Quantitative Echotexture Analysis for Prediction of Ovulation in Mares |
| |
| Authors: | Jacky Peng-Wen Chan VMDr Tsung-Hsiang Huang MS Shih-Te Chuang VMDr Feng-Pang Cheng PhD Hang-Poung Fung VMDr Chao-Ling Chen PhD Chi-Liang Mao VMDr |
| |
| Institution: | National Chung Hsing University, College of Veterinary Medicine, Department of Veterinary Medicine, 250-1, Kuo-Kuang Road, Taichung 402, Taiwan |
| |
| Abstract: | The aim of this study was to predict the ovulation in mares by quantitative analysis of the echotextural changes of preovulatory follicular walls. Four mares of breeding age with 32 preovulatory follicles and 11 anovulatory follicles were observed by ultrasonography. The slope of the regression line of the follicular wall and the echogenicity score of granulosa layer (GL) and anechoic layer (AL) were measured from the images on Days -3 (Day 0 = ovulation), -2, and -1, respectively. GL was scored from 1 (anechoic) to 3 (echoic), and prominence of AL was recorded from 1 (gray and thin) to 3 (black and thick). The results indicated that the regression line of the follicular wall for 81.3% (26/32) of preovulatory follicles had the slope value ≥19.0 on Day -1, in which 4 of the 26 preovulatory follicles were ≥19.0 on Day -2 already. Mean slope value on Day -1 (21.9 ± 1.5) was significantly greater (P < .01) than on Day -2 (15.0 ± 1.4) and Day -3 (14.0 ± 1.1). All of the slope values for the 11 anovulatory follicles were <19.0 on any given day. GL and AL scores of preovulatory follicles were significantly greater (P < .01) than in anovulatory follicles on Days -3, -2 and -1; nevertheless, only 28.1% (9/32) of preovulatory follicles scored 3 for both GL and AL simultaneously on Day -1. All anovulatory follicles scored <2 for both GL and AL on Day -1. It was concluded that the slope of the regression line of the follicular wall is useful in predicting preovulatory follicles within 48 hours of ovulation when the value is ≥19.0. Of these follicles (N = 26), 84.6% (22/26) were predicted to ovulate within 24 hours, and 15.4% (4/26) within 24 to 48 hours.IntroductionInsemination in mares by accurately predicting the time of ovulation may obtain maximum fertility with minimum use of semen, and therefore would definitely be a profitable advantage in the horse farming business. The optimal time for insemination with frozen-thawed semen usually include a shorter interval than if fresh semen or natural breeding is used. To achieve the maximal pregnancy rates with frozen-thawed semen, it is necessary to inseminate mares during a period between 12 hours pre- and 6 hours post-ovulation.1] Therefore, if the timing of ovulation could be predicted, it would be helpful for the veterinarian to inseminate a mare only once per cycle if performed very close to the time of ovulation. 2] In recent years, many indicators have been reported for predicting impending ovulation in mares, including measurement of electrical resistance of the vaginal mucus, 3] the distinguishable endometrial folding pattern of uterus in estrus, 4] changes in size and shape of the preovulatory follicles, 5, 6 and 7] and the echotexture changes in the preovulatory follicular wall. 8] The latter has been more efficient for predicting the imminence of ovulation; nevertheless, their assessment of criterions was scored subjectively. The hypothesis for this study was based on the published report from Gastal et al in 1998 8]; they found that 2 echotexture changes of the preovulatory follicle-increasing echogenicity of the granulosa layer and increasing prominence of an anechoic layer beneath the granulosa, were detected in the follicular wall as ovulation approached in mares. Computer-assisted image analysis is an advanced technology for diagnostic ultrasonography to improve the reproductive management of patients. 9, 10 and 11] The purpose of this study is to quantify the echotextural changes in the preovulatory follicular wall as ovulation approaches using computer-assisted image analysis, so that the quantified echotexture changes could serve as an indicator for prediction of ovulation in mares.Materials and MethodsAnimals and UltrasonographyFour non-lactating and nonpregnant mixed mares between 4 and 14 years of age and weighing between 450 and 550 kg were studied from January to December 2001. The geographic area of the mares in this study was in subtropical Taiwan of the northern hemisphere. All mares were maintained on alfalfa/grass hay and had access to water and mineralized salt. A teaser stallion was introduced to detect the estrus signs of mares about 2 weeks after the end of the last estrus. Follicular changes were monitored with a real-time B-mode linear assay ultrasound scanner, equipped with a 7.5-MHz transrectal probe (Model Scanner 200 Vet, Pie Medical, The Netherlands). Upon detection of a preovulatory follicle, ultrasound examination was performed daily and continued until ovulation. A total of 32 preovulatory follicles and 11 anovulatory follicles were identified from a retrospective determination.Ultrasonographic images were recorded on Hi-8 MP videotape with a Sony DCR-TRV 120 Digital-8 camera. The brightness and contrast controls of the monitor and the time-gain compensation of the scanner were standardized to constant settings throughout the observation period.Image AnalysisStill images were subsequently captured and saved as TIF files by computer using a digital image analysis program (Image-Pro Express V4.0 for Windows, Media Cybernetics, L.P., USA) with a resolution of 640 × 480 pixels and 256 shades of gray. Echotexture of the regions of interest was defined in terms of pixel intensity ranging from 0 (black) to 255 (white). Three ultrasonographic images of each preovulatory follicle at its distinctly discernible cross section were subsequently selected. To avoid the enhancement of through-transmission, sampling regions were located within the 10 or 2 o'clock position for measurement of pixel values (Fig 1). The pixel values were measured with the “Line Profile” tool, which involved sampling pixel values along a line traversing the follicle wall from the peripheral antrum, GL, AL, to the stroma. A graph of the pixel intensities along the line was produced ( Fig 2). The GL was defined as the highest pixel after which there was a sequential fall in gray-scale values. The pixel values along the curve (P0, P1, P2) were obtained as an average of 9 measurements (3 images per follicle and 3 lines per image) and were used to measure the slope of a regression line of the fall segment ( Fig 2). |
| |
| Keywords: | |
| 本文献已被 ScienceDirect 等数据库收录！ |
|
