Individual cow SCC was significantly related with BFVol only in the univariable model. These parameters can potentially be used to detect mastitis-related parenchymal alterations and evaluate udder health and productivity, both at individual cow level and at herd level. Based on these results, echotexture analysis could potentially be applied for early diagnosis of subclinical mastitis in lactation, aid the assessment of parenchymal alterations due to chronic subclinical mastitis and suggest a novel method to detect parenchymal alterations due to subclinical mastitis in the DP. Furthermore, we investigated the associations between udder parenchymal echotexture, blood flow in the milk vein, somatic cell count and milk yield of Holstein dairy cows. This could be due to the different technique applied in that study, measuring blood flow in cross-section instead of longitudinal section of the vein. Holstein cows’ in our study, with no significant difference regarding production status in either study. Changes in blood flow (A-C) and morphological (D-F) parameters of the milk vein during the study period: starting from 7 days before the dry period ("DP"), throughout the DP, 7 and 3 days prior to calving, at calving ("C") and from the 3rd to the 90th day in the new lactation period ("DIM": days in milk).

In our multivariable model (Table 3), milk yield and production stage were significantly associated with changes in echotexture features (NPV, PSD and Long-Run Emphasis). Figures presenting the progress of the 15 echotexture parameters throughout the study period: (2) Mean NPV; (3) PSD; (4) Skewness; (5) Entropy; (6) Run Percentage; (7) Correlation; (8) Runlength Distribution; (9) Contrast; (10) Percentage Non-Zero Gradients; (11) Excess; (12) Gradient Mean Value; (13) Gradient Variance; (14) Grey Value Distribution; (15) Long-Run Emphasis; (16) Homogeneity; (17) Nutritional content of rations. 1, 2, 3 are restricted cubic splines with four-knot terms, s1(t), s2(t), s3(t) are restricted cubic splines with four-knot terms and b1i, b2i, b3i are the corresponding random effects, NPV is the Mean Value for cow i at measurement day j, LRE is the Long-Run Emphasis for cow i at measurement day j, PSD is the Pixel standard deviation for cow i at measurement day j, LP2, LP3 and LP4 are the second, third and fourth lactation periods and QuarterB, QuarterC, QuarterD are the quarters of udder. 1, 2, 3 are restricted cubic splines with four-knot terms, s1(t), s2(t), s3(t) are restricted cubic splines with four-knot terms and b1i, b2i, b3i are the corresponding random effects, NPV is the Mean Value for cow i at measurement day j, PSD is the Pixel Standard Deviation for cow i at measurement day j, GMV is the Gradient mean value for cow i at measurement day j, white scrub jacket HG is the Homogeneity for cow i at measurement day j, GradVar is the Gradient Variance for cow i at measurement day j, ENT is the Entropy for cow i at measurement day j, LP2, LP3 and LP4 are the second, third and fourth lactation periods and QuarterB, QuarterC, QuarterD are the quarters of udder.

The multivariable model includes 3/15 echotexture parameters: Mean NPV (est. per 1 unit increase (95%CI), 0.09 (0.03, 0.15)), mean PSD (est. per 1 unit increase (95%CI), 0.57 (0.11, 1,03)), and Long-Run Emphasis (est. per 1 unit increase (95%CI), −0.23 (−0.46, −0.002)). Our study attempted to connect the dots between udder echotexture and blood flow in the milk vein with milk production and somatic cell count. BFVol: Mean blood flow volume of the milk vein. Associations between blood flow volume of the milk vein, daily milk yield and somatic cell count using linear mixed models. Ultrasonography was employed to examine dairy cows’ udder parenchyma and blood flow repeatedly from late lactation through to the DP and early lactation. Ultrasonographic examination of the mammary parenchyma and milk veins is a practical, non-invasive method on farm sites. Vmax: Mean maximum velocity of the milk vein. TAMV: Time-averaged mean velocity of the milk vein. A significant change was observed in the values of blood flow variables (BFVol, TAMV and Vmax) of the milk vein between late lactation, dry period and early lactation (Table 1, Figure 3). The highest value of mean BFVol was recorded at calving day, at 7.57 ± 2.18 L/min, and the lowest at mid-DP (21st day in the DP).

1) Video S1: color Doppler ultrasonographic examination of the blood flow in the milk vein of a cow. Blood flow parameters in the milk vein change significantly throughout late lactation, dry period and early lactation, corresponding to lactation needs, while morphological parameters remain constant. BFVol in the milk vein, keeping all other variables constant. To serve these needs, BFVol in the milk vein increases by 0.25 L/min for every 1 kg increase in DMY, keeping all other variables constant. The results of the univariable and multivariable LME models investigating the relationship between BFVol in the milk vein, DMY and iSCC are presented in Table 2. A significant association was observed between BFVol and DMY in the respective multivariable model, where side of the milk vein (right/left) appeared significant, too. Lactation period did not significantly affect BFVol in univariable or multivariable models. In our multivariable model (Table 4), quarter SCS was significantly associated with NPV, PSD, Gradient Mean Value, Homogeneity, Gradient Variance and Entropy, whilst udder quarter and time were significant, too.