The current study presents an accurate, sensitive and simple method for measuring increases and decreases in rate of flow of milk that represent milk ejection during expression of breastmilk with an electric pump. This method utilises a continuous weighing device with customised software (Showmilk, Medela AG) that records milk flow during breast expression. Increases and decreases in milk flow rate recorded by the Showmilk agreed 72% of the time with duct diameter changes measured by ultrasound (Figures 4 and 5). In addition, milk ejection occurred simultaneously in both breasts during double pumping.
The average number of identified milk ejections were similar between the ultrasound (4.2) and Showmilk (4.5) techniques in this study. These values are in agreement with previous research using ultrasound during breast expression for 15 minutes that found an average of 4.3  and 4.4  milk ejections. Furthermore, a study comparing intraductal pressure to observed milk flow during 15 minutes of breast expression  demonstrated 4 to 6 milk ejections. Interestingly, studies of breastfeeding using intraductal pressure measurements  and blood sampling for oxytocin  show similar numbers of milk ejections (4.4 and 3 to 5, respectively) without controlling for the length of breastfeeding. There are inherent difficulties when comparing the different techniques used to detect milk ejection, and as a result no technique has been classed as the gold standard. Each technique has measurement difficulties such as low pressures obtained by intraductal pressure measurement  and the influence of blood sampling rate, which can lead to under or overestimation of the number of milk ejections (oxytocin release). In addition, these invasive methodologies risk inhibiting the milk ejection reflex .
Compared to ultrasound, the Showmilk is cost-effective, requires little training and is portable. The technique is, however, dependent on the milk being removed from the breast, therefore the accuracy depends, in part, on both the effectiveness of the breast pump and the pumping performance of the mother. The results of this study are comparable to those obtained using other techniques to measure milk ejection, and it is concluded that the Showmilk is as accurate as the other techniques. In addition to detecting milk ejections, the Showmilk was also able to determine the time and relative strength (peak flow rate) of the milk ejections. This information could aid in the optimisation of the duration of the pumping session for individual mothers during established lactation.
The Showmilk detected all milk ejections imaged by ultrasound in 4 of the 14 women, and in 8 of the remaining 10 women the number of milk ejections were either under or over estimated by less than or equal to 2 milk ejections (Figure 4). There were 19 occasions (28%) where there was disagreement in the detection of milk ejection with ultrasound imaging and the Showmilk (Figure 4). In 8 of these 19 occasions, an increase in milk duct diameter was observed with ultrasound and no milk flow was detected. Of these, half occurred in the last 5 minutes of expression when the breast was becoming progressively drained. This phenomenon has also been observed in previous studies [17, 23].
On 11 of these 19 occasions, increased milk flow rates were observed with the Showmilk, without corresponding increases in milk duct diameter and the majority of these milk flow rate increases occurred in the last 5 minutes of expression (Figure 4). Aberrant milk flow rate recordings can occur with the release of pooled milk in the breast shield tunnel, resulting in the possibility of overestimating the number of milk ejections. As decreasing amounts of milk are removed from the breast it is possible the frequency of this artefact may increase. For this study, every effort was made to minimise these occurrences. Furthermore, it is worth noting that the ultrasound monitoring of milk ducts was performed in the non-expressed breast, whereas the expressed breast was changing in the degree of breast fullness due to milk removal. It is possible that milk duct diameter changes became less obvious over time in the non-expressed breast due to milk accumulating in the main ducts. Alternatively, the monitored milk duct may have increased only slightly at milk ejection and was not detected due to the limits of resolution of ultrasound. Also, whether all milk ducts increase in diameter simultaneously at milk ejection has not yet been studied. One animal study  found that neighbouring lobules of the mammary gland can respond differently to oxytocin. After suckling, two neighbouring lobules were observed, one with milk-filled alveoli, and one with empty alveoli. If different areas of the breast are responding differently at milk ejection, then not all milk ducts may be experiencing milk flow at the same time. The technique of ultrasound used in this study can only monitor one or two milk ducts accurately due to their erratic course . Further investigation is required to clarify the timing of ductal responses to oxytocin in women, and determine why the sensitivity of both techniques decreased after ten minutes of expression.
Pre-milk ejections were observed with the Showmilk after only 20 seconds of pumping. These milk flow rate increases had significantly lower peak flow rates and were not associated with increases in milk duct diameter, while the subsequent milk ejection was observed by both milk flow rate and duct diameter increases. Small volumes (approximately 2.5 mL) of milk removed prior to milk ejection during pumping have been reported previously . This milk may represent the milk present in the milk ducts that has been drawn from the duct by vacuum of the breast pump. However, we observed milk jets during this pre-milk ejection volume, suggesting an active role of the breast (positive pressure). In one case, a decrease in milk duct diameter was observed at a time where there was a small increase in milk flow rate (Figure 5). Therefore we suspect these pre-milk ejection volumes are being expressed at the end of a milk ejection that occurred prior to the beginning of breast pumping. Unfortunately, due to the lag time between changes in duct diameter and milk flow rate recordings, the decrease in duct diameter may not have been observed for all mothers. There are two reasons why it is critical that the pump be switched from the stimulation to expression mode during these pre-ejection episodes when jets of milk are observed at the nipple. Firstly, the stimulation phase is not effective at removing milk from the breast and therefore the duration of the measured milk ejection (by flow rate) will be artificially reduced. Secondly, these episodes are shorter and have a slower flow rate and therefore can be easily recognised and accounted for when analysing the flow rate data.
The Showmilk system allowed us, for the first time, to measure simultaneous milk ejections in both breasts during double pumping. It is assumed that milk ejection occurs at the same time in both breasts due to the systemic release of oxytocin; however, no study has yet confirmed this in lactating women. Nor has there been an assessment of how both breasts respond to the same stimulation by a breast pump. We found that the left and right breasts responded with simultaneous milk ejections 95.5% of the time, confirming that indeed most of the time milk ejection occurs in both breasts at the same time. Furthermore, with regard to milk ejection, this allows us to conclude that measurement of one breast is indicative of the process in both breasts. It can now also be assumed that monitoring a milk duct with ultrasound in one breast is consistent with duct dilations (milk ejection), in the opposite breast. These conclusions allowed us to conduct the other arm of the study, looking at ultrasound and milk expression simultaneously.
During 15 minutes of double pumping we measured the same number of milk ejections (5.0) in the left and right breasts. We did, however, find the number of milk ejections to be highly variable between mothers (3-8 milk ejections). These findings are consistent with other breast expression [17, 20] and breastfeeding  studies.
Unilateral milk flow rate increases were observed during double pumping, with the majority of discrepancies occurring toward the end of the expression session as the breast was emptied, similar to findings of the ultrasound and Showmilk component. The Showmilk will only record an increase in milk flow rate if there is sufficient milk in the breast to be removed by the pump. Therefore, if there is little milk available, it is possible milk may not flow despite milk ejection. In addition, during double pumping there may be large discrepancies between either the amount of milk in the breasts, or the emptying rates, which would also account for a milk ejection being detected in one breast and not the other. Despite this drawback, the Showmilk was still able to detect milk ejections even when low milk volumes were expressed (Figure 8), indicating that it may be possible to use this technique to detect milk ejections in mothers with low milk production.
The Showmilk records milk flow rate at a frequency of 8 Hz. Due to this high sensitivity, the recorded data requires smoothing to improve analysis. We investigated the effect of breast pump cycle frequency on milk flow rate to determine the appropriate smoothing parameters. The Symphony electric breast pump ranges from 45 to 78 vacuum cycles/minute. One vacuum cycle involves the application of vacuum to the nipple with the milk release valve closed off, followed by the release of the vacuum resulting in the opening of the milk release valve and milk flow from the breast shield, thus pulses of milk flow from the shield in 0.8-1.3 second cycles. We determined that a smoothing interval between 2.5 and 15 seconds provided a consistent measure of the rate of milk flow from the breast, therefore a smoothing interval of 10 seconds was used for this study (Figure 9).
During double pumping all electric breast pumps apply the same vacuum (set by the mother) to both breasts and do not allow independent adjustment of the applied vacuum. Because we pumped each breast independently we were able to adjust the vacuum separately according to the mothers' comfort. We found that the breast expression vacuum chosen by mothers for the left and right breasts was not different. This suggests that for healthy mothers in established lactation there is probably no advantage in regulating the vacuum levels separately for each breast during double expression. While both the vacuum and number of milk ejections did not differ between breasts, the amount of milk removed from each breast was highly variable. The percentage of available milk removed varied by almost 25% on average between the left and right breasts of a mother, despite the same vacuum being applied to the breast. Further research is required to identify factors other than milk ejection and vacuum level that may contribute to the variation in the proportion of available milk that can be removed from the breast by a breast pump.
In conclusion, the Showmilk records milk flow rate, using continuous weighing during breast expression and changes in the flow rate are indicative of milk ejection. The advantages are that it is simple, non-invasive and portable. Using this methodology we confirmed, in women, the hypothesis that milk ejection occurs synchronously in both the left and right breasts. The Showmilk is a simple technique that can be used in both clinical and research applications to undertake further investigation into milk ejection and elucidate factors influencing milk removal from the lactating breast.