Follicle Stimulating Hormone (FSH) levels increase in the follicular phase and are highest in the pre-ovulatory period, and maximum in the middle of the cycle [1]. At serum FSH concentrations between 15 to 20 mUI/mL serum estradiol rises above baseline. Subsequently, estradiol concentrations increase during the follicular phase. Thus, as the menstrual cycle progresses, estrogen stimulates the central production of FSH, and FSH in turn stimulates the production of estrogen progressively entering the bloodstream [2, 3]. This is the basis for the present contribution since there is a well-documented correlation between the days of the beginning of the cycle, with stimulation of serum estradiol around 291.25 $\pm$ 8.89 pmol/L (mean $\pm$ SEM). As the cycle progresses, the quality of daily estrogen released by the ovaries increase, reaching levels around 701.22 $\pm$ 16.28 pmol/L, (p 0.0001) [4]. FSH levels in the bloodstream are well characterized throughout the menstrual cycle [5, 6]. These changes occur because the dominant follicle responds to a threshold level of FSH, which raises estrogens both systemically as well as locally in the cervix [1]. Gradual changes in the characterization of cervical secretion, are a faithful mirror of the variations of the follicular dynamics and the ovulation process. Estimates of the day of ovulation (EDO) based on basal body temperature (BBT) were reported by Barrett & Marshall in 1969 [7], Schwartz in 1980 [8], and Royston in 1982 [9]. The American Society for Reproductive Medicine (ASRM) defines Fertile Window (FW) as the first day on which oestrone-3-glucuronide (E3G) is detected in urine until the second day after the peak of luteinizing hormone (LH) [10]. Currently, most studies define the period of greatest fertility from –6 days before ovulation to the first day after ovulation, and it is in this interval that the change from infertile days to days of greatest fertility generally occurs. This is determined by the progressive increase in serum oestrone-3-glucuronide (E3G) concentrations, from the day –6 days before EDO [11, 12, 13, 14]. Billings [15] first proposed a now commonly used method to assess the progress of the biophysical properties of cervical secretion: to infer form the analysis of the cervical secretion at the vulva, fertile or infertile conditions according to its type, as a function of its physical touch sensation, visual appearance and consistency. Many similarly observational methods have appeared in the literature, including those by Hilgers, Ferning and others [11, 16, 17, 18, 19, 20]. Cervical discharge is a hydrogel where the liquid component is an aqueous solution of various biochemical compounds (e.g., salts, minerals, sugars, amino acids, lipids, protein chains, enzymes, etc.) and the solid component mainly glycoproteins. These components determine the main biophysical parameters of the secretion of the cervix, which give it variations in quantity, appearance, transparency, viscoelasticity, and crystallization [21, 22, 23]. Odeblad described different types of crystallization patterns of dehydrated cervical secretions obtained by endocervical examination [24]. Type E (oestrogenic) secretion is abundant during the fertile phase and easily recognizable because it is clear, watery, elastic, and easy to quantify within the fertile phase, in quantity, appearance and consistency, characteristics that support an acellular environment suitable to sperm survival and motility [25]. The secretion type G (gestagen) is scarce, thick, opaque, and cellular which blocks sperm penetration. Since stretching and transparency in an estrogenic cervical secretion increases in the preovulatory phase their measurement and characterization can be used as basic indicators of FW [14, 15, 16, 17, 18, 19, 20, 24]. The internal structure of cervical secretion varies depending on the glycoprotein net that retains a liquid medium in its meshes. At the beginning and the end of the luteal cycle, the mesh diameter of the holes does not exceed 0.5 µm. During the follicular phase, it increases gradually under the effects of estrogens until it reaches a diameter between 2 µm and 5 µm in the pre-ovulatory period. Under physiological conditions, one observes that the dehydrated secretion reduces to a structure showing an arborization pattern reminiscent of “fern leaves”. This pattern appears to correlate with the concentration of circulating estradiol and is therefore more pronounced in the pre-ovulatory phase. The characteristic changes of each phase by Odeblad can only be determined and characterized by mixing percentages of the different crystallization patterns observed by microscopy and by using an endocervical approach. The daily secretion, during the pre-ovulatory period, is around 600 to 700 mg/day, and outside of that period it is 40 or 60 mg/day. Most studies attempting to determine the FW using observations of the cervical secretion utilize similar intervals –6 days EDO [14, 15, 16, 17, 18, 19, 20, 26, 27, 28]. Currently, most studies comparing FW with cervical discharge include the biophysical fertile window (BFW) [14, 28].

Crystallization of the cervical discharge of the vulvovaginal secretion is a consequence of the organization of sodium and potassium chloride crystals around a main axis formed by mucin in a process of dehydration of the cervical secretion [29]. Characteristic “fern-leaf” secretion L-type is proportional to the circulating estradiol concentration, which is higher in FW, and maximal in the pre-periovulatory phase [30, 31, 32]. L mucus exhibits “fern-like” topology, with a straight or curved central axis and branches that come out at 90° angle. These branches also act as an axis for new branches, again at 90° angles [29]. G mucus does not exhibit a predetermined form. Peak (P) mucus exhibits a star-like topology with six well-defined axes with hexagonal symmetry and branches protruding from the axes at 60° angles, resulting in 6 well defined subtypes [29]. S mucus exhibits a parallel arrangement of crystals that are close together yet not joined, and there are 3 subtypes S1–S3 [29]. These 4 patterns and are their corresponding mucus types were first reported by Odeblad in 1994 [32], and these results are successful and, like those obtained by Odeblad in 1983 [22] and Menárquez and Pastor in 2003 [29]. The presence of L and P endocervical secretions has been widely documented in the FW [22, 32]. Odeblad has developed the possibility of graphically estimating L-, S- and G-type crystallization patterns in endocervical samples in the menstrual cycle [24], but it was not possible to do with cervical discharge. The assessment of the FW by Creighton Model Fertility Care (CrMS) has a well-established biological basis, as the type of cervical secretion depends on estrogen concentration, which increases about 6 days before EDO [15, 32, 33, 34, 35]. The main objective of this work was to document, using computer analysis of crystallization, the possibility of pattern recognition of L-type samples obtained in vulvovaginal discharge.

A total of 29 captures were selected for analysis, of which 58.6% were positive as true positives. Thirty-four percent were true negatives, and 6.8% represented 2 cases in which they were falsely identified due to unstructured areas and were therefore erroneously classified as positives. No cases were identified as false negatives. This statistical difference was significant with a Chi-Square p test ( 0.001). Four samples corresponding to Figs. 3,4,5,6 classified in the true positive group are described in detail below. A detailed description of the result of a sample with a flat histogram is also included. The present selection of images is intended to show the characterization of the pattern found positive when compared to the directionality study of the main outcome of the measure. Fig. 3a shows the binarization of the scanned image, and Fig. 3b shows the corresponding histogram of frequencies distributed in three populations. A typical three-frequency pattern is observed, far from being isotropic. The mean frequency distribution starts in this case from 55 degrees and has a maximum peak of mean distribution around 70 degrees in the Gaussian curve description. Another group observed was formed by a symmetric population opposite to this group of frequencies with a mean frequency peak at –90 degrees, and a third population in the center representing the central branch, with a frequency peak between (–10 and +10 degrees). In Fig. 3c, the angle color of the skeletonized figure plots results in two representative right-angle populations in fuchsia and green, according to the histogram directionality that determines a right-angle orientation.

Fig. 3.

Binarization, skeletonization and staining of crystallization frequencies. (a) Binarization. The figure corresponds to the binarization of the scan of digital image. (b) Directionality histogram. (b) shows the frequency histogram and the relationship between the distribution of the histogram and the most preferred frequencies, which is observed in three populations. (c) Skeletonization and coloration of crystallization frequencies. The color representation of the angles in the skeletonized figure takes the form of two populations of colors representative of the orientation of the population of angles that occur most frequently in fuchsia and green. (d) Frequency plot. (d) corresponds to the frequency angles on the chromatic disk to indicate the color relationship of the crystallization angles analyzed by the frequency histogram.

Fig. 4a shows a preferred group of angles analyzed in the histogram at –90°, as can be seen in Fig. 4b. The corresponding orange coloring on the color scale can be observed, which corresponds to angles with a mean frequency range of –80°, as reflected in the right-angle skeleton diagram on the central axis branch in Fig. 4c.

Fig. 4.

Digitalization area, frequency diagram and colour diagram of the frequency angles obtained in the skeletonization process. (a) Digitalization area. (a) shows the scanned area. (b) Directionality histogram. (b) shows a histogram shifted at –80 degrees defining the preferred histogram population, characteristically marked at an angle close to 90 degrees to the branch axis. (c) Color skeletonization. The orange shading in the color range corresponds to –80° angles, as reflected in the central axis branch. (d) Frequency plot. (d) corresponds to a color disk to represent the angles of the frequency histogram.

The histogram in Fig. 5b shows similar behavior in the distribution of the preferred frequencies of the three populations. Two groups of preferred frequencies are observed with a maximum peak at 90°, and another symmetrical around an axis at –20° as can been seen in Fig. 5b. Two preferred shades, green and fuchsia, are identified in Fig. 5c, corresponding to 90° angles.

Fig. 5.

Digitalization area, crystallization frequencies and colour skeletonization. (a) Digital image scanning. The figure corresponds to the scanning of the digital image. (b) Directionality histogram. In this case, the population of preferred frequencies is greater than 75 degrees and two more frequency groups are observed; with a maximum peak at –90 degrees, and another asymmetric around the axis at –20 degrees, which tends to a preferred right-angle orientation as seen in the coloration of the (c) Color skeletonization. The chromatic skeletonization of the scanned sample represents two preferred colors, green and fuchsia which are observed at right angles. (d) Frequency plot. (d) corresponds to the color disc representing the angles of the frequency histogram.

The latter sample is characteristically organized around the central axis of the highest frequency and vertical orientation around 0°, as seen in Fig. 6a,b. The frequency symmetry corresponding to the green-fuchsia staining is due to the relationship with the branches perpendicular to the central axis, as shown in Fig. 6c.

Fig. 6.

Digitalization area, frequency diagram and colour diagram of the frequency angles obtained in the skeletonization process. (a) Digitization of the selected crystallization zone. (b) Directionality histogram. (b) is characteristically organized around the central axis of higher frequency and vertical orientation around 0 degrees. It represents a set of branches characteristically demonstrating two symmetrical populations at the extremes of the maximum peak at 80 and –80 degrees, symmetrical with a central group between (–20 and 20 degrees) of the maximum peak at zero degrees. (c) Color skeletonization. A right-angle orientation representative of the histogram is observed in the skeletonization coloration. Both preferred frequencies in red and blue-green colors represent respectively histogram angles of preferably perpendicular orientation (d) Frequency plot: The colors of the preferred angles are observed on the color disc.

The study of Fig. 7a reveals a fully flat histogram, without any preference for directionality, as seen in Fig. 7b. The color representation of the angles obtained from the skeletonization reveals a collage of colors in different orientations according to the colorimetry of Fig. 7c. In this sample, it can be said that it does not correspond to a 90-degree pattern, which could be representative of a type G secretion. Fig. 7d makes it possible to observe practically all the colors observed from various angles, since there are no preferred frequencies, so practically all the colors of the color discogram are represented, which coincides with the absence of a preferred branching pattern. It has not been possible to find samples of other characteristics in the vulvovaginal discharge in this series.

Fig. 7.

Scanned crystallization image, directionality of the obtained frequencies and skeletonization of the image. (a) Digitalized crystallization sample. (a) is another digitalized sample of the vaginal discharge of cervical secretion. In this case, an atypical area was selected which does not correspond to a crystallization organized around an axis. The zone framed by the yellow circle was studied. (b) Directionality histogram. (b), reveals a histogram with no preferential directionality in the direction of the sample, and no frequency peak can be detected. (c) Collage of coloration obtained from the skeletonization study. When observing the coloration of the skeletonization of the capture of the angles in (c), a collage of colors is recorded in different directions according to the colorimetric scale. In this sample, it can be said that characteristically it does not correspond to a 90-degree pattern, and it could be compatible with a type G sample. (d) Color discogram. (d) makes it possible to observe practically all the colors observed from various angles, since there are no preferred frequencies, so practically all the colors of the color discogram are represented, which coincides with the absence of a preferred branching pattern.

The skeletonization of an image of the crystallization of vulvovaginal discharge in this series allows us to identify three mean frequencies in the pattern of behavior of L-type samples. A definite pattern in FW is observed both in the study of the histogram angle frequencies and in the coloration of the branches, which follow an ordered pattern with a tendency to be perpendicular to 90 degrees. The four samples labeled with the numbers 3, 4, 5, and 6 show a similar behavior compared to the endocervical sample. Thus, this study of the directionality of branch connection of a cervical crystallization allows us a methodology to verify the L-type secretion in the vulvovaginal discharge. Sample 7 reveals a histogram with no preferred directionality. Looking at the skeletonization of the capture with the angles being recorded, a collage of colors in different directions is seen, which is characteristic of a lack of internal organization in the directionality of the angles and mixing of colors in the face of the impossibility of detecting a preferred angle. This pattern may be characteristic of a sample without symmetry, which could be characteristic of a G-type sample. In Fig. 7a, the attached digitized image reflects a capture with a type of branch that simulates an L-type secretion. These findings are well known in FW according to Odeblad’s scheme [24, 32]. G-type crystallization according to this scheme is also found in FW. Therefore, a smoothed multi-frequency curve with a flat histogram may be characteristic of a G-type secretion as seen in this figure. This sample may correspond to intermediate shots with part of mucus type L and G. This would explain the mixture of colors and angles in the same shot. It could be predicted that outside the fertile window, the absence of an orderly orientation of the branches is characteristicly higher, which would allow us to see multicolored skeletonization with no established pattern. Outside the FW we know that a type G pattern is more frequently identified according to the Odeblad´s scheme. Further studies could be performed to verify the presence of this previously described multicolor pattern outside the fertile window [24, 32].

We have used the CrMS to record FW days in the collected vulvovaginal samples. However, other classifications exist to study changes in the pattern of cervical secretion throughout the cycle [15, 16, 17, 18]. The billing method is another way to identify the FW, but in this series, it was impossible, not because of the method, but because of the technical difficulty described above in this series.

In this study, all samples were possible to integrate into the FW interval, which allowed us in this short series to integrate the observational branched crystallization patterns of the samples taken in the study. Under physiological conditions, the characteristic “fern-leaf” arborization is proportional to the circulating estradiol concentration and therefore maximal in the preovulatory phase [21, 22, 23, 24, 29, 30, 31, 32, 33, 34]. The distinctive changes of each phase can only be determined and characterized by mixing the percentages of the different crystallization patterns by microscopy, but this is only possible in an endocervical sample according to Odeblad’s scheme [32]. A quantifiable stage-by-stage characterization of the crystallization of cervical secretion was not possible by studying a cervical discharge sample. This series has allowed us, by studying the skeletonization, binarization, and directionality of a crystallized image to identify the mean frequencies that are repeated in the behavioral pattern of L-type samples. A definite pattern was observed in the FW in the study of the angular frequencies of the histogram and in the coloration of the branches, which follow an ordered pattern with a tendency to be perpendicular to 90 degrees. Therefore, we were able to see, a representative orientation of the right-angled histogram characteristic of 3 groups of populations of the studied angles, demonstrating the perpendicular relationship of the same, thus it was possible to differentiate this comparison by applying the comparison model of a perfect branch comparison sample of endocervical origin. The main outcome sample for this study was taken from endocervical sampling, which describes a 90-degree arborization within the cervical canal. It was the way to verify that the samples obtained for vulvovaginal crystallization have their origin in endocervical secretion.

It is well known that the volume of cervical secretion increases near the periovulatory period, with the greatest amount of secretion being observed around day –1 and 0 relative to EDO. Therefore, it is likely that it is possible to find this type of discharge in the middle of the cycle [35]. Finally, with this work, we have tried to present a scan in which the elasticity and transparency of the cervical secretion integrated with crystallization can be evaluated. The patterns of crystallization frequencies found depend on the ultrastructure of the crystallization of the glycoprotein network of the cervical crypts in hormonal response [32]. Thus, in this short series, we have been able to integrate the characteristics of the fluidity of cervical secretion in the preovulatory phase with the observed pattern of crystallization. Standardized characterization of these findings may be the subject of future developments, since the fluidity, elasticity, and transparency of cervical secretion is quantifiable, and are modified with hormonal changes allowing the detection of biophysical changes in cervical secretion.

However, it may be possible to use another reference point to identify the P day, such as ultrasound, the LH urine test, basal temperature if used appropriately, or isolated biophysical markers. The purpose of this brief study was to present the observation of cervical crystallization in FW. It can become an observable parameter in subfertile patients in order to aid in becoming pregnant. The crystallization process is an inexpensive and easily accessible technology for use by the general public.

In this study of the digitization of the crystallization of cervical secretion, we were able to identify a 90-degree L-type pattern in the FW. Considering the branching angles studied in the directionality study of the crystallization of cervical secretions, the presence in the histogram of right-angled frequencies allowed us to recognize the L-type pattern of cervical secretion. The skeletonization of the studied images together with a FW indicator allowed us to identify this crystallization pattern in samples at the local level in vulvovaginal secretion. Therefore, it was possible to recognize L-type secretion as a biomarker in vaginal discharge samples in FW.

The possibility of adding more data to the fertility diagnostic process may be worthwhile to help determine the most fertile days of FW by simply studying vulvovaginal secretions. This information could be useful for subfertile patients looking for a simple biomarker to help recognize the peak fertility time within FW.

The data sets used and/or analyzed during the present study are available upon reasonable request and describing the reason to the corresponding author.

JMML conceived the study and drafted the original manuscript, performed the conceptualization and the main analyses of the manuscript. JMML, CR and MLEE contributed to data acquisition. JS, MLEE, CR, JLA, MÁMC and OMM contributed to the conceptualization. JMML, MÁMC, OMM and JLA contributed to the analysis and interpretation of the data and statistical analyse. JMML and JS proofreaded and edited in preparation of the final draft. JMML, MÁMC and OMM made final adjustment of the graphic design. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

The subject gave their informed consent for inclusion before they participated in the study, which was conducted in strict accordance with the Declaration of Helsinki, and whose protocol was approved by the Ethics Committee of CEImLAR, Center for Biomedical Research of La Rioja (CIBIR) (approval number P.I.339).

We would like to thank all those who helped us during the construction of this manuscript, in special in the process of capturing imagines of digitalization by the microscopy of anatomy pathological laboratory.

This research received no external funding.

The authors declare no conflict of interest. José María Murcia Lora and Juan Luis Alcázar are serving as one of the Guest editors of this journal. We declare that José María Murcia Lora and Juan Luis Alcázar had no involvement in the peer review of this article and has no access to information regarding its peer review. Full responsibility for the editorial process for this article was delegated to Luca Roncati.