The Effect of Rider:Horse Bodyweight Ratio on the Superficial Body Temperature of Horse's Thoracolumbar Region Evaluated by Advanced Thermal Image Processing

Abstract

Appropriate matching of rider-horse sizes is becoming an increasingly important issue of riding horses' care, as the human population becomes heavier. Recently, infrared thermography (IRT) was considered to be effective in differing the effect of 10.6% and 21.3% of the rider:horse bodyweight ratio, but not 10.1% and 15.3%. As IRT images contain many pixels reflecting the complexity of the body's surface, the pixel relations were assessed by image texture analysis using histogram statistics (HS), gray-level run-length matrix (GLRLM), and gray level co-occurrence matrix (GLCM) approaches. The study aimed to determine differences in texture features of thermal images under the impact of 10-12%, >12 ≤15%, >15 <18% rider:horse bodyweight ratios, respectively. Twelve horses were ridden by each of six riders assigned to light (L), moderate (M), and heavy (H) groups. Thermal images were taken pre- and post-standard exercise and underwent conventional and texture analysis. Texture analysis required image decomposition into red, green, and blue components. Among 372 returned features, 95 HS features, 48 GLRLM features, and 96 GLCH features differed dependent on exercise; whereas 29 HS features, 16 GLRLM features, and 30 GLCH features differed dependent on bodyweight ratio. Contrary to conventional thermal features, the texture heterogeneity measures, InvDefMom, SumEntrp, Entropy, DifVarnc, and DifEntrp, expressed consistent measurable differences when the red component was considered.

Keywords: body mass index; color models; texture analysis; thermograph.

Figure 1

The thermal image processing steps included: (i) image acquisition, (ii) segmentation of ROIs, (iii) conversion to color components, and (iv) extraction features for both conventional and advanced digital image processing. WLI—white light; IRT—infrared thermal image; ROI—region of interest; DIP—digital image processing; Min—the minimal temperature; Max—the maximal temperature; Average—the average temperature; HS—histogram statistics and the example of image with its histogram; GLRLM—gray-level run-length matrix with the example of horizontal matrix; GLCM—gray level co-occurrence matrix with the example of matrix for a 1-pixel distance in the horizontal direction.

Figure 2

Features of (A,E,I,M) conventional thermography, (B,F,J,N) histogram statistics, (C,G,K,Q) gray-level run-length matrix and (D,H,L,P) gray level co-occurrence matrix for examined color components (R, red; G, green, B, blue) were found to be significantly different between the pre-exercise and post-exercise imaging for all rider groups (L, light; M, moderate; H, heavy). Data were presented separately for each region of interest (ROI 1-4) annotated here on the sample thermal image. Taver—average temperature; Tmax—maximal temperature; Tmin—minimal temperature; Skewness—skewness coefficient; Perc01, Perc10, Perc50, Perc90, Perc99—percentiles; Domn01, Domn10—dominants; Maxm01, Maxm10—maximum of moments; GLN—gray level non-uniformity; RLN—run-length nonuniformity; LRE—long-run emphasis; SRE—short-run emphasis; Fraction—the fraction of image in runs; MRLN—run-length nonuniformity moment; MGLN—gray level non-uniformity moment; AngScMom—angular second moment/energy; Correlat—correlation; SumOfSqs—sum of squares; InvDefMom—inverse different moment/homogeneity; SumAverg—summation mean; SumVarnc—summation variance; SumEntrp—summation entropy; DifVarnc—differential variance; DifEntrp—differential entropy.

Figure 3

Features of conventional thermography compared between the light (L), moderate (M), and heavy (H) groups in consecutive regions of interest (ROI 1, (A–C); ROI 2, (D–F); ROI 3, (G–I); ROI 4, (J–L)). Differences between groups were indicated with individual p-values when p < 0.05. Different superscripts on each plot were statistically different. (A,D,G,J) Taver—average temperature; (B,E,H,K) Tmax—maximal temperature; (C,F,I,L) Tmin—minimal temperature.

Figure 4

Features of (A,E,I,M) conventional thermography, (B,F,J,N) histogram statistics, (C,G,K,Q) gray-level run-length matrix and (D,H,L,P) gray level co-occurrence matrix for examined color components (R, red; G, green, B, blue) found to be significantly different between the light (L), moderate (M), and heavy (H) groups. Three groups of differences (Groups Dif.) were separated: I—groups L and H differed (marked as a, ab, b on plots); II—groups L/M and H differed (marked as a, a, b on plots); III—groups L, M, H differed (marked as a, b, c on plots). Data were presented separately for each region of interest (ROI 1–4) annotated here on the sample thermal image of the red component. Taver—average temperature; Tmax—maximal temperature; Tmin—minimal temperature; Skewness—skewness coefficient; Perc01, Perc10, Perc50, Perc90, Perc99—percentiles; Domn01, Domn10—dominants; Maxm01, Maxm10—maximum of moments; GLN—gray level non-uniformity; RLN—run-length nonuniformity; LRE—long-run emphasis; SRE—short-run emphasis; Fraction—a fraction of image in runs; MRLN—run-length nonuniformity moment; MGLN—gray level non-uniformity moment; AngScMom—angular second moment/energy; Correlat—correlation; SumOfSqs—sum of squares; InvDefMom—inverse different moment/homogeneity; SumAverg—summation mean; SumVarnc—summation variance; SumEntrp—summation entropy; DifVarnc—differential variance; DifEntrp—differential entropy.

Figure 5

Analysis of texture features for the red component compared between light (L), moderate (M), and heavy (H) groups in the first region of interest (ROI 1). Differences between groups were indicated with individual p-values when p < 0.05. Different superscripts on each plot were statistically different. (A) Mean; (B) Variance; (C) Skewness—skewness coefficient; (D) Kurtosis; (E–G) Perc01, Perc10, Perc50—percentiles; (H,I) Maxm01, Maxm10—maximum of moments; (J) GLN—gray level non-uniformity; (K) SRE—short-run emphasis; (L) Fraction—a fraction of image in runs; (M) MRLN—run-length nonuniformity moment; (N) MGLN—gray level non-uniformity moment; (O) AngScMom—angular second moment/energy; (P) Contrast; (Q) SumOfSqs—sum of squares; (R) InvDefMom—inverse different moment/homogeneity; (S) SumAverg—summation mean; (T) SumVarnc—summation variance; (U) SumEntrp—summation entropy; (V) Entropy; (W) DifVarnc—differential variance; (X) DifEntrp—differential entropy. Data are presented using minimum and maximum values, lower and upper quartiles, and median. The mean value is marked by a cross.

Figure 6

Analysis of texture features for the green component compared between light (L), moderate (M), and heavy (H) groups in the first region of interest (ROI 1). Differences between groups were indicated with individual p-values when p < 0.05. Different superscripts on each plot were statistically different. (A) Mean; (B) Variance; (C) Skewness—skewness coefficient; (D) Kurtosis; (E–G) Perc01, Perc10, Perc50—percentiles; (H,I) Maxm01, Maxm10—maximum of moments; (J) GLN—gray level non-uniformity; (K) SRE—short-run emphasis; (L) Fraction—a fraction of image in runs; (M) MRLN—run-length nonuniformity moment; (N) MGLN—gray level non-uniformity moment; (O) AngScMom—angular second moment/energy; (P) Contrast; (Q) Correlate; (R) SumOfSqs—sum of squares; (S) InvDefMom—inverse different moment/homogeneity; (T) SumAverg—summation mean; (U) SumVarnc—summation variance; (V) SumEntrp—summation entropy; (W) Entropy; (X) DifVarnc—differential variance; (Y) DifEntrp—differential entropy. Data are presented using minimum and maximum values, lower and upper quartiles, and median. The mean value is marked by a cross.

Figure 7

Analysis of texture features for the blue component compared between light (L), moderate (M), and heavy (H) groups in the first regions of interest (ROI 1). Differences between groups were indicated with individual p-values when p < 0.05. Different superscripts on each plot were statistically different. (A) Mean; (B) Skewness—skewness coefficient; (C) Kurtosis; (D–F) Perc50, Perc90, Perc99—percentiles; (G,H) Maxm01, Maxm10—maximum of moments; (I) RLN—run-length nonuniformity; (J) LRE—long-run emphasis; (K) SRE—short-run emphasis; (L) AngScMom—angular second moment/energy; (M) SumAverg—summation mean; (N) SumEntrp—summation entropy; (O) Entropy. Data are presented using minimum and maximum values, lower and upper quartiles, and median. The mean value is marked by a cross.

Figure 8

Analysis of texture features for the red, green, and blue components compared between light (L), moderate (M), and heavy (H) groups in the second region of interest (ROI 2). Differences between groups were indicated with individual p-values when p < 0.05. Different superscripts on each plot were statistically different. For the red component: (A) RLN—run-length nonuniformity. For the green component: (B) Mean; (C) Variance; (D) Skewness—skewness coefficient; (E) Kurtosis; (F,G) Perc10, Perc50—percentiles; (H) Maxm10—maximum of moments; (I) MGLN—gray level non-uniformity moment; (J) Contrast; (K) Correlat—correlation; (L) SumOfSqs—sum of squares; (M) SumAverg—summation mean; (N) SumVarnc—summation variance; (O) DifVarnc—differential variance; (P) DifEntrp—differential entropy. For the blue component: (Q) Mean; (R) Variance; (S,T) Perc50, Perc90—percentiles; (U) Maxm01—maximum of moments; (V) RLN—run-length nonuniformity; (W) AngScMom—angular second moment/energy; (X) Correlat—correlation; (Y) SumOfSqs—sum of squares; (Z) SumAverg—summation mean; (A’) SumVarnc—summation variance; (B’) SumEntrp—summation entropy; (C’) Entropy. Data are presented using minimum and maximum values, lower and upper quartiles, and median. The mean value is marked by a cross.

Figure 9

Analysis of texture features for the red component compared between light (L), moderate (M), and heavy (H) groups in the third region of interest (ROI 3). Differences between groups were indicated with individual p-values when p < 0.05. Different superscripts on each plot were statistically different. (A) Mean; (B) Variance; (C) Skewness—skewness coefficient; (D) Kurtosis; (E–G) Perc01, Perc10, Perc50—percentiles; (H,J) Maxm01, Maxm10—maximum of moments; (I) Domn01—dominant; (K) GLN—gray level non-uniformity; (L) LRE—long-run emphasis; (M) SRE—short-run emphasis; (N) Fraction—a fraction of image in runs; (O) MRLN—run-length nonuniformity moment; (P) MGLN—gray level non-uniformity moment; (Q) AngScMom—angular second moment/energy; (R) Contrast; (S) Correlat—correlation; (T) SumOfSqs—sum of squares; (U) InvDefMom—inverse different moment/homogeneity; (V) SumAverg—summation mean; (W) SumVarnc—summation variance; (X) SumEntrp—summation entropy; (Y) Entropy; (Z) DifVarnc—differential variance; (A’) DifEntrp—differential entropy. Data are presented using minimum and maximum values, lower and upper quartiles, and median. The mean value is marked by a cross.

Figure 10

Analysis of texture features for the green component compared between light (L), moderate (M), and heavy (H) groups in the third region of interest (ROI 3). Differences between groups were indicated with individual p-values when p < 0.05. Different superscripts on each plot were statistically different. (A) Mean; (B) Variance; (C) Skewness—skewness coefficient; (D) Kurtosis; (E–H) Perc01, Perc10, Perc50, Perc90—percentiles; (I,K) Maxm01, Maxm10—maximum of moments; (J) Domn01—dominant; (L) GLN—gray level non-uniformity; (M) SRE—short-run emphasis; (N) Fraction—a fraction of image in runs; (O) MRLN—run-length nonuniformity moment; (P) MGLN—gray level non-uniformity moment; (Q) Contrast; (R) Correlat—correlation; (S) SumOfSqs—sum of squares; (T) InvDefMom—inverse different moment/homogeneity; (U) SumAverg—summation mean; (V) SumVarnc—summation variance; (W) SumEntrp—summation entropy; (X) Entropy; (Y) DifVarnc—differential variance; (Z) DifEntrp—differential entropy. Data are presented using minimum and maximum values, lower and upper quartiles, and median. The mean value is marked by a cross.

Figure 11

Analysis of texture features for the blue component compared between light (L), moderate (M), and heavy (H) groups in the third region of interest (ROI 3). Differences between groups were indicated with individual p-values when p < 0.05. Different superscripts on each plot were statistically different. (A) Variance; (B) Skewness—skewness coefficient; (C) Perc99—percentile; (D,F) Maxm01, Maxm10—maximum of moments; (E) Domn01—dominant; (G) GLN—gray level non-uniformity; (H) SRE—short-run emphasis; (I) Fraction—a fraction of image in runs; (J) MRLN—run-length nonuniformity moment; (K) MGLN—gray level non-uniformity moment; (L) AngScMom—angular second moment/energy; (M) Contrast; (N) SumOfSqs—sum of squares; (O) InvDefMom—inverse different moment/homogeneity; (P) SumVarnc—summation variance; (Q) DifVarnc—differential variance; (R) DifEntrp—differential entropy. Data are presented using minimum and maximum values, lower and upper quartiles, and median. The mean value is marked by a cross.

Figure 12

Analysis of texture features for the red component compared between light (L), moderate (M), and heavy (H) groups in the fourth region of interest (ROI 4). Differences between groups were indicated with individual p-values when p < 0.05. Different superscripts on each plot were statistically different. (A) Mean; (B) Variance; (C) Skewness—skewness coefficient; (D) Kurtosis; (E–H) Perc01, Perc10, Perc50, Perc90—percentiles; (I,K) Maxm01, Maxm10—maximum of moments; (J) Domn01—dominant; (L) GLN—gray level non-uniformity; (M) LRE—long-run emphasis; (N) SRE—short-run emphasis; (O) Fraction—a fraction of image in runs; (P) MRLN—run-length nonuniformity moment; (Q) MGLN—gray level non-uniformity moment; (R) AngScMom—angular second moment/energy; (S) Contrast; (T) Correlat—correlation; (U) SumOfSqs—sum of squares; (V) InvDefMom—inverse different moment/homogeneity; (W) SumAverg—summation mean; (X) SumVarnc—summation variance; (Y) SumEntrp—summation entropy; (Z) Entropy; (A’) DifVarnc—differential variance; (B’) DifEntrp—differential entropy. Data are presented using minimum and maximum values, lower and upper quartiles, and median. The mean value is marked by a cross.

Figure 13

Analysis of texture features for the green component compared between light (L), moderate (M), and heavy (H) groups in the fourth region of interest (ROI 4). Differences between groups were indicated with individual p-values when p < 0.05. Different superscripts on each plot were statistically different. (A) Mean; (B) Variance; (C) Skewness—skewness coefficient; (D) Kurtosis; (E–H) Perc01, Perc10, Perc50, Perc90—percentiles; (I,K) Maxm01, Maxm10—maximum of moments; (J) Domn01—dominant; (L) GLN—gray level non-uniformity; (M) SRE—short-run emphasis; (N) Fraction—a fraction of image in runs; (O) MRLN—run-length nonuniformity moment; (P) MGLN—gray level non-uniformity moment; (Q) Contrast; (R) Correlat—correlation; (S) SumOfSqs—sum of squares; (T) InvDefMom—inverse different moment/homogeneity; (U) SumAverg—summation mean; (V) SumVarnc—summation variance; (W) SumEntrp—summation entropy; (X) Entropy; (Y) DifVarnc—differential variance; (Z) DifEntrp—differential entropy. Data are presented using minimum and maximum values, lower and upper quartiles, and median. The mean value is marked by a cross.

Figure 14

Analysis of texture features for the blue component compared between light (L), moderate (M), and heavy (H) groups in the fourth region of interest (ROI 4). Differences between groups were indicated with individual p-values when p < 0.05. Different superscripts on each plot were statistically different. (A) Variance; (B) Skewness—skewness coefficient; (C–E) Perc50, Perc90, Perc99—percentiles; (F,H) Maxm01, Maxm10—maximum of moments; (G) Domn01—dominant; (I) GLN—gray level non-uniformity; (J) SRE—short-run emphasis; (K) Fraction—a fraction of image in runs; (L) MRLN—run-length nonuniformity moment; (M) MGLN—gray level non-uniformity moment; (N) AngScMom—angular second moment/energy; (O) Contrast; (P) Correlat—correlation; (Q) SumOfSqs—sum of squares; (R) InvDefMom—inverse different moment/homogeneity; (S) SumVarnc—summation variance; (T) DifVarnc—differential variance; (U) DifEntrp—differential entropy. Data are presented using minimum and maximum values, lower and upper quartiles, and median. The mean value is marked by a cross.