Ultrasonographic anatomy of perineal structures during pregnancy and immediately following obstetric injury




To assess perineal anatomy using ultrasound before and immediately after delivery.


Structures in the perineum were studied by real-time two-dimensional transvaginal and endoanal ultrasound imaging using a combined linear and semicircular (up to 200° sector) probe. We examined 45 nulliparous pregnant women and 44 primiparae immediately after delivery (40 with anal sphincter tears and four without sphincter injury). In each case a single longitudinal image was later assessed by two observers in order to evaluate interobserver agreement.


In pregnancy, the perineal membrane, puboperineal muscles, conjoined longitudinal muscle and central point were identified on real-time examination in 91%, 98%, 100% and 100% of cases, respectively. At offline evaluation of the longitudinal images obtained for each of the pregnant women, the percentage of cases in which each structure was identified by both observers ranged from 64% to 100%. In the women who were examined postpartum, all structures were identified by both observers in all four of the women without sphincter injury. In the women with sphincter tears, the perineal membrane, puboperineal muscles, conjoined longitudinal muscle and central point were found by ultrasound to be intact in 10%, 10%, 55% and 18%, respectively. The agreement between two observers regarding identification of intact structures in a single longitudinal image was good for perineal membrane (kappa index, 0.66), fair for puboperineal muscles (kappa index, 0.40), and poor for conjoined longitudinal muscle and central point (kappa index, 0.08 and 0.17, respectively).


Ultrasonography might be helpful in the evaluation of perineal anatomy and extent of perineal tears. However, the relatively poor agreement between the two observers evaluating single linear transvaginal images implies that both transverse and longitudinal projections are necessary to obtain relevant information. Further studies are needed regarding the importance of specific sonographically identified structures and their role in pelvic floor dysfunction after delivery. Copyright © 2008 ISUOG. Published by John Wiley & Sons, Ltd.


Obstetric anal sphincter injury is associated with a significant risk of anal incontinence1. The consequences of tears in other vaginal and perineal structures have not received the same level of attention as that of the anal sphincter complex, even though the prevalence of anal incontinence after vaginal birth has been estimated to be around 25%—much higher than the rate of clinically diagnosed anal sphincter injuries2. Perineal trauma caused by vaginal delivery has serious consequences for women's health and well-being later in life3. Several studies have found that the extent of perineal tears is difficult to estimate immediately after delivery unless the examiner is specially trained or is using intra-anal ultrasonography4–6.

The perineal anatomy is complex, and controversies exist regarding topography and terminology. Understanding of normal anatomy has increased thanks to in-vivo studies using ultrasonography and magnetic resonance imaging (MRI). The fibrous structures of the perineum anchor the perineum to the skeleton at different sites. One part of this fibrous skeleton, the perineal body, was first named in 1880 by MacAlister7, 8. In sonographic descriptions, the perineal body is a structure proximal (cranial) to the external anal sphincter (EAS) where the internal anal sphincter (IAS), the conjoined longitudinal muscle, the rectovaginal septum and the caudal part of the perineal membrane are joined together. In some textbooks the perineal body is called the central tendon. The height of the perineal body can be measured by intra-anal ultrasound examination9, 10. Wilson described the central point of the perineum apart from the perineal body11. The central point was described as the point where the EAS, the superficial transverse perineal muscle, Colles' fascia and the median posterior border of the perineal membrane meet11. This point is located cranial to the perineal skin and caudal to the hymeneal ring where the perineal membrane is anchored12. The structures of perineum can be visualized using linear or circular transvaginal ultrasonography13–15. The subdivision of the levator ani muscle in this text will follow the description by DeLancey, in which the pubococcygeal part of the muscle is replaced by the pubovisceral muscle consisting of the pubovaginal muscle, the puboanal muscle and the puboperineal muscle16. The existence of the deep transverse perineal muscle has been questioned17. Instead the puboperineal muscle, arising from the pubic bone and inserting in the perineum, has been described using MRI18. The conjoined longitudinal muscle, also (excluding the smooth part) called the puboanal muscle, receives fibers from the pubovisceral muscle and the longitudinal layer of the rectum19. It encircles the IAS, continues between the EAS and the IAS, and inserts in the subcutaneous part of the EAS.

Obstetric textbooks do not always distinguish between perineal and vaginal tears. In this text we define perineal tears as those affecting structures caudal to the caudal part of the rectovaginal fascia. Tears cranial to this level, fornix tears or tears in the rectovaginal fascia are called vaginal tears. As far as we know vaginal tears will not affect anal continence if not causing a fistula.

The primary aim of this study was to evaluate whether the perineal membrane, puboperineal muscle, conjoined longitudinal muscle and central point of the perineum are identifiable on ultrasound investigation during pregnancy and immediately after delivery. The secondary aim was to evaluate the ultrasonographic appearance of these structures in a mid-sagittal vaginal view immediately after delivery in women with anal sphincter tears.


This descriptive study included three groups of women. Group I comprised 45 volunteer nulliparous women, examined during late pregnancy (at 32–40 gestational weeks). Group II included four healthy volunteers who were examined after delivery of their first child, with no tear or with first-degree perineal tears. This group was studied to make sure that the perineal structures could be seen directly postpartum. Group III consisted of 40 consecutive primiparous women with a clinical diagnosis of fourth-degree perineal rupture or third-degree rupture involving more than 50% of the anal sphincter. After inclusion in the study and before repair, the anal sphincter tears were classified by clinical examination and circular endoanal ultrasonography as partial third-degree tears, total third-degree tears (all of the EAS torn) or fourth-degree tears (involving anal mucosa). The ultrasonographic assessment was made in the delivery room before repair. A clinical description of the perineal tears and birth-related variables is given in Table 1. The study was approved by the Research Ethics Committee of Lund University. All participating women gave their informed consent.

Table 1. Clinical data for Group II (four women without sphincter tears) and Group III (40 women with third- or fourth-degree perineal tears)
ParameterGroup II (no tears) (n = 4)Group III
All (n = 40)Partial third-degree tears (n = 12)Total third-degree tears (n = 20)Fourth-degree tears (n = 8)
  • *

    Data missing for one patient.

Demographic data (median (range))
 Maternal age (years)29.5 (29–30)29 (19–37)30 (26–34)28.5 (19–37)28 (26–31)
 Birth weight (kg)3.5 (2.8–4.3)3.8 (2.6–4.7)3.6 (3.2–4.4)3.8 (3.0–4.7)3.7 (2.6–4.4)
 Body mass index (kg/m2)26 (23–30)27 (21–41)27 (23–35)27 (21–41)27 (25–30)
Mode of delivery (n)
 Fundal pressure02002
Mediolateral episiotomy013661
Clinical extension of the tear (n) (n = 39)* (n = 19)* 
 Intact perineal skin 12930
 No vaginal tear 6222
 Midline vaginal tear 10163
 Vaginal fornix tear 239113
De novo anal incontinence at 
 3-month follow-up (n) 
 Urge incontinence 6114
 Gas incontinence 12255
 Fecal incontinence 2002

The examinations were performed using real-time two-dimensional ultrasound imaging with a biplane endoscope probe (with a semicircular 200° window up to 9 MHz and a linear 10-MHz 64-mm window) (Hitachi EUB 6500, transducer U553, Hitachi, Tokyo, Japan). The women were in the dorsal lithotomy position during the examination. During transvaginal scanning, the probe was inserted into the vagina and the linear window was used to obtain one or more sagittal scans of the anal canal, including the perineum (Figure 1). Two scans were obtained after rotating the probe about 25° to the left and to the right. After a shift to the circular window, repeated scans of the anal canal from the level of the puborectal muscle down were made, as the probe was withdrawn from the vagina. Then endoanal linear scan images were obtained from 9 o'clock to 3 o'clock (at ‘hourly’ increments) and at 6 o'clock (totaling eight images). The circular scan was performed with the same technique as used for transvaginal scanning, with the center of the window located at 12 o'clock. The scans were stored on magneto-optical discs for offline analysis. All ultrasound examinations were performed by the same operator (A.K.Ö.).

Figure 1.

Schematic diagram of the linear vaginal probe placed in the vagina. The perineal structures can be observed in the mid-sagittal image. The beginning of the rectovaginal septum (RS) marks the cranial end of the perineum. Injuries in the structures cranial to the perineal body (1) are defined as vaginal tears in this study. In the center of the perineum the perineal body (1) dominates. In the perineal body all structures are hypoechogenic in this projection. The perineal membrane (2) anchors in the perineal body and follows the anterior contour of the puboperineal muscle (3). Cranial to the perineal skin and Colles' fascia the central point (4) can be found. The external anal sphincter (EAS; 5) has a close relationship with the conjoined longitudinal muscle (6), but this structure can also be observed outside of the internal anal sphincter (IAS) cranial to the EAS. The IAS (7) is hypoechogenic on ultrasound examination. The dorsal part of the EAS and the puborectal muscle (PR) are often not well visualized in this projection. (Lena Lyons, reproduced with permission from Ann-Kristin Örnö, Ultrasound studies on pelvic floor physiology and obstetric perineal tears, Faculty of medicine, Lund University, 2007.).

The ability to identify the perineal structures was assessed first by one investigator (A.K.Ö), based on all ultrasonographic images and all projection planes. Images from the transvaginal sagittal scans were evaluated first, and the observed abnormalities compared with those seen in other projection planes. In order to study the interobserver reproducibility of the linear ultrasound assessment of perineal structures, the linear vaginal sagittal scans were then evaluated by two observers (A.K.Ö. and A.H.), the latter being blinded to the clinical diagnosis. The anatomical structures were described as being either identifiable (with a reasonably normal appearance) or not visualized (or appearing clearly abnormal). The agreement was statistically evaluated by calculating the kappa index, and classified as poor (0–0.20), fair (0.21–0.40), moderate (0.41–0.60), good (0.61–0.80) or very good (0.81–1.00). The women with perineal tears were followed up for median 13 weeks after delivery with a structured interview using the Wexner score20, with an additional question on the possible presence of fecal urge (5-min time limit), based on a validated questionnaire mailed to the women around 1 month before the interview.


In scans taken during pregnancy (Group I), the following structures were identified: IAS (hypoechogenic), rectal columns (hyperechogenic), conjoined longitudinal muscle (hyperechogenic), EAS (mixed echogenicity), central point (mixed echogenicity), puboperineal muscle (hyperechogenic) and perineal membrane (hypoechogenic). The perineal body was seen as a hypoechogenic conglomerate of structures between the rectovaginal septum and the EAS in all mid-sagittal scans from pregnant women (Figures 2, 4 and 5). The perineal body was not described in the tables because we were not able to recognize it or the surrounding structures after delivery when they were torn.

Figure 2.

Transvaginal sagittal midline ultrasound image of the anal canal and the perineum during pregnancy. The perineal body (1) is hypoechogenic. The perineal membrane (2) is anchored in the perineal body and stretches towards the central point (4), and is relatively hypoechogenic. The hyperechogenic puboperineal muscles (3) are positioned anterior to the external anal sphincter (5). The hypoechogenic central point (4) is located dorsal to the vaginal introitus, gathering fibers from the perineal membrane, the superficial vaginal muscles, Colles' fascia and the external anal sphincter. Between the external (5) and the internal (7) anal sphincters is the thin hyperechogenic conjoined longitudinal muscle (6), which is interrupted by the perineal body immediately cranial to the external sphincter. The rectovaginal septum (8) is seen cranial to the perineal membrane. Inside the anal canal, the rectal columns (9) fill the lumen of the anal canal.

The first observer identified the EAS, IAS, conjoined longitudinal muscle and central point in all 45 pregnant women when using both linear and circular transvaginal and endoanal ultrasonography. The puboperineal muscle and perineal membrane were identified in 44 and 41 women, respectively (Table 2). These structures, with few exceptions, were also identifiable in a single longitudinal image. The findings were essentially reproduced by the second observer, except those for the conjoined longitudinal muscle and central point, which were identified only in 29 and 34 of the 45 women, respectively (Table 2).

Table 2. Ability to identify intact perineal structures in 45 pregnant nulliparous women (Group I) with multiplanar transvaginal ultrasonography by Observer 1, and in an offline analysis of a single sagittal midline image by Observers 1 and 2
StructureIdentified on multiplanar ultrasonography(Observer 1)Identified in a single sagittal image
Observer 1Observer 2
  • Values are n (%).

  • *

    Kappa index 0.58 and 0.48 for agreement in the identification of perineal membrane between the multiplanar sonography and assessment of a single image by Observer 1 and by Observer 2, respectively. Kappa index 0.70 for agreement between the evaluation of a single image by the two observers. For the other structures, calculation of kappa index was not possible or meaningful, as the number of negative observations on multiplanar sonography was zero or one.

External anal sphincter45 (100)45 (100)45 (100)
Internal anal sphincter45 (100)45 (100)45 (100)
Perineal membrane*41 (91)36 (80)38 (84)
Puboperineal muscle44 (98)44 (98)44 (98)
Longitudinal muscle45 (100)45 (100)29 (64)
Central point45 (100)43 (96)34 (76)

In Group II (women with no or first-degree tears), all structures were identified by both observers on the longitudinal transvaginal scans after delivery (Figure 3).

Figure 3.

Transvaginal sagittal midline ultrasound image of the perineum, rectovaginal septum and anal canal immediately after delivery in a woman with a first-degree perineal tear with all perineal structures preserved. Marked edema is seen in the perineal membrane (2), rectovaginal septum (8) and perineal body (1). The fibers in the external anal sphincter (5) are intact, but it has a different appearance from that seen before delivery. 3, puboperineal muscle; 4, central point; 6, conjoined longitudinal muscle; 7, internal anal sphincter; 9, rectal columns.

Among the 40 women in Group III with third- and fourth-degree perineal tears, 21 had been delivered with ventouse or forceps, and in 13 a mediolateral episiotomy had been performed. Eight women had fourth-degree tears; in these tears the perineum, the perineal body and parts of the rectovaginal septum were completely divided. Twelve of the 32 third-degree tears were partial and 20 were total (Table 1, Figures 6 and 7). Of 30 spontaneous tears in the skin and Colles' fascia observed clinically, 28 followed the midline and two were parallel to it. The vaginal tears seen cranial to the perineal body were of two types, dividing the midline or following one or both vaginal fornices (Table 1). The vaginal tears were described clinically, but we were not able to describe and identify the structures in these tears by sonography. Caudal to the perineal body, a tear in the conjoined longitudinal muscle could be transverse. In all other perineal structures the tears had a sagittal direction. The puboperineal muscles had a tendency to retract towards their origin on the pelvic sidewalls, together with the structures in the perineal body and the perineal membrane. These structures therefore had to be identified by forceps during the repair. The circular endoanal ultrasound image was needed to confirm the diagnosis of a damaged EAS. Table 3 shows the involvement of the perineal and vaginal structures, as assessed by ultrasound imaging, according to degree of the tear.

Figure 4.

Endoanal semicircular ultrasound image taken during pregnancy. The perineal membrane (2) and the puboperineal muscle (3) meet in the midline. In this area the internal anal sphincter (7) and the conjoined longitudinal muscle (6) are attached to the perineal body; in some scans this can give a shadow with mixed echogenicity that is easily mistaken for a tear in the internal anal sphincter. 5, external anal sphincter.

Figure 5.

Transvaginal semicircular transverse ultrasound image of the perineal body and anal canal during pregnancy. The perineal membrane (2) and the puboperineal muscle (3) meet in the midline in the perineal body (1). 5, external anal sphincter; 7, internal anal sphincter.

Figure 6.

Transvaginal sagittal midline ultrasound image of the anal canal immediately after delivery in a woman with a fourth-degree perineal tear. All structures in the perineum including the rectovaginal septum are torn, and are missing from the image. The following structures can be identified dorsal to the anal canal: external anal sphincter (5), conjoined longitudinal muscle (6), internal anal sphincter (7) and puborectal muscle (PR). Anterior to the anal canal, the cranial part of the internal anal sphincter (7) remains intact. 9, rectal columns.

Figure 7.

Endoanal sagittal midline ultrasound image of the perineum immediately postpartum. The hypoechogenic internal anal sphincter (7) and the hyperechogenic conjoined longitudinal muscle (8) are squeezed by the linear-array ultrasound probe. The perineal membrane is torn (arrow) as is the puboperineal muscle. The external anal sphincter (5) appears partly torn in this image, but turned out to be totally torn at the site of mediolateral episiotomy. 1, perineal body.

Table 3. Detection rate of intact perineal structures in 40 women with third- or fourth-degree perineal tears (Group III) at ultrasound examination with linear and semicircular transvaginal and endoanal scans in the delivery room
StructureAll (n = 40)Partial third-degree tears (n = 12)Total third-degree tears (n = 20)Fourth-degree tears (n = 8)
  1. Values are n (%).

Internal anal sphincter31 (78)12 (100)19 (95)0 (0)
Puboperineal muscle4 (10)4 (33)0 (0)0 (0)
Longitudinal muscle22 (55)9 (75)13 (65)0 (0)
Perineal membrane4 (10)3 (25)1 (5)0 (0)
Central point7 (18)6 (50)1 (5)0 (0)

Table 4 summarizes data on the ability to identify the perineal structures in the linear sagittal transvaginal ultrasound image after delivery. The intraobserver agreement between the single image and the initial multiplanar evaluation (Observer 1) was moderate or good for all structures, except the perineal membrane, for which it was fair, and the central point, for which it was poor (Table 4). For the second observer, who was blinded to the clinical data, the agreement with the initial multiplanar sonography was good or moderate for the EAS and perineal membrane, fair for the central point and IAS, and poor for the puboperineal muscle and conjoined longitudinal muscle. Comparing the two observers' evaluation of longitudinal images, the agreement was good for perineal membrane, moderate for EAS, fair for puboperineal muscle, and poor for IAS, conjoined longitudinal muscle and central point.

Table 4. Ability to identify perineal structures in a single transvaginal sagittal ultrasound image immediately after delivery in 40 women with clinically diagnosed anal sphincter tears (Group III) compared with findings at multiplanar ultrasonography
Evaluation on multiplanar sonography (n = 40)Structure identified on single sagittal imageObserver 1 vs. Observer 2
Observer 1 Observer 2 
  1. Kappa index calculated for agreement between the assessment of the sagittal image by each of two observers with the initial multiplanar sonographic investigation, and for agreement between the two observers.

External anal sphincter
 Partially torn (n = 12)6 10 
 Totally torn (n = 28)1 4 
 Kappa index (95% CI) 0.53 (0.21–0.84) 0.66 (0.41–0.91)0.56 (0.27–0.86)
Internal anal sphincter
 Normal (n = 31)31 21 
 Torn (n = 9)2 3 
 Kappa index (95% CI) 0.84 (0.63–1.0) 0.27 (−0.06 to 0.60)0.14 (−0.21 to 0.48)
Perineal membrane
 Normal (n = 4)3 3 
 Torn (n = 36)8 5 
 Kappa index (95% CI) 0.29 (−0.11 to 0.70) 0.42 (0.0–0.85)0.66 (0.38–0.94)
Puboperineal muscle
 Normal (n = 4)4 3 
 Torn (n = 36)7 13 
 Kappa index (95% CI) 0.45 (0.08–0.82) 0.17 (−0.18 to 0.52)0.40 (0.09–0.70)
Conjoined longitudinal muscle
 Normal (n = 22)17 6 
 Torn (n = 18)6 3 
 Kappa index (95% CI) 0.44 (0.16–0.72) 0.10 (−0.19 to 0.39)0.08 (−0.21 to 0.36)
Central point
 Normal (n = 7)1 3 
 Torn (n = 33)6 3 
 Kappa index (95% CI) 0.04 (−0.53 to 0.45) 0.36 (−0.07 to 0.79)0.17 (−0.30 to 0.65)

At 3-month follow-up, 2/8 women with fourth-degree tears reported de-novo fecal incontinence. Among the women with third-degree tears, one reported problems with fecal urgency and fecal incontinence before pregnancy; her problems had worsened at the 3-month follow-up. One woman reported gas incontinence during pregnancy; her problems were unchanged, except that she had de-novo fecal urge after the delivery. Two of 12 women with partial third-degree tears, 5/20 with total third-degree tears and 5/8 with fourth-degree tears were incontinent for gas at least once a week at 3 months after delivery (Table 1).


To our knowledge, this is the first study that has attempted to identify the extent of ruptures in the tissues surrounding sphincter tears by ultrasonography immediately after delivery. Descriptions of how to identify the perineal and vaginal structures, other than the anal sphincter, involved in obstetric tears have previously been lacking21. The vaginal linear sagittal scan of the perineum gave an understanding of the dimensions of and relationships between the perineal structures that it was not possible to obtain on circular two-dimensional imaging. On intra-anal ultrasound examination the curvature of the anal canal is lost, resulting in a distorted topography even if three-dimensional scanning is used.

Both the muscles and the fibrous network of fascias and ligaments are needed to maintain normal function of the pelvic floor12. Careful surgical repair of these structures is mandatory after obstetric tears affecting the midline, where the healing process is impaired22 and recurrence after prolapse surgery is common23. The topographic description of the perineum in late pregnancy may be of help in identifying torn structures at repair.

The perineal membrane gives stability to the posterior compartment, and damage to this structure may lead to rectocele24. Knowledge of the variation of the shape of the perineal membrane may be of importance at repair of this structure, which may sometimes be hard to identify25.

The puboperineal muscle demonstrates a close relationship with the EAS on linear transvaginal and endoanal scanning. In semicircular endoanal imaging it can be seen as a bundle of transverse fibers cranial to the EAS, just next to the more caudal fibers of the EAS that create a complete circle (Figure 4). In circular vaginal scanning the relationship between the puboperineal muscles and the perineal membrane can be evaluated (Figure 5). The puboperineal muscles are torn together with the EAS and have a tendency to retract along the vaginal walls in the same manner as the EAS. The function of the puboperineal muscle is not known. The function of the central point is also unclear, but it has been suggested that it gives stability to the perineum11.

Both linear and semicircular scanning was necessary to fully visualize the central point. The conjoined longitudinal muscle was not easy to detect. It follows the IAS, and could be seen in the rectovaginal septum and under the EAS in endoanal circular and longitudinal scans (Figures 4 and 7). The muscle might be a factor to consider when repairing the anal sphincter muscle complex. The conjoined longitudinal muscle is described in most anatomical texts—it inserts into the subcutaneous part of EAS, and might have a role in the anal continence mechanism. No method of surgical repair has been presented when describing the repair of obstetric anal sphincter tears26.

We have not tried to describe the perineal body after delivery. In most scans it was not possible to point it out as a separate structure as the perineal membrane, the IAS and the conjoined muscle all have the same appearance in this area. The height of the perineal body is dependent on whether or not these structures were torn. The term perineal body has been used by some authors to name all fibrous structures in the perineum. We have made a distinction between different fibrous parts. We feel that these names help in sorting out torn structures when diagnosing a tear in the delivery room, as all parts need to be approximated to fully restore the anatomy.

A weakness of this study is that one of the investigators was performing the ultrasound examinations as well as the repair of most of the tears. As inspection and palpation are helpful in evaluating obstetric tears, this knowledge might have made it easier to interpret the ultrasound images. Because ultrasonography was used in determining the degree of injury, it may have resulted in too high an accuracy of the ultrasound examinations. To compensate for this, at least to some extent, the other observer was blinded as to whether each scan had been carried out during pregnancy or postpartum, and to the severity of the tears. The interobserver agreement was high for examinations during pregnancy, but quite low for the postpartum examinations, especially regarding the IAS, the longitudinal muscle and the central point. This implies that further studies are needed to assess the reliability of ultrasound imaging in evaluating these structures in perineal tears. Ultrasonography is very much operator dependent. Thus, the congruence between the two observers might have been higher if both had had the opportunity to perform the ultrasound examination27.

Another concern is that we have interpreted the structures in the perineum from their location according to descriptions in the literature and from their identification during surgical repair. Thus, it cannot be fully ascertained that structures not previously described ultrasonographically were correctly identified, as we did not have any ‘gold standard’ available from, for example, dissections of cadavers.

In summary, linear and semicircular transvaginal and endoanal ultrasonography seems to provide valuable information on perineal anatomy before and after parturition. Ultrasonography might be helpful in evaluating the extent of perineal tears before surgical repair. However, the relatively poor agreement between the two observers evaluating single linear transvaginal images implies that a combination of transverse and longitudinal projections is necessary to obtain relevant information. Further studies are needed regarding the importance of specific sonographically identified structures and their role in postpartum pelvic floor dysfunction.


Ann-Kristin Örnö was supported by the Grants to Researchers in the Public Health Service from the Swedish Government and from the Region Skåne. Ultrasound equipment was kindly provided by Hitachi Ultrasound Holding AG, Zug, Switzerland.