BRIEF DESCRIPTION OF THE DRAWINGS
An embodiment of the invention will now be described, by way of example only, with reference to the accompanying figures, in which:
FIG. 1 shows a 35 minute STAN® recording in 1st stage of labour in which the fetus is exposed to slowly developing hypoxia;
FIG. 2 shows a STAN® recording during the last stages of delivery showing the FHR overacting with an “Alarm reaction” to developing hypoxia;
FIG. 3 shows a plot of FHR variation with the identified primary fetal heart rate component superimposed;
FIG. 4A shows a plot of the FHR with the identified primary FHR component superimposed;
FIG. 4B shows the resulting FHR residual component of FIG. 4A;
FIGS. 5A and B respectively show the FHR and a plot of the 95th percentile of the FHR residual component in a situation where the fetus is displaying a lack of reactivity;
FIG. 6 shows a plot of the frequency distribution of 3-4 ms FHR residuals in a fetus showing a lack of reactivity;
FIGS. 7A and B respectively show the FHR and a plot of the 3-4 ms frequency distribution in an instance where the fetus is gradually losing reactivity;
FIGS. 8A and B respectively show the FHR and a plot of the 3-4 ms frequency distribution in an instance where the FHR data alone is ambiguous;
FIG. 9 shows a plot of the 95th percentile of the FHR residual component in respect of the fetus recorded in FIG. 2.
FIG. 10 provides a flow chart indicating the different steps in the residual analysis process according to one embodiment of the invention.
FIG. 11 is a schematic illustration of the apparatus according to one embodiment of the present invention.
DETAILED DESCRIPTION
During labour the FHR is monitored for signs of fetal distress (lack of oxygen). FIG. 1 shows a 35 minute STAN® recording of a 1st stage of labour. The case illustrates a situation of lack of FHR variability and reactivity (preterminal FHR pattern) that at time of the recording was not recognised in spite of the ST event. The figure also depicts low measurements of “residual” markers of FHR reactivity and variability (Frequency distribution within the 3-4 ms range of residual measurements (FD3-4 ms) being <1% and 95th percentile of residual measurements being <3 ms).
The FHR 10 is displayed together with uterine activity 12. Here, although the FHR 10 does rise in conjunction with an increase in uterine activity 12 (i.e. contractions), the FHR variation is less than usually expected at this stage. While most experienced obstetricians would be aware that the FHR 10 was abnormal, such a reading would not necessarily trigger other alarms (such as an ST event). Therefore there are occasions when such an abnormal reading would be missed.
FIG. 2 shows another example for an abnormal FHR that can sometimes be overlooked upon a visual observation only. This STAN® recording was taken during the final stages delivery and again shows the FHR 20 and uterine activity 22. During this stage of labour it is usual for the FHR 20 to be very erratic and therefore many obstetricians would interpret this reading as normal. Even if other alarms, such a ST events, were raised, confusion over the interpretation of the FHR 20 can lead to delays.
In this case, we have a progressive increase in T/QRS ratio identified by the ST log. The FHR 20 was over-reacting (Alarm reaction) to developing hypoxia. Due to uncertainty over the analysis of the FHR 20, the fetus was born 44 minutes later with signs of oxygen deficiency. Note the marked increase in beat-to-beat variation.
Extreme situations of reduced or increased variability in the heart rate cause problems for interpretation as the above examples show. Using the method described above, the FHR can be separated into a primary component and a residual component.
An approximation of the primary FHR component can be calculated by means of polynomial curves fitting. FIG. 3 shows a plot of duration (in ms) of consecutive heart beats (RR intervals) over time (in seconds). The polynomial curve fit 34 is shown over the actual RR data 30. Beat-to-beat residual variation corresponds to the difference between the RR data 30 and the curve fit 34 and comprises the FHR residual component. Beat-to-beat variation would correspond to the difference between the RR data and the curve fit.
In order to prevent “ringing” in the primary FHR component, the FHR data is divided into small adjacent regions and individual approximations are performed in each region. FIGS. 4A and 4B show a piecewise polymomial approximation to the HR data. FIG. 4A shows a plot of FHR (beats per minute) over time (s). The FHR 40 has been split into 5 adjacent regions 421, 422, 423, 424, 425 of 20 consecutive heart rate samples. A polynomial approximation is carried out in each region 421, 422, 423, 424, 425 to provide individual primary FHR components 441, 442, 443, 444, 445. As the approximations are independent they do not induce “ringing” in the primary FHR component. However, in order to give a smooth continuous curve, it is important to ensure that the primary FHR components 441, 442, 443, 444, 445 align at the boundaries of each region 421, 422, 423, 424, 425. To achieve this, each neighbouring pair of regions shares one heart rate sample. These samples are known as knot points 46. At each knot point 46 the neighbouring polynomial curves are constrained to meet two conditions. Firstly that they have an equal value and secondly that their first derivatives (gradients) are equal. Further constraints can be added so that higher order derivatives must also be equal at the knot points 46, but in this application there is no demonstrated benefit.
Once the primary FHR component has been identified, it is subtracted from the FHR to leave the FHR residual component. FIG. 4B shows the FHR residual component 48 which is derived from FIG. 4A. Once the FHR residual component 48 has been extracted from the RR or FHR data this can be analysed in a number of ways to monitor the health of the fetus.
FIG. 5A shows the FHR 50 together with the uterine activity 54 in a case of preterminal fetal heart rate pattern recorded during 80 minutes. FIG. 5B displays the 95th percentile residual measurements with a 20 minute running median. The 95th percentile shows the level that 95% of the residual component has been below during a certain time period. This reflects changes in the FHR variability and FHR reactivity. From a study of STAN recordings it has been found that a consistent reading of below 3 ms, such as that shown in FIG. 5B, indicates a lack of reactivity.
Another area of interest is the 3-4 ms frequency distribution. The 3-4 ms band has been identified as the highest band which most consistently contains FHR residuals regardless of the fetal heart rate and can thus be used regardless of whether the fetus is in a sleep or active state. It has been found that loss of reactivity can be defined as a situation where the maximum 3-4 ms frequency distribution is of <7%. An illustration of this is given in FIG. 6 applying the fetal heart rate sequence of FIG. 5A.
Some examples, which highlight the benefits of the present invention, are now given.
FIG. 7A shows the FHR 70 together with the uterine activity 74. In this situation a gradual loss of FHR variation is occurring, which from this data alone can often be missed or at least cause a delay before the loss of reactivity is detected. In this example, the 3-4 ms distribution frequency is obtained as described above and this is shown in FIG. 7B. It will be seen that in this plot the loss of reactivity is much more apparent and dramatic and therefore easier to detect. With such data available to the obstetrician, diagnosis and action regarding the health of the fetus can be taken more rapidly.
FIG. 8A shows another plot of FHR 80, together with uterine activity 84. Here again the FHR variability appears relatively low and from a visual observation may appear to indicate a loss of reactivity. However, from a study of the 3-4 ms frequency distribution of the FHR residual component, given in FIG. 8B, it is apparent that the residuals are regularly greater than 7%. Therefore, the fetus is healthy and is showing signs of reactivity. Without being able to study this plot, it is possible the obstetrician would make a wrong diagnosis which would lead to unnecessary intervention with the delivery.
FIG. 9 shows the 95th percentile plot of the FHR residual component of FHR 20 recorded in FIG. 2. Whereas the FHR readings 20 were indecisive, it is clear from the 95th percentile plot that during the period covered by FIG. 2 there is an extreme rise in FHR variability, indicating fetal distress.
The method of the present invention has been applied to stored FHR recordings obtained from thousands of deliveries. Careful study of the FHR residual components in these cases has lead to recommendations for its use according to Table I.
FIG. 10 provides a flow chart indicating the different steps in the residual analysis process, according to one embodiment of the present invention.
FIG. 11 is a schematic illustration of the apparatus 100 according to one embodiment of the present invention, comprising means for determining a fetal heart rate (110), for a fetus (170), means for identifying a primary fetal heart rate component which is required to shift a volume of blood from the heart to the cardiovascular system (120), means for subtracting the primary component from the determined fetal heart rate to determine a residual component (130), means for using said residual component to estimate the fetal heart rate beat-to-beat variation (140), computer means (150), and data storage means (160).
As discussed above, the present invention provides a new and improved way of analysing and monitoring the FHR. This method can be applied during pregnancy and is particularly useful during labour and delivery, when the FHR is non-stationary. The identification of the primary and residual FHR components is a new concept. Careful study of FHR recordings has revealed the most effective methods of applying statistical analysis to the residual FHR component to provide indications of hypoxia.
Specific embodiments of the invention have now been described. However, it should be pointed out that the present invention is not limited to the realizations described above. Several alternatives are possible, as would be apparent for someone skilled in the art. For example, the implementation of the method as discussed above could be accomplished in different ways, such as in especially dedicated hardware or in software, or by a combination of the two. Further, the means for executing various method steps could be arranged as separate units or entities, but alternatively several method steps could be executed by one single unit. Accordingly, a single unit may perform the functions of several means recited in the claims. Such and other obvious variations must be considered to be part of the present invention, as it is defined in the appended claims.
Patent Application
20080045849
