Computerized histogram analysis of the canine placenta during normal pregnancy

Abstract

In this study, the aim was to evaluate digitalized images obtained via B-mode sonography (USG) during different stages of pregnancy and to assess features of the canine placenta, typical for the different gestational periods. For this purpose, ultrasound pictures were digitalized and a computerized histogram analysis performed. Finally, 151 USG images were obtained from 37 healthy, pregnant dogs of different breeds, aged 4.4 ± 1.8 years, between days 17 and 62 of gestation, and 755 Regions of Interest (ROI) within these images were evaluated. Five ROI per picture were situated in the ventral and dorsal uterus/placenta (early stages) or exclusively the placenta (later stages). The data were grouped into GI: gestational day 17–25, GII: gestational day 26–40, GIII: gestational day 41–50 and GIV: gestational day 51–62.The mean grey value (MGV), standard deviation of grayscale value (StdDGV), minimum grey value (MinGV), maximum grey value (MaxGV), Modal/Mode grey value (ModGV), median grey value (MedGV), homogeneity (Angular Second Moment, ASM), contrast (CONT), correlation (COR), regional homogeneity (Inverse Difference Moment, IDM) and the entropy (ENT) parameters in the Region of Interest (ROI) were calculated.In the course of pregnancy, the grey value parameters, CONT and the local homogeneity (IDM) changed characteristically. The MGV increased between days 17 and 40 (p < 0.0001) and decreased towards late gestation (GIV, p < 0.0001). In parallel, the ModGV, MedGV and MaxGV values changed significantly (p < 0.0001). Similarly, CONT value significantly increased between implantation (GI) and later gestational periods, together with the increasing grayscale variation, probably indicating the massive increase in vascularisation during mid-gestation. This is supported by the fact that local homogeneity (IDM) was highest during implantation and decreased towards late pregnancy (p < 0.01). There was no difference between groups in terms of ASM, COR and ENT (p > 0.05). We conclude that some of the parameters evaluated by computational analysis increased or decreased with the progression of pregnancy, and correlated with the gestational period of bitches. Further studies will reveal whether this technique can contribute to early diagnosis of placental alterations.