to be explored, and detailed descriptions about the anatomical shape variation in the equine femur and tibia are still lacking. One common method to characterize complex anatomical geometry is statistical shape modeling (SSM), which has been employed to quantify the anatomical morphological variability in various joints in humans, such as carpometacarpal joint and patellofemoral joint [9, 10]. In horses, SSM has been used to evaluate the equine third metacarpal and also an articulating model of the equine distal limb [11, 12]. SSM creates a point-to-point-based correspondence within samples and extracts the most prominent modes of shape variation within the population, named principal components or modes, using principal component analysis (PCA), which mathematically reduces the shape dimensionality of the region of interest [13, 14]. Several studies utilized SSM to provide better understanding regarding the correlation between joint morphology and joint function, which enabled prognosis and determination of progression in humans for joint OA [15, 16]. With advanced imaging technologies such as computed tomography (CT) becoming more common in veterinary practice, equine SSM studies are now being published and in future could be used to guide therapies and evaluate risks based on morphologies [11, 12]. This study aimed to create statistical shape models of the equine femur and tibia to provide an overarching description of morphological variability in the equine stifle, which could serve as a baseline for future investigations on correlation between equine joint morphology and degenerative joint diseases such as equine stifle OA. The subjective of this study was to describe the shape variations of the equine femur and tibia using shape models within ±3 standard deviations (SD) in each mode, which is commonly used in SSM to capture 99.7% of the shape variation [9]. We hypothesize that the main variation within both femur and tibia models would be the scaling of the bones. Materials and methods Population This study was approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Florida (Approval Number 201809196, August 7th, 2019). Adult horses from the hospital and research populations were used for this prospective study. Horses were euthanized for reasons unrelated to any hindlimb injuries. Colic was the main reason for euthanasia for clinical animals, and owner permission was obtained prior to use in the current research project. Horses euthanized at the conclusion of their in vivo research studies were also utilized in this study. All animals were euthanized via intravenous lethal injection of pentobarbital sodium (Beuthanasia-D) at 0.2ml/kg. Adult horses from the hospital and research populations were used for this prospective study. Horses were euthanized for reasons unrelated to any hindlimb injuries. No restriction was placed on age, breed, sex or body mass for inclusion in this study, other than the animal was required to be an adult and fully mature, which was classified as greater than 5 years of age. Sample collection and imaging protocol All hindlimbs used were transected at the coxofemoral joint; no preference was given to laterality and the decision to harvest a left or right limb was based on a coin toss. Once harvested, limbs were scanned within 24 hours of collection. Bone algorithm volumetric data Digital Imaging and Communications in Medicine (DICOM) images of all specimens, from the proximal femoral head at the coxofemoral joint to the distal tibia at the tibiotarsal joint, were created using a 160-slice Toshiba Aquilion CT Scanner (slice thickness of 0.5 mm, and in-plane resolution of 0.3 mm). Once CT images were acquired, scans were reviewed to evaluate for any obvious pathology (e.g., subchondral bone cysts, osteophytes etc.) and excluded if any obvious pathology was present. Statistical shape modeling Segmentation for the femora and tibiae from DICOM data were performed in medical imaging software (Mimics 16.0, Materialise, Leuven, Belgium). Images were segmented using the program’s predefined bone threshold (i.e., 226 to 3017 Hounsfield units). The segmented femora and tibiae were then exported as 3D objects represented by triangular meshes into 3-Matic (Materialise, Leuven, Belgium). All right limbs were mirrored to the left to create the SSM model. All samples underwent minimal preprocessing, such as smoothing, to eliminate noise from the CT images while the native anatomic geometry was mostly preserved. Briefly, the 3D samples were processed with smoothing and wrap functions using a gap closing distance and minimal defect size of 0.2 mm. The original femur and tibia samples were represented by approximately 55000 to 95000 and 45000 to 75000 triangles of various sizes, respectively. To achieve better results of shape registration with scaling and fitting as well as minimizing the computational expense, femur and tibia samples were remeshed with respective uniform triangles of 4.5 mm and 1.5 mm average edge lengths, which enabled a reduction of triangles to approximately 14000 to 18000 and 11000 to 14000 for femora and tibiae, respectively. For each model, one sample was arbitrarily chosen to be the template for the first iteration of point correspondence registration. Alignment with scaling of the samples were carried out with each sample being automatically rigidly aligned to the template and scaled. To minimize the computational expense while maintaining good mesh fidelity, sample points were set at 75% with 50 iterations for each alignment. A warp function was used to create point correspondence between the template and the sample after the alignment. Procrustes alignment was carried out to eliminate the position differences and ensure only shape variations were captured. Two separate PCAs were carried out to generate mean shapes for the femur and tibia models as well as to extract all modes of variation. The number of modes obtained was limited to n-1, where n was the number of samples used for each model building [17]. To minimize the potential error resulting from the arbitrarily chosen template, the mean model generated by the first iteration of the statistical model was used as the new template for the subsequent shape model building (Fig 1). The same process was followed, and a leave-one-out analysis was performed to evaluate the generalizability of each model. All principal components were named “modes” in the current study, for instance, the first principal component of the equine femur shape model would be mode 1 of the femur model. 10.1371/journal.pone.0287381.g001 Fig 1 Pipeline to generate statistical shape model of the tibia from segmented 3D samples exported from Mimics. Image with blue and orange bones overlapping demonstrates non-rigid alignment between the template and the samples. The tibia statistical shape model (in green) created using the PCA generated mean model, was then used in the second iteration of SSM as a template. Shape analysis Based on anatomical landmarks, measurements were performed on the femur and tibia instances generated by the respective shape model within ±3SD (i.e., mean, ±1SD, ±2SD and ±3SD instances) in 3-Matic. A cumulative variation of 95% within the population was used to determine the number of modes in the shape models to be included. Shape analysis with detailed measurements were performed on regions of interest as outcome measures: full length of the bone, condylar dimensions, femoral head size and orientation, femoral trochlear tubercle orientation, anatomical and mechanical axes orientation, femoral version angle, tibial plateau angles in the coronal and sagittal planes. Femur biometrics For the femur shape model, the length of the femur was measured from the most proximal point at the greater trochanter to the most distal point at the lateral tubercle of the femoral trochlea in the sagittal plane within mode 1 (Fig 2A). The bicondylar width was measured in mode 1 from the utmost medial point to the utmost lateral point on the femoral condyles in a transverse view (Fig 2B). The medial condylar width and depth were taken from a caudal view and a medial view of the femur in mode 1, respectively (Fig 2C and 2D). A spherical approximation to the medial femoral condyle was created in mode 1 using the condylar surface, and the radius of the best-fit sphere was used as measurement for the femoral condylar curve (Fig 2E). Similarly, a spherical approximation to the femoral head surface was created to quantify the radius of the femoral head within mode 1 instances (Fig 2F). The anatomical axis was created using the surface of the femoral diaphysis. The orientation of the femoral neck was the angle between the femoral head direction, which was defined by using a best-fit function to the femoral neck surface, and the anatomical axis measured in mode 2 (Fig 3A) [18]. The center of the femoral condyles was determined to be the midpoint of the centers of the best-fit spheres to the medial and lateral condyles. The mechanical axis was determined by the direction of the line connecting the center of the femoral head and the center of the femoral condyles, and the femoral mechanical-anatomical (FMA) angle was analyzed using ±3SD instances in the coronal plane in mode 2 (Fig 3B) [19, 20]. To facilitate describing the morphology of the medial tubercle of the femoral trochlea demonstrated by the shape model, a best-fit sphere to the most prominent region of the medial tubercle was created on instances generated in mode 3. The estimated direction of the tubercle follows the line connecting the center of the femoral condyles and the center of the best-fit sphere of the tubercle (Fig 4A). In a transverse plane that is normal to the anatomical axis, the femoral version angle was measured between the axis of the femoral neck and the transcondylar axis on instances generated in mode 4 (Fig 4B) [21]. 10.1371/journal.pone.0287381.g002 Fig 2 Biometrics demonstrated on mode 1 instances generated by the femur SSM model: (2A) length of the femur, (2B) femoral bicondylar width, (2C, 2D) medial femoral condylar width and depth, (2E) estimation of the medial femoral condyle radius, and (2F) the femoral head radius. 10.1371/journal.pone.0287381.g003 Fig 3 Biometrics demonstrated on mode 2 instances generated by the femur SSM model: (3A) femoral neck angle and (3B) FMA angle. 10.1371/journal.pone.0287381.g004 Fig 4 Demonstration of (4A) femoral trochlear tubercle orientation on a mode 3 instance and (4B) femoral version angle on a mode 4 instance. Tibia biometrics In the tibia shape model, the length was measured in mode 1 from the most proximal point on the medial intercondylar eminence of the tibia (MICET) to the most distal point of the distal intermediate ridge of the tibia in the coronal view (Fig 5A). The bicondylar width was measured from the most medial point to the most lateral point on the periphery of the tibial plateau in a transverse plane on mode 1 generated instances (Fig 5B). From proximal to distal in a transverse view, the medial condylar width was measured from the most medial point of the tibial plateau to the border of the MICET, which was characterized by a curvature analysis performed in 3-Matic (Fig 5C). The medial condylar depth was the distance from the most cranial point to the most caudal point on the tibial plateau measured in the sagittal plane on mode 1 instances (Fig 5D). The coronal tibial plateau (CTP) angle, represented by the joint line connecting the most medial and most lateral point on the tibial plateau and the anatomical axis, was analyzed in a cranial view on instances generated in mode 2 (Fig 6A) [22]. The medial and lateral caudal tibial slope (CTS) angles were measured in mode 3, the slope was defined by the angle between the line perpendicular to the anatomical axis of the tibial diaphysis and the line connecting the most cranial and caudal point on the tibial plateau in the sagittal plane (Fig 6B) [23]. 10.1371/journal.pone.0287381.g005 Fig 5 Biometrics demonstrated on mode 1 instances generated by the tibia SSM model: (5A) length of the tibia, (5B) tibial bicondylar width, (5C) medial tibial condylar width, and (5D) medial tibial condylar depth. Note that in 5C the border was defined by a Gaussian curvature analysis, where the curvature in each point is calculated based on the local minimum and maximum values using a Gaussian interpolation. The higher the curvature value, the more curved the region is. The triangle edge/point selected to be the border between the medial tibial plateau and MICET was determined by the mean +1SD curvature value, and the line (solid blue line) connecting the edge and the most medial point was perpendicular to the line directed from the most caudal to the most cranial point of the tibial plateau border defined by the curvature analysis shown as the blue dotted line. 10.1371/journal.pone.0287381.g006 Fig 6 Demonstration of (6A) CTP angle on a mode 2 instance and (6B) CTS angle on a mode 3 instance. Statistics To evaluate the correlations between the observed biometrics and the corresponding mode of variation, correlation analyses using Pearson’s correlation coefficient R with statistical significance set at p<0.05 were performed using 100 shape model generated femur instances with 4 modes at arbitrary standard deviations. The same analyses were also performed on 100 tibia instances with 3 modes at randomly generated standard deviations. In addition, a Pearson’s correlation analysis and a paired T test with statistical significance set at p < 0.05 was performed to evaluate the relationships between the CTS angle in the medial and lateral compartments. Results Fifteen equine hindlimbs with appropriate institutional approval were used in the study. One Morgan horse, two Quarter Horses and twelve Thoroughbred (9 females, 1 stallion and 5 geldings; mean age ± SD = 14.3 ± 4.6 years old; weighing 488.7 ± 46.8 kg) were used. Only entire femora or entire tibiae were used; one tibia that was sectioned at the mid-diaphysis was excluded, thus fifteen femora and fourteen tibiae were used. All specimens were examined through CT images and displayed no signs of injury or obvious deformation in the stifle joint. Model evaluation Cumulative variance accounted for by ordered modes decreases for both femur and tibia statistical shape models (Fig 7). With 6 and 3 modes, approximately 95% of the total geometric variations in the sampled population were described for the femur and tibia shape models, respectively. The leave-one-out analysis for both shape models showed a decrease in root-mean-square error as the number of modes used in the prediction model increases (Fig 8). The average geometric error for femur and tibia shape models were 2.12 mm and 1.37 mm, respectively. 10.1371/journal.pone.0287381.g007 Fig 7 Cumulative variation explained by the femur and tibia statistical shape models. 10.1371/journal.pone.0287381.g008 Fig 8 Boxplot of mean geometric error evaluated in the leave-one-out analysis for (8A) femur and (8B) tibia statistical shape models as a function of the number of modes in the model prediction. Error was calculated using the part comparison function in 3-matic (Mimics) that evaluated the mean geometric distance between the left-out model and the prediction model. Femur shape model In the equine femur statistical shape model (Fig 9), with 6 modes, 95.7% of the total variation was described (Table 1). Mode 1 mainly described the shape differences in scaling of all structures, which in total accounted for approximately 81.6% of the variation (Table 2). All biometrics showed strong positive correlations with the mode 1. The length of the mean femur instance was 460.6 mm and the bicondylar width was 106.7 mm. The medial condyle showed a width of 41.6 mm and a depth of 67.8 mm, respectively. Moreover, the medial condyle demonstrated a best-fit radius of 36.0 mm, and the femoral head showed a similar radius of 36.1 mm. Mode 2, which accounted for 4.8% of the total variation, demonstrated the differences in the FMA angle and the femoral neck angle (Table 3). Both biometrics demonstrated moderate correlations with the mode 2 with FMA angle being negatively correlated. Mode 3 displayed shape changes in the medial tubercle of the femoral trochlea and a moderate positive correlation with the mode 3, accounting for 3.4% of the total variation (Table 4). Mode 4 accounted for 2.5% of total variation and captured the shape differences in femoral version angle (Table 5). A weak correlation was shown between the femoral version angle and the mode 4. 10.1371/journal.pone.0287381.g009 Fig 9 Instances of the femur statistical model over ±3SD for biometrics in 4 different modes. Note that in each graph, the measurement indicator illustrations were created to emphasize the morphology rather than showing accurate measurements. 10.1371/journal.pone.0287381.t001 Table 1 Cumulative variability percentages and description of mode characteristics for the femoral shape model. Mode CUMULATIVE VARIABILITY (%) MODE DESCRIPTION 1 81.6 Scaling 2 86.4 FMA and femoral neck angle 3 89.8 Femoral trochlear tubercle orientation 4 92.2 Femoral version angle 5 94.1 Third trochanter and lesser trochanter width 6 95.7 Third trochanter position 10.1371/journal.pone.0287381.t002 Table 2 Equine femur model measurements in mode 1. Measurement -3SD -2SD -1SD MEAN 1SD 2SD 3SD R Length (mm) 400.9 419.5 439.3 460.6 476.7 496.5 515.9 0.993*** Bicondylar width (mm) 91.7 96.7 101.1 106.7 111.9 117.1 122.4 0.869*** Medial condylar width (mm) 35.8 38.1 41.3 41.6 42.8 45.5 46.0 0.849*** Medial condylar depth (mm) 57.1 60.8 63.8 67.8 71.3 74.4 78.1 0.846*** Medial condylar curve (mm) 29.5 31.8 33.1 36.0 36.3 38.9 40.3 0.814*** Femoral head radius (mm) 30.4 31.9 33.8 36.1 37.4 39.0 40.9 0.889*** *** P < 0.001 10.1371/journal.pone.0287381.t003 Table 3 Equine femur model measurements in mode 2. Measurement -3SD -2SD -1SD MEAN 1SD 2SD 3SD R FMA angle (°) 8.6 8.0 7.8 6.7 5.8 5.3 4.6 -0.457* Femoral neck angle (°) 144.6 149.1 146.6 157.3 161.3 157.3 156.8 0.432*** * P < 0.05 *** P < 0.001 10.1371/journal.pone.0287381.t004 Table 4 Equine femur model measurements in mode 3. Measurement -3SD -2SD -1SD MEAN 1SD 2SD 3SD R Angle between femoral trochlear tubercle and the anatomical axis (°) 10.7 14.2 13.7 15.3 18.5 18.3 18.2 0.314** ** P < 0.01 10.1371/journal.pone.0287381.t005 Table 5 Equine femur model measurements in mode 4. Measurement -3SD -2SD -1SD MEAN 1SD 2SD 3SD R Femoral version angle (°) 37.3 33.8 38.9 43.5 44.1 45.9 55.0 0.250* * P < 0.05 Tibia shape model In the tibia statistical shape model (Fig 10), approximately 95% of the total shape variation within the tibia population was captured with 3 modes (Table 6). Mode 1 described the scaling of the tibia, accounting for 91.2% of the total variation (Table 7). All biometrics observed within the mode displayed strong positive correlations with mode 1. Mode 2 characterized the CTP angle, accounting for 2.3% of the variation (Table 8). A significantly positive strong correlation was seen between the CTP angle and the mode 2. Mode 3 described the CTS angles, which represented 1.3% of the total variation (Table 9). CTS angle in the medial compartment was moderately correlated with the mode 3 and the correlation was significant, whereas the lateral CTS angle demonstrated a non-significant weak correlation. However, there was a strong positive correlation between the medial and lateral CTS angles (R = 0.774, p < 0.001), with the lateral CTS angle being significantly larger than the medial (difference mean ± SD: 5.07° ± 0.37°, p < 0.001). 10.1371/journal.pone.0287381.g010 Fig 10 Instances of the tibia statistical model over ±3SD for biometrics in 3 different modes. 10.1371/journal.pone.0287381.t006 Table 6 Cumulative variability percentages and description of mode characteristics for the tibial shape model. Mode CUMULATIVE VARIABILITY (%) MODE DESCRIPTION 1 91.2 Scaling 2 93.5 CTP angle 3 94.8 CTS angle 10.1371/journal.pone.0287381.t007 Table 7 Equine tibia model measurements in mode 1. Measurement -3SD -2SD -1SD MEAN 1SD 2SD 3SD R Length (mm) 358.2 374.8 392.4 410.3 427.8 446.8 464.7 0.998*** Bicondylar width (mm) 104.2 109.0 111.4 118.3 123.7 127.8 133.3 0.922*** Medial condylar width (mm) 32.6 35.1 35.1 35.8 41.5 44.6 44.9 0.797*** Medial condylar depth (mm) 61.6 72.5 75.8 79.7 83.9 86.8 90.3 0.740*** *** P < 0.001 10.1371/journal.pone.0287381.t008 Table 8 Equine tibia model measurements in mode 2. Measurement -3SD -2SD -1SD MEAN 1SD 2SD 3SD R CTP angle (°) 87.1 88.4 88.4 88.9 89.2 89.8 90.5 0.815*** *** P < 0.001 10.1371/journal.pone.0287381.t009 Table 9 Equine tibia model measurements in mode 3. Measurement -3SD -2SD -1SD MEAN 1SD 2SD 3SD R Medial CTS angle (°) 7.4 8.0 9.3 9.6 10.9 10.9 12.4 0.529* Lateral CTS angle (°) 14.4 15.7 16.3 17.6 18.6 19.6 20.9 0.142 *** P < 0.001. Discussion The current study is the first report on inter-subject variability in equine femur and tibia geometry using SSM methods. The equine femur and tibia statistical shape models can be powerful tools for future studies as they provide a wide range of functionalities: facilitating bone reconstruction, implant design and preoperative surgical planning. The statistical shape models support our hypothesis that the main mode of variation in the equine femur and tibia is due to scaling. The biometrics of the mean femur and tibia models could serve as a baseline for future references in describing average equine stifle anatomy as well as facilitating design of surgical tools and techniques such as stifle and hip arthroplasties. In mode 1 of the femur model, the length of the femur increased by approximately 29% from -3SD to +3SD, and the femoral condyles demonstrated a similar increase of 33% in bicondylar width. Comparably, the medial condyle showed a 28% increase in condylar width. A 37% increase in condylar depth and in condylar radius were observed in the medial femoral condyle, demonstrating the largest proportional changes in the femur shape model. In mode 2, the femur shape model described variations in the FMA angle. Joint alignment, one of the mainstays of successful arthroplasties in human knees, requires the knowledge of the FMA angle in healthy joints, which was reported to range from 2.6° to 7.4° with a mean of 5.1° [24–27]. The current study displayed a slightly higher mean FMA angle of 6.7° ranging from 4.6° to 8.6°. In mode 3, shape variation was observed in the tubercle of the femoral trochlea. The angle between the direction of the tubercle and the anatomical axis increased by 12.4° from 6.6° to 19° within ±3SD. The result displays a wide range of trochlear positions, which could have some great impact on the position and alignment of the patella. As the direction of the tubercle aligned closer to the anatomical axis (tubercle positions more vertically), the patella would sit more laterally, which could increase the tension and stress in the medial patellar ligament resulting ligamentous tears or being more prone to injury. In mode 4, the version angle increased from 33.8° to 55° within ±3SD. In humans, the femoral neck anteversion is an essential indicator of torsion in the femur, which varies up to 30° in adults and affects the biomechanics of the knee and hip [28]. Future investigations are needed to determine the effect of different femoral version angle on the equine stifle and hip joint kinematics and kinetics. In the tibia shape model, mode 1 accounted for 91.2% of the variation within the population. As the length of the tibia increased by approximately 30%, the bicondylar width and the medial condylar width increased by 28% and 38%, respectively. The medial condylar depth displayed a more sizable growth of 47% as the tibia length increased. These results showed that the width of the tibial plateau experienced a more similar scale growth to the tibia size, whereas the tibial condylar depth showed a higher upscaling rate than the width. Mode 2 characterizes the geometric variation in equine CTP angle. A number of studies have suggested CTP angle plays an important role in describing tibia anatomy and the joint alignment in human knees [29, 30]. In the current study, the shape model displayed a 3.4° variation in CTP angle with the mean instance demonstrating 1.1° in valgus. In mode 3, the CTS angle ranged from 7.4° to 12.4° and 14.4° to 20.9° in the medial and lateral compartments, respectively. As shown in the results, the CTS in medial and lateral compartments are positively correlated, with the lateral CTS angle being significantly larger than the medial. In human knees, the description of the posterior tibial slope was well-documented, and reports have suggested the range of the posterior tibial slope varied between races, sexes and measuring methods used [30–33]. Multiple studies have shown strong correlations between the increased posterior tibial slope and anterior cruciate ligament injury [34, 35]. In canine stifles, the CTS angle, which was reported by several groups to be 22.6° or greater, was much more acute than what was seen in equine stifles and human knees [36, 37]. Similarly, increased tibial slope was believed to predispose dogs to cranial cruciate ligament rupture via excessive cranial translation of the tibia [36, 38]. Therefore, future studies could utilize results derived from the current study as a baseline to determine the impact of an increased CTS on cranial cruciate ligament injury due to altered biomechanics in equine stifles. One possible explanation for the extremely large percentage of the shape variation that mode 1 in current shape models accounted for was the size of the bones and scattered regions of interest. A common practice for modeling large bone structures with concentrated region of interest is performing customized transections on the samples, by which the investigated morphological variables would not be affected [39, 40]. However, the current study aimed to establish a comprehensive description on the femur and tibia morphology. The anatomical structures such as the diaphysis of the femur and tibia were essential to establish the anatomical and mechanical axis, thus all bone samples were intact. It is conceivable that a focus on the femoral condyles or tibial plateau would result in variation in different modes other than scaling. One limitation to the current study is the limited sample size, which could also have contributed to the large percentage variation in mode 1 in both femur and tibia shape models. It is important to note that the modes of variation and the percentage each mode accounted for are dependent on the sampled population. The current study utilized a relatively homogenous population with 12 out of 15 specimens being Thoroughbred. Therefore, in addition to larger sample size, it would be beneficial to include more stifles from different horse breeds. Additionally, although FMA angle and CTP angle are important biometrics for investigating relationships between the joint alignment and joint disorders due to altered biomechanics, a model incorporating intact femorotibial joints would be more suited for characterizing the variability in joint alignment. Another limitation is the potential error resulted from geometric differences in the left and right equine stifles. Prior literature has shown that the mean bilateral difference in all morphological variables in canine femurs were less than 5%, thus it was assumed that errors were minimal when all right limbs were mirrored to the left in the current study [41]. The main objective of the models constructed in the current study is to provide a quantitative description of the equine femur and tibia morphological variability. As the current imaging technology does not have the capacity to allow clinicians to CT scan an equine stifle in vivo, the models presented in the current study could possibly assist the prediction of a patient-specific bone geometry. By using different combination of modes in the current shape models, a fitting prediction geometry would be generated based on the radiographic projections. Once the matching is complete, a 3D representation of the patient’s bone would be available for 3D printing, which could help clinicians to visualize or prepare for the surgical treatment. Conclusion The current study presented statistical shape models of the equine femur and tibia, describing morphological variability within the sampled population in a compact representation. These population models can serve as a baseline for future studies that investigate the relationship between equine stifle morphology and joint disorders such as OA and injuries. Furthermore, development of novel stifle arthroplasties, implant design, and surgical planning could be assisted by using shape model generated instances. This work was supported by Surgical Translation And 3D Printing Research Lab and Gary J. Miller Ph.D. Orthopedic Biomechanics Lab at University of Florida. 10.1371/journal.pone.0287381.r001 Decision Letter 0 Shaheen Aliah Faisal Academic Editor © 2023 Aliah Faisal Shaheen 2023 Aliah Faisal Shaheen https://creativecommons.org/licenses/by/4.0/ This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Submission Version0 23 Mar 2023 PONE-D-22-28433Anatomical variations of the equine femur and tibia using statistical shape modelingPLOS ONE Dear Dr. Biedrzycki, Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process. Please note that I have taken the decision to go ahead with a single review given the length of time it has taken to secure a review. Can you please address the comments raised by the reviewer? Please submit your revised manuscript by May 07 2023 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file. Please include the following items when submitting your revised manuscript:A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter. If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols. We look forward to receiving your revised manuscript. Kind regards, Aliah Faisal Shaheen Academic Editor PLOS ONE Journal Requirements: When submitting your revision, we need you to address these additional requirements. 1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf 2. Please provide additional information regarding the euthanasia of the animals included in the study and provide the reasons why the animals were euthanised. 3. Please include your animal ethics statement in the Methods section of your manuscript. In the Methods section of your revised manuscript, please include the animal ethics information you provided in the Cover Letter of your manuscript, the full name of the IACUC that approved the protocol, the approval or permit number that was issued, and the date that approval was granted. 4. Thank you for stating in your Funding Statement: “HH The project was partially funded by the University of Florida Graduate Student Fellowship Award. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.” Please provide an amended statement that declares *all* the funding or sources of support (whether external or internal to your organization) received during this study, as detailed online in our guide for authors at http://journals.plos.org/plosone/s/submit-now.  Please also include the statement “There was no additional external funding received for this study.” in your updated Funding Statement. Please include your amended Funding Statement within your cover letter. We will change the online submission form on your behalf. Additional Editor Comments (if provided): Given the length of time it has taken to find reviewers for this article, I have taken the decision to go ahead with a single review. Please address the comments of the reviewer, including the rationale for developing an isolated model of the femur and tibia and technical comments related to element size. [Note: HTML markup is below. Please do not edit.] Reviewers' comments: Reviewer's Responses to Questions Comments to the Author 1. Is the manuscript technically sound, and do the data support the conclusions? The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented. Reviewer #1: Yes ********** 2. Has the statistical analysis been performed appropriately and rigorously? Reviewer #1: Yes ********** 3. Have the authors made all data underlying the findings in their manuscript fully available? The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified. Reviewer #1: No ********** 4. Is the manuscript presented in an intelligible fashion and written in standard English? PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here. Reviewer #1: Yes ********** 5. Review Comments to the Author Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters) Reviewer #1: The authors present the results of a statistical shape analysis of the femur and tibia of the horse, which has been the focus of limited prior equine research. What is the rationale for focusing on models of the femur and tibia in isolation, rather than developing a combined model that would describe full limb variation, or more specifically, stifle morphology? This would be more beneficial to the stated application of assisting with virtual surgical planning and treatment of OA. Line 245: Exactly how was the geometric error defined/quantified? How did these errors change when they were focused around the primary region of interest (stifle) rather than the full bone (which I would expect to have very small errors along the shaft of the bone, which would bias the reported error towards smaller values)? Fig. 8: Inclusion of additional modes seems to have a fairly small effect on the model accuracy. Why do the authors think this is? What are the leave-one-out errors if the subject-specific models were instead just compared to a (rigid) scaled version of the mean geometry? It may be that no particularly useful information (apart from size) is gained from the PCA. “femur and tibia samples were remeshed with uniform triangles of 4.5 mm and 1.5 mm average edge lengths”. Is not clear if this means femurs were meshed with edge length of 4.5mm and tibia were meshed with edge length of 1.5mm. If this is the case, what was the difference for the meshing discrepancy between bones? If each femoral mesh (for example) has a element size of 4.5 mm, how have the authors accounted for variability that occurs as a result of the nodal points not being directly aligned (i.e. noise) rather than true variation between bones? Line 30-31: This is a not result and does not add value to the abstract. I suggest removing or reframing more clearly as a hypothesis. The proposed usage for the model/data is quite vague. Clearer applications of how this work would be applied would be more helpful. Fig. 5C. It is not clear what the reader should make of the colormap data presented in this figure. I think this represents the curvature of the mean geometry, but there is no information about how curvature changes with changing modes of variation, so it is difficult to see how this contributes to the rest of the analysis. Lines 224-227: Why were correlation analyses limited to the relationship between CTP and medial and lateral compartments, rather than comparing the PC scores with each of the variables qualitatively identified as being representative of variation with a mode? ********** 6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files. If you choose “no”, your identity will remain anonymous but your review may still be made public. Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy. Reviewer #1: No ********** [NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.] While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step. 10.1371/journal.pone.0287381.r002 Author response to Decision Letter 0 Submission Version1 8 May 2023 Response to Reviewers Dear Editor, Thank you for giving us the opportunity to submit a revised draft of the manuscript “Anatomical variations of the equine femur and tibia using statistical shape modeling” for publication in the PLOS ONE. We appreciate the time and effort that you and the reviewer dedicated to providing feedback on our manuscript and are grateful for the insightful comments on and valuable improvements to our paper. We have incorporated most of the suggestions made by the reviewers. Those changes are highlighted within the manuscript. A point-by-point response to the reviewers’ comments and concerns was included below. Reviewer #1: 1. The authors present the results of a statistical shape analysis of the femur and tibia of the horse, which has been the focus of limited prior equine research. What is the rationale for focusing on models of the femur and tibia in isolation, rather than developing a combined model that would describe full limb variation, or more specifically, stifle morphology? This would be more beneficial to the stated application of assisting with virtual surgical planning and treatment of OA. Author response: Thank you for the question. First, building a shape model of the full limb would require consistent joint pose acquired by CT scans, which can be challenging. Second, the current study aims to focus on providing a comprehensive description of the bone morphology of the equine femur and tibia. Including the full limb as a shape model would introduce pose variations, which are irrelevant to the goal of the current study. 2. Line 245: Exactly how was the geometric error defined/quantified? How did these errors change when they were focused around the primary region of interest (stifle) rather than the full bone (which I would expect to have very small errors along the shaft of the bone, which would bias the reported error towards smaller values)? Author response: Thank you for the questions. The leave-one-out (LOO) analysis is a standard test that is commonly used for testing the generalizability of the shape model, which indicates how well the model can be generalized to new subjects. The geometric error in line 245 was referring to the LOO result, which was defined by comparing the instance generated by the prediction model with the left-out model. There are two main sources for the reported error: inaccuracy in the alignment and the model prediction. Before implementing the part comparison analysis, a global registration tool in 3-Matic was used for rigid alignment through translation and rotation. Then, the error was determined using the part comparison that calculates the mean distance between all vertices of the two models. The final result reported was the mean of the errors derived from all prediction models. Therefore, both inaccuracy in the alignment and the SSM’s prediction would contribute to geometric errors. The reviewer expected the error to be larger if the primary region of interest was the stifle rather than the full bone because of expected smaller errors along the shaft, this is not necessarily true based on what we observed from the models. In fact, most of the errors were seen at locations beyond the stifles. For example, multiple comparison showed maximum distance at the tibial malleolus and the femoral supracondylar fossa. Therefore, if the model only focused on the stifle, the overall error would be expected to be smaller. 3. Fig. 8: Inclusion of additional modes seems to have a fairly small effect on the model accuracy. Why do the authors think this is? What are the leave-one-out errors if the subject-specific models were instead just compared to a (rigid) scaled version of the mean geometry? It may be that no particularly useful information (apart from size) is gained from the PCA. Author response: Thank you for the questions. The reason why the inclusion of additional modes did not have a huge impact on the overall accuracy is because of the large scale of the bone models. This was reflected in the variation of modes as mode 1 in the femur and tibia shape models accounted for 81.6% and 91.2%, respectively. Moreover, the percentage that the subsequent modes accounted for decreases as the number of modes increases, which means the variation in subsequent modes were much smaller in magnitude compared to scaling. LOO analysis was conducted to test the generalizability of the shape models using N-1 samples. So, they are essentially shape models with one fewer sample. The results showed when using the models to fit to new instances, one might expect an average geometric error of 2.12 mm and 1.37 mm when using the femur and tibia shape models with 14 and 13 PC modes, respectively. This result should be similar when the LOO model was to compare to a rigidly scaled mean geometry if the scaled model does not exceed ± 3SD in each mode. The statistical shape model describes a collection of semantically similar objects. In our case, the shape models were built using a relatively homogenous breed of horses. If the model was to predict the geometry of the femur or the tibia of a pony, it would fail to generate an accurate instance. Moreover, the comparison would likely not be meaningful as the 3D representation of a bone model is not just consist of a set of scaled biometrics. PCA is a technique that extracts the modes of variations from a large dataset and reduces the dimensionality of the variables, which enables us to characterize complex geometries like the femur and tibia in an extremely compact manner. The biometrics reported in the current study by no means concludes all the variations encoded in each mode. Although the subsequent modes explained for much smaller percentages of the total variation compared to scaling, the morphological variations in anatomical angulation observed are still valuable and could be used for explaining joint pathology. For example, the coronal tibial slope is associated with tibiofemoral joint alignment, which is considered a major risk factor for developing osteoarthritis in human knees. Only 3.4° of difference was seen between the -3SD and +3SD instances in our model, thus there’s a homogeneity present in the equine tibia, which could explain the lack of varus/valgus joint deformity in equine stifles. Another advantage of using PCA is the ability to localize the anatomical differences. With every mode of variation being independent of each other, we can identify the shape changes that each mode corresponds to through manipulating the standard deviation. 4. “femur and tibia samples were remeshed with uniform triangles of 4.5 mm and 1.5 mm average edge lengths”. Is not clear if this means femurs were meshed with edge length of 4.5mm and tibia were meshed with edge length of 1.5mm. If this is the case, what was the difference for the meshing discrepancy between bones? If each femoral mesh (for example) has an element size of 4.5 mm, how have the authors accounted for variability that occurs as a result of the nodal points not being directly aligned (i.e. noise) rather than true variation between bones? Author response: Thank you for pointing this out. The texts have been revised in materials and methods line 125: “…femur and tibia samples were remeshed with respective uniform triangles of 4.5 mm and 1.5 mm average edge lengths.” As the femurs have much larger surface than the tibiae, more triangles are required to accurately represent the geometry of the femurs. The larger the edge length of the mesh is, the fewer triangles we need for meshing. Literature on statistical shape model of the full human femur had used approximately 8,000 nodes to identify the geometric variations. Therefore, we aimed for a similar number of nodes on the femur and tibia samples. As mentioned in materials and methods line 130, 4.5 mm edge length for the femur mesh and 1.5 mm for the tibia mesh were used and yielded approximately 14,000 to 18,000 (approximately 8000 nodes) and 11,000 to 14,000 (approximately 7000 nodes) triangles for femora and tibiae, respectively. Multiple steps were carried out after remeshing to minimize the geometric errors resulted from alignment. All samples were rigidly aligned with scaling to the template through 50 iterations using 75% of the sample points after remeshing. As for the warp errors, also referred as elastic registration errors in some literature, they were negligibly small and often were not included. In our model, the unsigned errors were 0.184 ± 0.061 mm and 0.151 ± 0.045 mm in the femur and tibia shape models, respectively. Finally, the Procrustes alignment method was used (integrated in create SSM function in 3-Matic) before PCA to eliminate position differences and ensure only shape variations were captured. The texts from 131 to 136 were revised to clarify the preprocess preparing for the PCA: “Alignment with scaling of the samples were carried out with each sample being automatically rigidly aligned to the template and scaled. To minimize the computational expense while maintaining good mesh fidelity, sample points were set at 75% with 50 iterations for each alignment. A warp function was used to create point correspondence between the template and the sample after the alignment. Procrustes alignment was carried out to eliminate the position differences and ensure only shape variations were captured.” 5. Line 30-31: This is a not result and does not add value to the abstract. I suggest removing or reframing more clearly as a hypothesis. Author response: Thank you for the suggestion. We absolutely agree and the text has been removed as suggested. 6. The proposed usage for the model/data is quite vague. Clearer applications of how this work would be applied would be more helpful. Author response: Thank you for the suggestion. Currently, the model was used to generated multiple instances for facilitating designing implants and surgical instruments for an unicompartmental equine stifle arthroplasty. Moreover, the model can be used in assisting equine anatomy education. As discussed in line 396 to 400, for the use of the patient-specific bone matching, prediction model can be generated using different combination of modes in statistical shape model. Then the prediction geometry will be used to match the radiograph of the patient. Once matching is complete, the SSM can export the 3D model of the prediction geometry for clinicians to observe and practice on. Lines from 397 to 400 were revised as followed: “the models presented in the current study could possibly assist the prediction of a patient-specific bone geometry. By using different combination of modes in the current shape models, a fitting prediction geometry would be generated based on the radiographic projections. Once the matching is complete, a 3D representation of the patient’s bone would be available for 3D printing, which could help clinicians to visualize or prepare for the surgical treatment. ” Another helpful application is the model can serve as a representation of the healthy equine stifle population. Future studies can build a shape model using stifles with joint pathologies. By comparing the healthy model and the joint model with pathologies, this could help identify the abnormal anatomical variations that contribute to the pathologies. 7. Fig. 5C. It is not clear what the reader should make of the colormap data presented in this figure. I think this represents the curvature of the mean geometry, but there is no information about how curvature changes with changing modes of variation, so it is difficult to see how this contributes to the rest of the analysis. Author response: Thank you for your question. The colormap in Fig 5C indeed represents the curvature of the medial tibial plateau of the mean geometry. As there is no clear definition where the medial tibial plateau ends, it is challenging to define the width of the medial tibial plateau manually without introducing human errors. Therefore, we decided to use the curvature as an indicator with an automated algorithm to consistently separate the plateau and the MICET. Because the curvature analysis was only used to identify the border of the medial plateau, the magnitude change of the curvature was irrelevant. Furthermore, all curvature analysis was carried out on mode 1 instances as mode 1 characterizes the scaling of the condylar dimensions. The use of the curvature analysis for determining the width of the medial tibial plateau was stated in line 201 to 203. 8. Lines 224-227: Why were correlation analyses limited to the relationship between CTP and medial and lateral compartments, rather than comparing the PC scores with each of the variables qualitatively identified as being representative of variation with a mode? Author response: Thank you for your question. First of all, we want to apologize for the typo in statistics line 247, the correlation analysis was performed on the CTS in the medial and lateral compartments as shown in the results instead of the CTP. The text has been revised in the manuscript. Second, the initial rationale for only including the CTS correlation was to emphasize the interest on tibial slope comparison as it’s an important biometric associated with joint stability and joint kinematic performance in many species. However, the authors agree that it’s a great suggestion to add correlation analysis between the PC score and each biometrics. Thus, we generated 100 arbitrary femur and 100 tibia instances using the shape models with randomized standard deviation in each mode, the biometrics in each instance were measured using a script in MATLAB. Pearson’s correlation coefficient R and the p-value were calculated all analyses. An additional Pearson’s correlation analysis and a paired T test were performed to evaluate the relationships between the medial and lateral CTS angles. The manuscript in materials and methods statistics line 246 to 254 and in results were revised. Attachment Submitted filename: Response to Reviewers.docx Click here for additional data file. 10.1371/journal.pone.0287381.r003 Decision Letter 1 Shaheen Aliah Faisal Academic Editor © 2023 Aliah Faisal Shaheen 2023 Aliah Faisal Shaheen https://creativecommons.org/licenses/by/4.0/ This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Submission Version1 5 Jun 2023 Anatomical variations of the equine femur and tibia using statistical shape modeling PONE-D-22-28433R1 Dear Dr. Biedrzycki, We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements. Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication. An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org. If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org. Kind regards, Aliah Faisal Shaheen Academic Editor PLOS ONE Additional Editor Comments (optional): Thank you for addressing the reviewers' comments. The manuscript is now accepted for publication. Reviewers' comments: Reviewer's Responses to Questions Comments to the Author 1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation. Reviewer #1: All comments have been addressed ********** 2. Is the manuscript technically sound, and do the data support the conclusions? The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented. Reviewer #1: (No Response) ********** 3. Has the statistical analysis been performed appropriately and rigorously? Reviewer #1: (No Response) ********** 4. Have the authors made all data underlying the findings in their manuscript fully available? The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified. Reviewer #1: (No Response) ********** 5. Is the manuscript presented in an intelligible fashion and written in standard English? PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here. Reviewer #1: (No Response) ********** 6. Review Comments to the Author Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters) Reviewer #1: (No Response) ********** 7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files. If you choose “no”, your identity will remain anonymous but your review may still be made public. Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy. Reviewer #1: No ********** 10.1371/journal.pone.0287381.r004 Acceptance letter Shaheen Aliah Faisal Academic Editor © 2023 Aliah Faisal Shaheen 2023 Aliah Faisal Shaheen https://creativecommons.org/licenses/by/4.0/ This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. 22 Jun 2023 PONE-D-22-28433R1 Anatomical variations of the equine femur and tibia using statistical shape modeling Dear Dr. Biedrzycki: I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department. If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org. If we can help with anything else, please email us at plosone@plos.org. Thank you for submitting your work to PLOS ONE and supporting open access. Kind regards, PLOS ONE Editorial Office Staff on behalf of Dr. Aliah Faisal Shaheen Academic Editor PLOS ONE ==== Refs References 1 Fowlie JG , Richardson DW , Ortved KF . Chapter 101—Stifle. In: Auer JA , Stick JA , Kümmerle JM , Prange T , editors. Equine Surgery (Fifth Edition). W.B. Saunders; 2019. 2 D Lasalle J , Alexander K , Olive J , Laverty S . Comparisons Among Radiography, Ultrasonography and Computed Tomography for Ex Vivo Characterization of Stifle Osteoarthritis in the Horse. Veterinary Radiology & Ultrasound. 2016;57 (5 ):489–501. doi: 10.1111/vru.12370 27237699 3 Auer JA , Fackelman GE , Gingerich DA , Fetter AW . Effect of hyaluronic acid in naturally occurring and experimentally induced osteoarthritis. Am J Vet Res. 1980 Apr 1;41 (4 ):568–74. 7406275 4 Bolam CJ , Hurtig MB , Cruz A , McEwen BJE . Characterization of experimentally induced post-traumatic osteoarthritis in the medial femorotibial joint of horses. American Journal of Veterinary Research. 2006 Mar 1;67 (3 ):433–47. doi: 10.2460/ajvr.67.3.433 16506905 5 Matsuda S , Miura H , Nagamine R , Mawatari T , Tokunaga M , Nabeyama R , et al . Anatomical analysis of the femoral condyle in normal and osteoarthritic knees. Journal of Orthopaedic Research. 2004;22 (1 ):104–9. doi: 10.1016/S0736-0266(03)00134-7 14656667 6 Beynnon B , Yu J , Huston D , Fleming B , Johnson R , Haugh L , et al . A Sagittal Plane Model of the Knee and Cruciate Ligaments With Application of a Sensitivity Analysis. Journal of Biomechanical Engineering. 1996 May 1;118 (2 ):227–39. doi: 10.1115/1.2795965 8738789 7 Hashemi J , Chandrashekar N , Gill B , Beynnon BD , Slauterbeck JR , Schutt RC , et al . The Geometry of the Tibial Plateau and Its Influence on the Biomechanics of the Tibiofemoral Joint. J Bone Joint Surg Am. 2008 Dec 1;90 (12 ):2724–34. doi: 10.2106/JBJS.G.01358 19047719 8 Masouros SD , Bull AMJ , Amis AA . (i) Biomechanics of the knee joint. Orthopaedics and Trauma. 2010 Apr 1;24 (2 ):84–91. 9 Rusli WMR , Kedgley AE . Statistical shape modelling of the first carpometacarpal joint reveals high variation in morphology. Biomech Model Mechanobiol. 2020 Aug 1;19 (4 ):1203–10. doi: 10.1007/s10237-019-01257-8 31754950 10 Fitzpatrick CK , Baldwin MA , Laz PJ , FitzPatrick DP , Lerner A L. , Rullkoetter PJ . Development of a statistical shape model of the patellofemoral joint for investigating relationships between shape and function. Journal of Biomechanics. 2011 Sep 2;44 (13 ):2446–52. doi: 10.1016/j.jbiomech.2011.06.025 21803359 11 Liley H , Zhang J , Firth EC , Fernandez JW , Besier TF . Statistical modeling of the equine third metacarpal bone incorporating morphology and bone mineral density. PLOS ONE. 2018 Jun 6;13 (6 ):e0194406. doi: 10.1371/journal.pone.0194406 29874224 12 Van Houtte J , Vandenberghe F , Zheng G , Huysmans T , Sijbers J . EquiSim: An Open-Source Articulatable Statistical Model of the Equine Distal Limb. Frontiers in Veterinary Science. 2021 Mar 3;8 :623318. doi: 10.3389/fvets.2021.623318 33763462 13 Cootes TF , Cooper DH , Taylor CJ , Graham J . Trainable method of parametric shape description. Image and Vision Computing. 1992 Jun 1;10 (5 ):289–94. 14 Lorenz C , Krahnstöver N . Generation of Point-Based 3D Statistical Shape Models for Anatomical Objects. Computer Vision and Image Understanding. 2000 Feb 1;77 (2 ):175–91. 15 Gregory JS , Waarsing JH , Day J , Pols HA , Reijman M , Weinans H , et al . Early identification of radiographic osteoarthritis of the hip using an active shape model to quantify changes in bone morphometric features: Can hip shape tell us anything about the progression of osteoarthritis? Arthritis & Rheumatism. 2007;56 (11 ):3634–43. doi: 10.1002/art.22982 17968890 16 Haverkamp DJ , Schiphof D , Bierma-Zeinstra SM , Weinans H , Waarsing JH . Variation in joint shape of osteoarthritic knees. Arthritis & Rheumatism. 2011;63 (11 ):3401–7. doi: 10.1002/art.30575 21811994 17 Bredbenner TL , Eliason TD , Francis WL , McFarland JM , Merkle AC , Nicolella DP . Development and Validation of a Statistical Shape Modeling-Based Finite Element Model of the Cervical Spine Under Low-Level Multiple Direction Loading Conditions. Frontiers in Bioengineering and Biotechnology. 2014. doi: 10.3389/fbioe.2014.00058 25506051 18 Doherty M , Courtney P , Doherty S , Jenkins W , Maciewicz RA , Muir K , et al . Nonspherical femoral head shape (pistol grip deformity), neck shaft angle, and risk of hip osteoarthritis: A case–control study. Arthritis & Rheumatism. 2008;58 (10 ):3172–82. doi: 10.1002/art.23939 18821698 19 Moreland JR , Bassett LW , Hanker GJ . Radiographic analysis of the axial alignment of the lower extremity. J Bone Joint Surg Am. 1987 Jun 1;69 (5 ):745–9. 3597474 20 Luo CF . Reference axes for reconstruction of the knee. The Knee. 2004 Aug 1;11 (4 ):251–7. doi: 10.1016/j.knee.2004.03.003 15261208 21 Yoshioka Y , Cooke TD . Femoral anteversion: assessment based on function axes. J Orthop Res. 1987;5 (1 ):86–91. doi: 10.1002/jor.1100050111 3819914 22 Yoshioka Y , Siu DW , Scudamore RA , Cooke TDV . Tibial anatomy and functional axes. Journal of Orthopaedic Research. 1989;7 (1 ):132–7. doi: 10.1002/jor.1100070118 2908904 23 Slocum B , Devine T . Cranial tibial wedge osteotomy: a technique for eliminating cranial tibial thrust in cranial cruciate ligament repair. J Am Vet Med Assoc. 1984 Mar 1;184 (5 ):564–9. 6706801 24 D’Lima DD , Chen PC , Colwell CW . Polyethylene contact stresses, articular congruity, and knee alignment. Clin Orthop Relat Res. 2001 Nov;(392 ):232–8. doi: 10.1097/00003086-200111000-00029 11716388 25 Fang DM , Ritter MA , Davis KE . Coronal alignment in total knee arthroplasty: just how important is it? J Arthroplasty. 2009 Sep;24 (6 Suppl ):39–43. doi: 10.1016/j.arth.2009.04.034 19553073 26 Cherian JJ , Kapadia BH , Banerjee S , Jauregui JJ , Issa K , Mont MA . Mechanical, Anatomical, and Kinematic Axis in TKA: Concepts and Practical Applications. Curr Rev Musculoskelet Med. 2014 Jun 1;7 (2 ):89–95. doi: 10.1007/s12178-014-9218-y 24671469 27 Wang Y , Zeng Y , Dai K , Zhu Z , Xie L . Normal Lower-Extremity Alignment Parameters in Healthy Southern Chinese Adults as a Guide in Total Knee Arthroplasty. The Journal of Arthroplasty. 2010 Jun 1;25 (4 ):563–70. doi: 10.1016/j.arth.2009.03.021 19447000 28 Scorcelletti M , Reeves ND , Rittweger J , Ireland A . Femoral anteversion: significance and measurement. J Anat. 2020 Nov;237 (5 ):811–26. doi: 10.1111/joa.13249 32579722 29 Okazaki K , Miura H , Matsuda S , Takeuchi N , Mawatari T , Hashizume M , et al . Asymmetry of mediolateral laxity of the normal knee. Journal of Orthopaedic Science. 2006 May 1;11 (3 ):264–6. doi: 10.1007/s00776-006-1009-x 16721527 30 Matsuda S , Miura H , Nagamine R , Urabe K , Ikenoue T , Okazaki K , et al . Posterior tibial slope in the normal and varus knee. Am J Knee Surg. 1999 Jan 1;12 (3 ):165–8. 10496466 31 Utzschneider S , Goettinger M , Weber P , Horng A , Glaser C , Jansson V , et al . Development and validation of a new method for the radiologic measurement of the tibial slope. Knee Surg Sports Traumatol Arthrosc. 2011 Oct 1;19 (10 ):1643–8. doi: 10.1007/s00167-011-1414-3 21298254 32 Weinberg DS , Williamson DFK , Gebhart JJ , Knapik DM , Voos JE . Differences in Medial and Lateral Posterior Tibial Slope: An Osteological Review of 1090 Tibiae Comparing Age, Sex, and Race. Am J Sports Med. 2017 Jan 1;45 (1 ):106–13. doi: 10.1177/0363546516662449 27587744 33 Haddad B , Konan S , Mannan K , Scott G . Evaluation of the posterior tibial slope on MR images in different population groups using the tibial proximal anatomical axis. Acta Orthopædica Belgica. 2012 Dec 1;78 (6 ):757. 23409572 34 Todd MS , Lalliss S , Garcia E , DeBerardino TM , Cameron KL . The Relationship between Posterior Tibial Slope and Anterior Cruciate Ligament Injuries. Am J Sports Med. 2010 Jan 1;38 (1 ):63–7. doi: 10.1177/0363546509343198 19737987 35 Bredbenner TL , Eliason TD , Potter RS , Mason RL , Havill LM , Nicolella DP . Statistical shape modeling describes variation in tibia and femur surface geometry between Control and Incidence groups from the Osteoarthritis Initiative database. Journal of Biomechanics. 2010 Jun 18;43 (9 ):1780–6. doi: 10.1016/j.jbiomech.2010.02.015 20227696 36 Guastella DB , Fox DB , Cook JL . Tibial plateau angle in four common canine breeds with cranial cruciate ligament rupture, and its relationship to meniscal tears. Vet Comp Orthop Traumatol. 2008;21 (2 ):125–8. doi: 10.3415/vcot-07-02-0015 18545714 37 Dismukes DI , Tomlinson JL , Fox DB , Cook JL , Witsberger TH . Radiographic Measurement of Canine Tibial Angles in the Sagittal Plane. Veterinary Surgery. 2008;37 (3 ):300–5. doi: 10.1111/j.1532-950X.2008.00381.x 18394079 38 Slocum B , Devine T . Cranial tibial thrust: a primary force in the canine stifle. Journal of the American Veterinary Medical Association. 1983 Aug 1;183 (4 ):456–9. 6618973 39 Schumann S , Tannast M , Nolte LP , Zheng G . Validation of statistical shape model based reconstruction of the proximal femur—A morphology study. Medical Engineering & Physics. 2010 Jul 1;32 (6 ):638–44. doi: 10.1016/j.medengphy.2010.03.010 20435501 40 Sarkalkan N , Weinans H , Zadpoor AA . Statistical shape and appearance models of bones. Bone. 2014 Mar 1;60 :129–40. doi: 10.1016/j.bone.2013.12.006 24334169 41 Kuhn JL , Goulet RW , Pappas M , Goldstein SA . Morphometric and anisotropic symmetries of the canine distal femur. Journal of Orthopaedic Research. 1990;8 (5 ):776–80. doi: 10.1002/jor.1100080520 2388117