INTRODUCTION

Magnetic resonance imaging (MRI) is increasingly used by gynecologists and especially reproductive surgeons to assess the genital tract in a variety of clinical situations that require precise detail of the reproductive organs. Sufficient detail is provided in most circumstances by transvaginal ultrasonography (TVUS) or saline infusion sonohysterogram. In some instances, such as in determining the size and location of fibroids or discerning adenomyosis, neither TVUS nor saline infusion sonohysterogram is definitive. Furthermore, accurate determination of a müllerian anomaly may allow more precise preoperative assessment and even potentially eliminate a surgical step, such as a diagnostic laparoscopy for resection of a uterine septum.

Magnetic resonance imaging is gaining widespread acceptance and, in many instances, is a cost-effective tool in the evaluation of abnormal uterine bleeding. It is noninvasive, differentiates uterine anatomy in response to exogenous hormones or the normal menstrual cycle, and reliably localizes pelvic pathology and size of lesions. When uterine conservation is desired in women with fibroids, and TVUS or saline infusion sonohysterogram is indeterminate in localizing the depth of myometrial involvement of a fibroid, MRI should be considered as a part of the clinical algorithm. The precision of MRI localization of submucosal fibroids can obviate the need for hysterectomy and permit a skilled surgeon to hysteroscopically resect the fibroids. If the clinical examination is suspicious for adenomyosis and the TVUS is nondiagnostic, the clinician should strongly consider MRI. When the results of the imaging study would influence surgical route and planning, MRI should be considered in the preoperative evaluation.

This chapter gives an overview of the physics of MRI and gives a visual overview of the potential clinical applications of this imaging modality.

PRINCIPLES OF MAGNETIC RESONANCE IMAGING

Clinical MRI takes advantage of the physical properties of the protons in the body to produce anatomic images. All nuclei—in this case, hydrogen nuclei—have the physical properties of charge and spin. Depending on the charge and spin, certain atoms, including hydrogen nuclei (protons), behave as tiny magnets and have both polarity and an associated magnetic field with the properties of magnitude and direction. When placed in a magnetic field, they tend to align with the field in a lower energy state or against the field in a higher energy state.

When the system is at rest, in equilibrium, there are a relatively tiny number of excess protons in the lower energy state. These are the protons that are available for use in MRI. Because of the physical properties of the system, these magnetic fields do not remain perfectly aligned with the main magnetic field but rather they rotate or precess about the central axis of the main magnetic field, much as a spinning top precesses in the presence of gravity. This precession is very important in MRI because it is essentially the only parameter in the system that we can either change or measure.

The Larmor equation tells us that the precessional frequency is equal to the strength of the magnetic field in which a particular proton of interest resides times a constant called the gyromagnetic ratio. The gyromagnetic ratio is fixed for a given nucleus. At the magnetic field strengths used in MRI, the precessional frequencies are in the radiofrequency range. If we know the strength of the magnetic field in a particular location, we can predict the precessional frequency of the protons in that location. Analogously, if we change the magnetic field in a predefined manner and if we can measure the precessional frequencies of various protons, we can calculate their positions as long as we know the magnetic field variations.

Precessing protons will only absorb energy applied at their precessional frequency. When we introduce radiofrequency energy at the appropriate frequencies, the protons precessing at these frequencies absorb the energy and move to a higher energy state. As these protons relax back to equilibrium, radiofrequency signals can be detected at their particular precessional frequencies.

In MRI, we use these principles to create images. During imaging, appropriate radiofrequency energy is put into the system while supplementary magnetic fields, called gradients, are added to or subtracted from the main magnetic field in a predefined linear fashion. Because the scanner can vary the gradients in all directions, the process of spacial localization, identifying the location of a particular signal in three-dimensional space, can be achieved. Multiple iterations of the imaging examination are performed and the scanner collects and reconstructs the obtained data, assigning the signal from the protons to the appropriate location on the image until the entire image is formed.

Varying certain acquisition parameters, especially timing, changes the images so that their contrast varies depending more on their interactions with each other (T₁-weighted imaging) or on their interactions with their local environment (T₂-weighted imaging). Many other techniques and strategies are employed to vary the contrast, sensitivity, and efficiency of MRI and much more complete descriptions are available to the interested reader.¹

MRI VERSUS OTHER IMAGING MODALITIES

MRI has many advantages over ultrasound and computed tomography (CT) for the evaluation of pelvic anatomy, congenital anomalies, and pathologic disorders. Ultrasound, the workhorse of pelvic imaging, is considered convenient, efficient, and cost-effective. Ultrasound is easily accessed by gynecologists for evaluation of their patients and is an ideal screening examination. Among its many uses are the identification and follow-up of cysts, endometrial abnormalities, and fibroids. However, even in the best of hands, ultrasound remains operator-dependent and patient limited.

Pelvic CT, performed in the axial plane, is limited by surrounding osseous structures and by the soft tissue contrast produced by differential absorption of the X-ray beam. The patient is exposed to ionizing radiation and likely to an intravenous contrast medium to exploit variations in tissue blood supply. MRI is also useful in the identification and differentiation of simple, proteinaceous, or hemorrhagic fluid as well as in the identification of fluid septations and loculations.

The strengths of MRI include sensitivity to soft-tissue contrast unparalleled in other imaging modalities, multiplanar capability, less limitation by uterine and in most cases patient size (especially relative to ultrasound), and the lack of ionizing radiation. With recent advances in sequence and hardware technology, the classic MRI limitation of sensitivity to motion, whether it derives from patient respiration or peristalsis, is decreasing.

The following sections include illustrative cases of normal anatomy and benign pathology frequently encountered by reproductive surgeons. The focus is on congenital and acquired pathology, but we do not discuss differentiation of a benign or malignant cyst. The assessment of malignant versus benign cysts is most commonly determined by ultrasonography (see Chapter 30).

NORMAL ANATOMY

The normal uterus, cervix, vagina, endometrium, and ovaries can be readily identified on T₂-weighted images in a variety of planes. Also the plicae palmitae, the anterior and posterior fornix, the cul-de-sac, rectum, urinary bladder, and urethra are easily identified (Fig. 31-1). The uterine MRI anatomy can be subdivided. The signal intensity and thickness of these layers can vary based on the hormonal milieu. The myometrium is seen externally and tends to be intermediate in signal intensity. Centrally is the endometrium, which tends to be of increased signal on T₂-weighted sequences. Between the two layers is a much darker, uniform layer, called the junctional zone. Although histologically similar to the adjacent myometrium, the difference in signal intensity is believed to result from a lower water content relative to the adjacent myometrium³ and the compact longitudinally oriented smooth muscle bundles oriented parallel to the endometrium⁴ (see Fig. 31-1). The radiologist, when assessing the uterus for pathology, evaluates the thickness, signal characteristics, and distinctness of each of these zones.

Figure 31-1 Normal anatomy. Axial (A,E), sagittal (B,C,F), and coronal (D) T₂-weighted images. Normal structures including the cervix (c), vagina (v), endometrium (e), ovaries (o), plicae palmitae (pl), the anterior (a) and posterior (p) fornix, the cul-de-sac (s), rectum (r), urinary bladder (b), and urethra (ur), are visible. The myometrium (m), endometrium (e), and junctional zone (jz), as well as nabothian cysts (arrow) are also identified. Also visible are the external cervical os (os), junctional zone (jz), lateral fornices (lf), and coccyx (cc).

CONGENITAL ANOMALIES OF THE REPRODUCTIVE TRACT

The MRI appearance of müllerian anomalies directly corresponds to the findings on both hysterosalpingography and gross examination. Several potential anatomic observations can be made. In cases of uterine agenesis, no vaginal or uterine tissue is present (Fig. 31-2).

Figure 31-2 Vaginal and uterine agenesis. Sagittal T₂-weighted image of a patient with vaginal and uterine atresia. Bladder (b) and rectum (r) are identified; uterus, cervix, and vagina are absent.

The arcuate uterus is a frequent müllerian anomaly that usually has little clinical significance and is identified as a contour or “bulge” of the fundus, with no changes associated with the endometrial canal (Fig. 31-3).

Figure 31-3 Arcuate uterus. Oblique axial T₂-weighted image demonstrating the contour bulge (b) at the fundus without external fundal contour deformity.

Several anomalies are easily determined by MRI, such as a unicornuate uterus where a single horn can be identified (Fig. 31-4). However, others are more difficult, especially when the decision of whether to proceed to surgery relies on the MRI diagnosis.

Figure 31-4 Unicornuate uterus. Axial T₂-weighted image of a unicornuate uterus demonstrating a solitary uterine horn (u); urinary bladder (b).

The most frequent differential diagnosis that a radiologist is asked to make on MRI is between a septate and bicornuate uterus. In a septate uterus, the septum can be seen extending from the fundus to any level in the uterine cavity up to the internal os. MRI can distinguish between a muscular and a fibrous septum based on the signal characteristics. A fibrous septum is of low signal on T₂-weighted sequences, whereas the signal of a muscular septum will follow that of the myometrium, if obtained in the correct plane (Figs. 31-5 to 31-9). The basic concept holds that a bicornuate uterus would have a muscular septum and a septate uterus would be more fibrous. A septate uterus will have some “dimpling” of the fundus of the uterus, but the external surface of the uterus does not form two horns. In a bicornuate uterus the two horns are identified and the external fundal surface is clearly divided (Figs. 31-10 and 31-11). In a uterus didelphys the separate canals and cervices are easily identified. In cases where there is a vaginal septum, it is also easily identified (Fig. 31-12).

Figure 31-5 Septate uterus. Coronal T₂-weighted image through the uterus demonstrating a myometrial septum (ms) in the fundus and the fibrous portion of the septum (fs) extending to the cervix (c). Note the fundal contour deformation (fd) accentuated by the urinary bladder.

Figure 31-6 Septate uterus. A, In this case, the axial T₂-weighted image shows that the muscular septum (s) has the same signal as the myometrium and extends nearly the entire length of the canal; cervix (c), fundus (f), junctional zone (jz). B, Coronal image through the uterine body; myometrium (m), endometrium (e), and bladder (b). C, A coronal T₂-weighted image through the abdomen in a patient with a septate uterus demonstrates a solitary kidney (k). The liver (l) and spleen (spl) are also seen.

Figure 31-7 Septate uterus. A, In this case, the oblique coronal T₂-weighted images demonstrate the shallow fundal deformity (fd) and the thick muscular septum (s) that extends to the lower uterine segment. B, The oblique axial T₂-weighted image demonstrates the muscular septum (s) and the endometrial canals (e). The bladder (b) and normal ovary (o) are included on these images.

Figure 31-8 Septate uterus. The uterus of this patient, best seen on reconstructed oblique axial T₂-weighted images (A,B), has a dark fibrous septum (fs) that extends into the cervix (c).

Figure 31-9 Septate uterus. This is best appreciated on oblique coronal T₂-weighted image through the uterine canal demonstrating myometrial septum (ms).

Figure 31-10 Bicornuate uterus with atretic right horn (ah). T₂-weighted images in the axial (A) and coronal (B) planes show a normal-appearing left horn (h) and an atretic horn, which lacks the signal characteristics of endometrium and junctional zone.

Figure 31-11 Bicornuate uterus. Coronal image through the plane of the uterus demonstrating the large gap between the two uterine horns (h), which join at the lower uterine segment (lus), and the single cervix (c).

Figure 31-12 Uterus didelphys. Series of contiguous axial images (A–C) with two uterine canals and cervices.

In patients who present at puberty with acute pain and amenorrhea, the differential diagnosis should include an intact hymen, vaginal septum, and congenital absence of the cervix. In these cases, blood products of varying ages may be identified in the vaginal canal or uterus. Blood products range from bright through the spectrum of grey to dark. The darkest grey (almost black) represents hemosiderin (Fig. 31-13).

Figure 31-13 Uterus didelphys with obstructive hymen and atretic right uterus (r). A, Axial T₁-weighted high-signal material in dilated cervical canal (c); B, through a lower level. C and D, Axial T₂–weighted images at the same level as A and B, with sagittal (E) and coronal (F,G) images through dilated cervix and vagina with blood central and along the periphery (as in B and D). G also includes each uterus, left (l) and right (r). The signal of the blood in the cervical canal and vagina varies depending on the age and oxygenation of the blood products (see text). Uterus didelphys with obstructive hymen and atretic right uterus (r). (E) Sagittal and coronal (F,G) images through dilated cervix and vagina with blood central and along the periphery (as in B and D). G also includes each uterus, left (l) and right (r). The signal of the blood in the cervical canal and vagina varies depending on the age and oxygenation of the blood products (see text).

ACQUIRED ANOMALIES OF THE REPRODUCTIVE TRACT

The most common acquired anomalies of the reproductive tract that may require radiologic help to decide on appropriate treatment are leiomyomas and adenomyosis.

Adenomyosis

The diffuse or localized forms of adenomyosis can be accurately detected with MRI. In diffuse adenomyosis, the thickness and signal characteristics of the junctional zone are evaluated. If the junctional zone is greater than 10 to 12mm thick, adenomyosis has been shown to be present. Rarely, uterine contractions⁵ have been shown to produce segmental thickening of the junctional zone, hence the reliance on these values. In adenomyosis, the junctional zone may contain regions of bright or high signal intensity on the T₁– or T₂-weighted images, corresponding to regions of hemorrhage⁶ (Figs. 31-14 to 31-17).

Figure 31-14 Adenomyosis. Sagittal (A,B) and coronal (C) transvaginal ultrasound demonstrates a heterogeneous focal region felt at the time to represent a fibroid. Axial T₁-weighted (D), axial T₂–weighted (E), and coronal (F) images from an MRI performed on the same patient show the changes of adenomyosis, a region of decreased signal in T₂ containing punctate foci of high signal on T₁-weighted and T₂-weighted sequences, corresponding to blood degradation products. Note in the affected area the loss of a well-defined junctional zone.

Figure 31-15 Adenomyosis. A, Axial T₂ -weighted and B, T_1-weighted images through focal adenomyosis with loss of junctional zone and foci of high signal on both sequences.

Figure 31-16 Diffuse adenomyosis. Adenomyosis involves the majority of the uterus; axial T₁-weighted (A), axial T₂-weighted (B), sagittal T_2-weighted (C), and coronal T_2-weighted (D) images with loss of the junctional zone in the affected region. Foci of high signal on both T₁-weighted and T₂-weighted sequences represent blood products.

Figure 31-17 Diffuse adenomyosis. Sagittal T₂-weighted (A) and axial T₂-weighted (B) images show complete involvement of the uterus with adenomyosis. C, The axial T₁-weighted image shows small foci of high signal.

A focal adenomyoma may appear as mass with heterogeneous to low signal intensity, with an appearance that may be similar to a leiomyoma. However, the margins should be less distinct, corresponding to the interdigitation of the adenomyoma with the myometrium (see Fig. 31-15).

Leiomyoma

These benign tumors have a characteristic appearance on T₂-weighted images. The margins are distinct from the surrounding myometrium, and the signal intensity is less than the surrounding myometrium, approaching that of the junctional zone. The degenerated myomas contain blood, blood degradation products, mucin, or hyaline and will have a heterogeneous appearance. Although MRI can distinguish these more complex myomata from the “typical” myoma, it cannot distinguish which of these substances are present. MRI has also proven useful in distinguishing fibroids from other solid pelvic masses and for localization, which may affect therapy⁷ (Figs. 31-18 and 31-19). However, there are no clear criteria that can distinguish malignant from benign lesions.

Figure 31-18 Fibroids. T₂-weighted sagittal (A) and axial (B) images of multiple uterine fibroids of low signal relative to the myometrium throughout the uterus in multiple locations, including submucosal (sm), myometrial (m), and exophytic (exo).

Figure 31-19 Fibroids. T₂-weighted sagittal (A) and coronal (B) images of myometrial fibroid (f) with internal degeneration, represented by internal areas of high signal.

Post-Cesarean Section Defects

The high frequency of cesarean sections and potential dehiscence or rupture at the scar site has led to some interesting although clinically obscure observations on MRI. The scar or myometrial defect that corresponds to the surgical site can be identified on MRI (Fig. 31-20). However, there are no good data on the correlation between the scar characteristics and potential future reproductive outcomes.

Figure 31-20 Cesarean section defect. The T₂-weighted sagittal image shows the contour deformity (arrow) in the lower uterine segment.

Cervical Anomalies

The anatomy of the cervix has similar components as the uterus. The benign abnormality identified is nabothian cysts. Because nabothian cysts contain fluid, they follow fluid on the T₂-weighted sequences and are sharply contrasted from the surrounding stroma (Fig. 31-21).

Figure 31-21 Nabothian cysts. T₂-weighted coronal (A) and sagittal (B) images of the uterus, showing a cluster of small cysts aligned along the endocervical canal (arrow).

OVARY

Normal ovaries are typically identified by the follicles and cysts (Fig. 31-22; see also Fig. 31-1). A major indication for MRI is often to distinguish a persistent physiologic cyst from a pathologic one. Hemorrhagic cysts are identified by high signal on a T₁-weighted sequences without loss of signal on fat-suppressed sequences. This indicates blood or proteinaceous material within the cyst lumen (Fig. 31-23).

Figure 31-22 Ovary. Normal physiologic follicle (p) in right ovary on axial T₂-weighted image through the pelvis.

Figure 31-23 Blood products in an ovarian cyst. In a right ovarian cystic lesion (o) there is homeogeneous high signal on the axial T₁–weighted (upper) image. On the T₂-weighted (lower) image there is a gradient of progressive signal loss called shading. This represents a gradient of blood degradation products, as can be seen in endometriomas or more chronic hemorrhagic ovarian cysts.

Dermoid cysts are identified mainly by the fat component, with a sequence that will have fat as high signal intensity (bright) and a sequence that suppresses the fat signal (dark) lesion (Fig. 31-24).

Figure 31-24 Dermoid tumor. Axial images through the pelvis show a left adnexal mass that is high signal on T₁-weighted (A) and T₂-weighted (C) images in a frondlike low signal central region (arrow). The high signal component loses signal on B, a T₁–weighted, fat-saturated sequence. The loss of signal on the fat saturation signal corresponds to fat. These characteristics are diagnostic for a dermoid tumor.

Endometriosis cysts can be identified by the cyst contents, which will include blood or proteinaceous fluid of varying ages. Typically, there is a peripheral dark signal, indicating hemosiderin. There may be a relatively large amount of associated soft tissue, which would represent the associated inflammatory response or adhesions. Although MRI may identify endometriomas as small as 5 mm, it cannot identify the sheetlike peritoneal involvement (Figs. 31-25 and 31-26).

Figure 31-25 Endometriosis. A, T₁–weighted and B, T₂-weighted sagittal sequence showing a soft-tissue signal mass (m) in the cul-de-sac that contains small foci of high signal on both sequences. C, T₁-weighted image after IV gadolinium shows enhancement through this tissue.

Figure 31-26 Endometrioma. A, Axial T₁-weighted image through the pelvis demonstrating high signal collection (r) in the right adnexa and in the left adnexa, two locules: a medial locule (ml) and lateral locule (ll). B, fat-saturated T_1-weighted image shows no change in the signal characteristics of the collections. The medial locule has high signal and the lateral locule is of low signal This excludes extracellular fat producing the high signal in (r) and (ml). C, T₂-weighted axial image at the same level. The signal characteristics change based on the age of the blood degradation products (see text).

In peritoneal inclusion cysts, typically the ovary is not visualized within the fluid. The pelvic fluid will appear as a loculated collection (Fig. 31-27). Potential malignancies can be identified on MRI by similar criteria as those used in ultrasonography.

Figure 31-27 Peritoneal inclusion cysts. A, Axial and B, multiple sagittal T₂-weighted images show pelvic fluid collection in the cul-de-sac (cds) containing multiple septations. C, axial T₁-weighted, fat-saturated sequence before and after IV gadolinium shows enhancement in the septae (sep) and uterus (u) but no enhancement in the fluid.

URETHRA

Chronic pelvic pain may have any of a large number of potential causes. MRI may help in identifying the cause; this imaging modality is often ordered to evaluate for adenomyosis or endometriomas as a cause of chronic pain. However, occasionally a urethral diverticulum is observed that some believe may contribute to the pain syndrome.

A urethral diverticulum is identified as a fluid collection adjacent to the urethra (Fig. 31-28). At times, the neck of the diverticulum can be identified. On occasion, complex fluid (typically proteinaceous) is identified. Only the wall of the diverticulum should enhance, without any nodularity, which would be concerning for a neoplastic process.

Figure 31-28 Urethral diverticulum. A, Periurethral fluid collection (arrow) extending from 12 to 6 o’clock position on axial T₁-weighted sequence; the contents have low signal. B, T₁-weighted fat-saturated sequence after IV gadolinium with peripheral enhancement and C, T₂-weighted sequence; the contents have high signal. These signal characteristics indicate that the diverticulum contains fluid (not blood or proteinaceous material). No nodule is within the fluid.