**5. Confounding factors**

A highly significant diurnal rhythm was observed in an on-line, computerised study of sheep (Brace and Moore, 1991). In that study, the urine flow rate was measured continuously over a period of days. In 8/9 animals, the peak flow rate occurred at around 9 pm, while it occurred at 9.30 am in the remaining sheep fetus. The maximum urine flow was 28 ± 5% above the 24-hour mean. Although this significant diurnal variation was demonstrated in an animal model, it is important not to disregard a possible diurnal variation in human fetal urine production as well.

The urine production rate was estimated two hours before and two hours after maternal breakfast in 25 AGA and 15 IUGR fetuses. After breakfast, the HFUPR increased in AGA fetuses but did not change in IUGR fetuses (Yasuhi et al., 1996). The PI of the fetal renal artery was significantly reduced after maternal food ingestion in uncomplicated pregnancies (Yasuhi et al., 1997). To our knowledge, no conflicting reports have been published and, to avoid the confounding variation due to diurnal variation and maternal meal ingestion, it is recommended that the estimation of fetal urine production should be performed under standardised conditions. Also, maternal water ingestion might influence the fetal urine production rate (Flack et al., 1995; Oosterhof et al., 2000).

The HFUPR was determined by 2D ultrasound immediately before cordocentesis for blood gas analysis in 27 Small for Gestational Age (SGA) and 101 AGA fetuses (Nicolaides et al., 1990). The HFUPR was reduced in the group of SGA fetuses in comparison with AGA fetuses. Furthermore, the reduction in urine production for the SGA fetuses was correlated with the degree of fetal hypoxemia, while the degree of fetal hypoxemia did not correlate

Several studies demonstrate associations between increased impedance in the fetal renal arteries and factors suggestive of compromised fetal conditions and, in some studies, also reduced urine production rates (Mikovic et al., 2003; Miura, 1991; Stigter et al., 2001; Vyas et al., 1989). In one study, the renal artery flow-velocity wave forms were examined in normal and hypoxemic human fetuses (Vyas et al., 1989). The Pulsatility Index (PI), which is peak systolic velocity minus end diastolic velocity over mean velocity, was higher in SGA than in AGA fetuses. Furthermore, using cordocentesis in the SGA fetuses, a significant, direct correlation was found between blood oxygen deficit and increased renal artery PI (Vyas et al., 1989). Moreover, in a study of 35 IUGR fetuses, the PI in the fetal renal arteries was significantly increased (Mikovic et al., 2003). In studies of fetal urine production, it was demonstrated that the PI in the renal artery was higher in IUGR than in AGA fetuses and that it displayed a negative correlation with the urine production rate and the amniotic fluid volume (Miura, 1991). In spite of varying results regarding the PI in fetal renal arteries (Silver et al., 2003; Stigter et al., 2001), the data suggest that, in fetal hypoxemia, there is a redistribution of blood flow, with a decrease in renal blood perfusion and a decrease in HFUPR. These findings may be important, as it would be of great clinical interest to detect whether or not a particular fetus with growth restriction is

A highly significant diurnal rhythm was observed in an on-line, computerised study of sheep (Brace and Moore, 1991). In that study, the urine flow rate was measured continuously over a period of days. In 8/9 animals, the peak flow rate occurred at around 9 pm, while it occurred at 9.30 am in the remaining sheep fetus. The maximum urine flow was 28 ± 5% above the 24-hour mean. Although this significant diurnal variation was demonstrated in an animal model, it is important not to disregard a possible diurnal

The urine production rate was estimated two hours before and two hours after maternal breakfast in 25 AGA and 15 IUGR fetuses. After breakfast, the HFUPR increased in AGA fetuses but did not change in IUGR fetuses (Yasuhi et al., 1996). The PI of the fetal renal artery was significantly reduced after maternal food ingestion in uncomplicated pregnancies (Yasuhi et al., 1997). To our knowledge, no conflicting reports have been published and, to avoid the confounding variation due to diurnal variation and maternal meal ingestion, it is recommended that the estimation of fetal urine production should be performed under standardised conditions. Also, maternal water ingestion might influence the fetal urine

**4.1 Renal artery flow velocity and urine production in fetuses with hypoxemia** 

with the degree of fetal smallness.

further compromised.

**5. Confounding factors** 

variation in human fetal urine production as well.

production rate (Flack et al., 1995; Oosterhof et al., 2000).

#### **6. Methods for estimating the volume of the fetal urinary bladder**

Calculated bladder volumes cannot be validated in living human fetuses, as the true fetal urinary bladder volumes are not known. The reliability of the estimated bladder dimensions, on the other hand, has been evaluated using the 2D ultrasound technique (Fagerquist et al., 2001; Fagerquist et al., 2003; Fagerquist et al., 2002). The calculation of the measurement error was based on the variability in repeated estimations of identical bladder volumes. Standard deviation (SD) was used because the variation was normally distributed. Furthermore, there was a linear relationship between bladder volume and the measurement error, which has been thoroughly documented in three previous studies (Fagerquist et al., 2001; Fagerquist et al., 2003; Fagerquist et al., 2002). This is a prerequisite for using a linear regression function (Skrepnek, 2005).

Fig. 4. The SD was calculated when estimating the bladder volume of 120 fetuses. Different methods were used and this gave rise to 222 relationships between SD and bladder volume. The maximum and minimum bladder volumes were 80.5 mL and 0.1 mL respectively. The distribution of the SDs supports a linear relationship (correlation coefficient 0.36).

Renal Function and Urine Production in the Compromised Fetus 141

Fig. 6. The image that was going to be used for volume computation according to the sumof-cylinders method was created in an interactive process. Firstly, the operator traced the bladder borders with a red digital marker on the computer screen using the Microsoft-Paint

The software determines the co-ordinates for the bladder boundary pixels. The image was analysed by the software and each column of pixels was scanned 1) from left to right on the screen, moving from top of the image to bottom in each column to identify the red marked top pixel of the bladder border and 2) from right to left on the screen, moving from bottom of the image to top in each column to identify the red marked bottom pixel of the bladder border. The length of each vertical strip (from top to bottom red pixels in each column) perpendicular to the long axis was computed. The length of each strip was regarded as the diameter of a cylinder, one pixel high. The total bladder volume was calculated by adding

computer program.

all the cylinders.

Fig. 5. The distribution of the residuals supports a linear relationship between the SD and bladder volume based on 222 relationships between the SD and bladder volume. The maximum and minimum of the residuals were 8.2 mL and -1.9 mL respectively. The mean was 0.00 mL and the median -0.37 mL.

One important finding was that the fetal urinary bladder can be regarded as a rotational body. This means that an appropriate 2D longitudinal image of the fetal bladder has all the information that is needed for volume calculation. This discovery simplifies the way the volume can be calculated.

The method for estimating the volume of the fetal urinary bladder was gradually developed in order to reduce the SD for volume estimation. In the first paper, the volume measurement errors (SD) when using the conventional method and the ellipsoid formula were analysed (Fagerquist et al., 2001). The SD for the measurement error was 17.3-10.9% for bladder volumes of 5-40 mL.

By introducing the sum-of-cylinders method, the SD was significantly reduced to 8.8-3.5% (p=0.0032) (Fagerquist et al., 2003).

#### **6.1 The 2D technique and the sum-of-cylinders method**

Before volume computation, the operator performed an interactive process using the Microsoft-Paint and Microsoft Excel computer programs. The bladder image documented in the Microsoft-Paint format and the data of calibration in an Excel sheet were activated in the MathCad computer program.

Fig. 5. The distribution of the residuals supports a linear relationship between the SD and bladder volume based on 222 relationships between the SD and bladder volume. The maximum and minimum of the residuals were 8.2 mL and -1.9 mL respectively. The mean

One important finding was that the fetal urinary bladder can be regarded as a rotational body. This means that an appropriate 2D longitudinal image of the fetal bladder has all the information that is needed for volume calculation. This discovery simplifies the way the

The method for estimating the volume of the fetal urinary bladder was gradually developed in order to reduce the SD for volume estimation. In the first paper, the volume measurement errors (SD) when using the conventional method and the ellipsoid formula were analysed (Fagerquist et al., 2001). The SD for the measurement error was 17.3-10.9% for bladder

By introducing the sum-of-cylinders method, the SD was significantly reduced to 8.8-3.5%

Before volume computation, the operator performed an interactive process using the Microsoft-Paint and Microsoft Excel computer programs. The bladder image documented in the Microsoft-Paint format and the data of calibration in an Excel sheet were activated in the

was 0.00 mL and the median -0.37 mL.

volume can be calculated.

volumes of 5-40 mL.

(p=0.0032) (Fagerquist et al., 2003).

MathCad computer program.

**6.1 The 2D technique and the sum-of-cylinders method** 

Fig. 6. The image that was going to be used for volume computation according to the sumof-cylinders method was created in an interactive process. Firstly, the operator traced the bladder borders with a red digital marker on the computer screen using the Microsoft-Paint computer program.

The software determines the co-ordinates for the bladder boundary pixels. The image was analysed by the software and each column of pixels was scanned 1) from left to right on the screen, moving from top of the image to bottom in each column to identify the red marked top pixel of the bladder border and 2) from right to left on the screen, moving from bottom of the image to top in each column to identify the red marked bottom pixel of the bladder border. The length of each vertical strip (from top to bottom red pixels in each column) perpendicular to the long axis was computed. The length of each strip was regarded as the diameter of a cylinder, one pixel high. The total bladder volume was calculated by adding all the cylinders.

Renal Function and Urine Production in the Compromised Fetus 143

On the other hand, when using the 2D ultrasound technique, the operator can avoid disturbing shadows by selecting an appropriate longitudinal bladder image. The bladder is defined by the operator who electronically marks the pixels, which are included in the bladder contour. Typically, this corresponds to 200-300 pixels in the boundary. In this way, the technical limit for

When estimating the HFUPR, the measurement error is made when assessing the volume of the bladder, and that SD has been estimated for the 2D technique and the sum-of-cylinders method. There are some other factors that influence the HFUPR measurement error, the magnitude of the HFUPR and the number and time points of bladder image capture. To date, no information relating to the SD for bladder volume estimation by 3D ultrasound, which is a prerequisite for the subsequent analysis of estimation accuracy, is available.

When utilising the HFUPR for fetal surveillance, it is necessary to know whether the estimated HUFPR is pathologically low, i.e. below the 5th percentile point. It is therefore necessary to answer the question: 1) What is the risk of false readings at the 5th percentile point, for example, even though the true HFUPR is at a higher percentile point? Furthermore, it is necessary to answer the question: 2) How much of an observed HFUPR change (for example, during daily controls) can be explained exclusively by measurement error? The implication of the volume measurement error was demonstrated in detail in a publication publication

concerning the 2D technique and the sum-of-cylinders method. (Fagerquist et al., 2010).

The data suggest that, in fetal hypoxemia, there might be a redistribution of blood flow, with a reduction in both renal perfusion and fetal urine production rate, which can be estimated by ultrasound. These findings may be important, as it would be of great clinical interest to determine whether or not a particular fetus with growth restriction is further compromised. To utilise the HFUPR, for fetal surveillance, a program is available for estimating the risk of false readings at a low percentile point, even though the true HFUPR is at a higher percentile point, and the degree to which an observed HFUPR change can be

I would like to thank Ingemar Kjellmer, Professor at the Department of Pediatrics, Sahlgrenska University Hospital, Gothenburg, for his enthusiasm and encouragement, Anders Odén, Adjunct Professor of Biostatistics, Chalmers University of Technology, Gothenburg, for his patience, and Ulf Fagerquist, Tech Lic, my brother and friend, who created the computer programs, which gave us the opportunity to approach the true volume

Moreover, I would like to express my sincere thanks to Hans Steyskal, Professor and Mathematics Consultant, Concord, MA, USA, for his advice, and Sture G. Blomberg, MD,

optimal precision, one pixel, is reached; this is the smallest unit of display resolution.

**7. The measurement error when estimating the HFUPR** 

**8. Conclusion** 

**9. Acknowledgments** 

of the fetal urinary bladder.

explained exclusively by measurement error.

PhD, for his valuable criticism and support.

Fig. 7. Using the MathCad computer programs, the vertical distances between marked pixels included in the bladder border was estimated. The bladder image was then electronically subdivided in vertical cylinders and the sum of these cylinders with the height of one pixel equals the bladder volume.

#### **6.2 The 3D ultrasound technique**

The 3D ultrasound technique and integrated software, such as the "Virtual Organ Computer-aided AnaLysis" system (VOCALTM), are already available for volume estimation and measurements in the in vitro setting and are both reliable and valid (Raine-Fenning et al., 2003). Moreover, this technique has been applied to the fetal urinary bladder (Lee et al., 2007; Touboul et al., 2008). Unfortunately, the 3D technique is prone to the same types of problem encountered in 2D ultrasound imaging, plus others unique to volume acquisition and visualisation (Nelson et al., 2000). When selecting the initial bladder image, the operator can avoid shadows from the fetal pelvis. However, according to the VOCAL system, the subsequent process for volume estimation is automatic and disturbing shadows are not avoided. Furthermore, in this program, only 40 electronic points are available for the contour marking, which is another disadvantage.

On the other hand, when using the 2D ultrasound technique, the operator can avoid disturbing shadows by selecting an appropriate longitudinal bladder image. The bladder is defined by the operator who electronically marks the pixels, which are included in the bladder contour. Typically, this corresponds to 200-300 pixels in the boundary. In this way, the technical limit for optimal precision, one pixel, is reached; this is the smallest unit of display resolution.
