Computer-assisted image analysis is a technological extension of reproductive ultrasonography and allows the quantitative assessment of the luteal echotexture, which is related to changes in histological features and, consequently, to steroidogenesis. The aim of this study was to determine the efficiency of luteal echotexture evaluation as a tool to assess luteal function in different phases of the estrous cycle in Toggenburg goats. Nulliparous goats (n = 21), 8 months in age, 33.52 ± 1.22 kg of body weight, and body score condition of 3.5 ± 0.07 (1 to 5 scale), which showed estrus within a 48-h period during the natural breeding season (March and April), were used. After estrous detection (Day 0) and mating, ovarian sonographic evaluations were performed daily using a portable ultrasound device (Aloka SSD 500, Aloka Co.) equipped with an adapted linear transrectal 5-MHz probe. The examinations were preceded by blood sample collections, which were stored until radioimmunoassay for progesterone (P4). Images were recorded in VHS tapes, then digitized to TIFF files (resolution of 1500 × 1125 pixels) using a video capture board. A representative elementary area of 5625 pixels (0.31 cm2) was defined for the luteal tissue according to the criterion proposed by Van den Bygaart and Protz 1999. Computer-assisted analyses were performed using custom-developed software (Quantporo®). Each pixel received a numeric value ranging from 0 (black) to 255 (white). Luteal echotexture and plasma P4 data were analyzed by ANOVA, and differences among means were determined by Tukey’s test. Correlations were established by Pearson’s correlation method. Results are shown as mean ± SEM. Corpora lutea size increased progressively (P < 0.001) until Day 9, when it reached the maximum area (1.26 ± 0.32 cm2). No increase in size was detected on the subsequent days (P > 0.05). Plasma P4 levels increased until a maximum value on Day 9 (6.31 ± 0.46 ng mL–1), and no increase was observed further (P > 0.05). In nonpregnant animals (n = 7), luteolysis was characterized by an abrupt decrease in plasma P4 concentration, which dropped to values lower than 1 ng mL–1 24 h after the onset of the process, whereas luteal area decreased gradually. Plasma P4 concentration was correlated to luteal area during luteogenesis and luteolysis (r = 0.63 and r = 0.50, respectively; P < 0.05). Mean pixel value showed a progressive increase during luteogenesis and reached the maximum value on Day 13 (54.33 ± 1.83). During corpus luteum (CL) regression, mean pixel value decreased to lower values 48 h after the onset of natural luteolysis (P < 0.05). Through both luteogenesis and luteolysis, positive correlations were observed between mean pixel values and luteal area (r = 0.34 and r = 0.26, respectively; P < 0.05) and also between mean pixel values and plasma P4 concentration (r = 0.24 and r = 0.37, respectively; P < 0.05). Pixel heterogeneity was not correlated to luteal area nor plasma P4 levels. These results suggest an association between CL echotexture and steroidogenic function; therefore, the quantitative assessment of the pixel brightness has a potential to be used for luteal function evaluation in goats.
FAPEMIG and CAPES.
Image analysis and processing methods in verifying the correctness of performing low-invasive esthetic medical procedures
