**Abstract**

The classic imaging diagnosis of endometriomas encounters multiple limitations, including the subjective evaluation of medical examinations and a similar imaging appearance with other adnexal lesions, especially the functional hemorrhagic cysts. For this reason, a definite diagnosis of endometriomas can be made only by pathological analysis, which reveals particular features in terms of cellularity and biochemical components of their fluid content. It is theorized that these histopathological features can also be reflected in medical images, altering the pixel intensity and distribution, but these changes are too subtle to be assessed by the naked eye. New quantitative imaging evaluations and emerging computer-aided diagnosis techniques can provide a detailed description of image contents that can be furtherly processed by algorithms, aiming to provide a more accurate and noninvasive diagnosis for this disease.

**Keywords:** computer-aided diagnosis, endometrioma, endometriosis, MRI, texture analysis

## **1. Introduction**

Laparoscopic biopsy of suspicious-looking lesions, followed by histologic confirmation, is the gold standard for diagnosing pelvic endometriosis [1]. The first line imaging of ovarian endometriotic lesions (endometriomas) remains transvaginal ultrasonography (TVUS) which is able, in most situations, to offer sufficient information for adequate preoperative planning [2]. Other diagnostic procedures, such as magnetic resonance imaging (MRI), are used in certain circumstances based on the results of the TVUS and the severity of the symptoms. The pelvic MRI scan provides for full lesion mapping, with a high detection rate for both anterior and, particularly, posterior lesions, whereas TVUS shows lower sensitivity rates [3]. MRI shows higher accuracy for the detection and characterization of endometriotic lesions than other imaging modalities, therefore it is often used to evaluate adnexal masses and monitor treatment response, potentially avoiding the need for a follow-up laparoscopy [4]. Moreover, this method is especially usefull in the detection of deep infiltrating

endometriosis, being absolutely required in this subcohort of patients. As with any other imaging modality, the MRI evaluation of endometriosis is limited by the technique itself, but more by the examiner's experience and level of training, since all classic MRI signs of this disease are qualitative and entire subjectively evaluated.

One of the most challenging tasks in the diagnosis of endometriomas is distinguishing these lesions from functional hemorrhagic ovarian cysts (HCs) since they share many both imaging and histological characteristics. To avoid unnecessary surgery, it's critical to correctly distinguish the two lesions [5]. As a result, the difference in imaging between the two entities has a significant impact on the subsequent medical and surgical treatment options [6]. This chapter focuses on the emerging quantitative imaging modalities that can improve the diagnostic and characterization of endometriomas and may aid to distinguish these lesions from HCs.

#### **2. MRI signal intensity measurements**

The "T2 shading" sign is currently considered a characteristic MRI finding of endometriomas [7]. This sign refers to a cystic lesion with a high signal on T1-weighted (T1W) sequences and subsequent T2 shortening which results in hypointensity on T2-weighted (T2W) images, as a result of in-lesion hemorrhage and accumulation of blood products and proteins [8, 9].

The "T2 shading" was first described by Togashi and colleagues [10] in 1991, who attributed it high diagnostic value (98% sensitivity and 96% specificity). Following further investigation, more recent studies demonstrated that this sign is not as specific to endometriomas as originally thought, mostly due to the subjective character of the imaging findings [11] and the occurrence of this sign in other ovarian cysts accompanied by bleeding [12–14]. The acknowledgment of these limitations drastically decreased the utility of this sign in further studies (with a sensitivity and specifity as low as 68% and sensitivity and 14.2%, respectively specificity) [11, 12].

It is important to remember that some HCs have a delayed regression and may accumulate blood products, which also leads to a decrease in their intrinsec signal on T2W sequences [14]. However, it is expected that HCs would accumulate f blood products to a lesser degree compared to endometriomas, since they usually regress within a few menstrual cycles and do not exhibit cyclic intra-lesional bleeding. In this regard, Outwater et al. [11] concluded that endometriomas tend to


#### **Table 1.**

*The diagnostic ability of the "T2 shading" sign for identifying endometriomas.*

*Quantitative Imaging Parameters in the Diagnosis of Endometriomas DOI: http://dx.doi.org/10.5772/intechopen.101561*

#### **Figure 1.**

*The magnetic resonance (MRI) representation of the "T2 shading" sign, based on the examination of a patient with two histologically-proven endometriotic lesions. (A) On the axial T1-weighted image, both lesions express similar high-signal intensity (green arrows). (B) On the T2-weighted image, the two lesions express different degrees of signal drop (green arrows).*

have higher T1 and lower T2 signal intensities (SI) than HCs, thus creating a more abrupt "shading" phenomenon. However, previous studies did not provide a clear definition of the extent of the signal difference between T1W and T2W sequences, and the degree required for this in order to become a confident diagnostic criterion (**Table 1**). An adequate example of the variation of this signs' appearance is demonstrated in **Figure 1**.

Lately, a small-cohort study [18] aimed to differentiate endometriomas from HCs by quantifying the "T2 shading" sign through signal intensity (SI) measurements made by placing regions of interest (ROIs) within the lesions on T1 and T2 weighted-images (WI). The signal intensity difference was quantified by subtracting SI values between T1 and T2 WI (A = T1 − T2). There were statistically significant differences between the two entities only when comparing T1 SIs (p = 0.0003) due to the much higher values obtained by endometriomas, while the T2 SIs were very similar and did not differ significantly (p = 0.27). As expected, endometriomas demonstrated a higher loss (median SI loss = 432.95 units) while HCs recorded negative results (median SI loss = −46.8 units). In most cases, the values recorded by HCs on T2 WI were higher than the ones on T1 WI, which could be a consequence of the HCs' blood content being in different stages of degradation and to an overall lesser amount of blood products or protein accumulation (or to a higher percentage of intrinsically present fluid – which is usually not present in endometriomas). These results are in accordance with the early observations of Outwater et al. [11] that were mentioned earlier.

Through the quantitative appreciation of the "shading" sign, using a cut-off for the signal drop of more than 31.3 SI units, endometriomas could be differentiated from HCs with 100% sensitivity and 81.82% specificity. Therefore, this study [18] concluded that the key in identifying endometriomas relies upon a brighter T1 appearance of the lesions, which cannot be always appreciated by the visual evaluation (**Figure 2**). As the proposed measurement technique is rather basic, it could be easily translated into daily clinical practice, possible offering more confidence in the MRI diagnosis of these lesions.

#### **Figure 2.**

*Quantitative assessment of the "T2 shading" sign in endometriomas and functional hemorrhagic cysts. (A and B) The MRI examination of a 32-year old patient with endometrioma. The difference in signal intensity between (A) T1 (1146.61 units) and (B) T2 (318.93 units) was 827.68 units. (C and D) The MRI examination of a 27-year old patient with an ovarian hemorrhagic cyst. The difference in signal intensity between (C) T1 (413 units) and (D) T2 (420.19 units) was* −*7.19 units.*

### **3. Diffusion-weighted imaging**

MRI incorporates a functional technique known as diffusion-weighted imaging (DWI). DWI sequences provide information about the Brownian motion of water molecules in a tissue [19]. Based on these sequences, apparent diffusion coefficient (ADC) maps can be computed, and together they offer qualitative and quantitative information about tissue density [20]. Tissues that are highly cellular or have cellular swelling show lower water diffusion coefficients which translate to higher SI on DWI sequences and lower values on ADC maps. The ADC value can be quantified as SI values through ROI placement, and can also be used as a marker of cellularity [20]. In recent years, these sequences have evolved as a new tool for the molecular characterization of pathological fluid collections [21]. Technically, the DWI sequences are constructed by acquiring T2-based images at different b-values, through diffusion-sensitizing gradients turned on at various strengths [22]. These b-values reflect the strength and time of the gradients employed to generate such diffusion-weighted images. Subsequently, the diffusion effects are closely linked to the b value [23]. In classic pelvic examinations, ADC maps are automatically generated using all acquired b values. An adequate example is displayed in **Figure 3**. The utility of this technique in differentiating endometriomas from HCs was also investigated in several studies, with contradictory results (**Table 2**).

Overall, the studies conducted by Lee [12] and Balaban [24] showed a statistically significant difference between the ADC values measured in the two groups, while no such difference was observed in a third study conducted by Lupean [18]. Interestingly, in the studies coordinated by Balaban [24] and Lupean [18], the recorded ADC values were higher for HCs than for endometriomas, while Lee [12] obtained opposite results. It is possible that due to the short-living nature of HCs, these lesions do not have the necessary time to build up blood and degradation products, and therefore they produce have a higher motion degree of the water molecules and therefore a lesser decrease of ADC values. However, the differentiation of the two entities based on diffusion

*Quantitative Imaging Parameters in the Diagnosis of Endometriomas DOI: http://dx.doi.org/10.5772/intechopen.101561*

#### **Figure 3.**

*(A) Axial T2 fat sat, (B) Axial DWI (b = 200), and (C) ADC map of a left-ovarian endometriotic lesion (red arrows). (D) Axial T2, (E) axial DWI, and (F) ADC map of a right-ovarian hemorrhagic cyst (green arrows). There is an obvious lower ADC signal intensity for the endometrioma (red arrow in (C)) than for the hemorrhagic cyst (green arrow in (F)).*


*n, number of patients; HCs, hemorrhagic cysts; ADC, median ADC values expressed as number × 10−3 mm2 /s; cut-off, ADC cut-off value expressed as number × 10−3 mm2 /s; p-value, univariate analysis result; Se, sensitivity; Sp, specificity.*

#### **Table 2.**

*The main results that were obtained by studies that focused on the role of differentiating endometriomas from hemorrhagic cysts through apparent diffusion coefficients.*

sequences seems unreliable, considering the changing nature of HCs and therefore the variable results obtained in different studies. Therefore, the MRI diffusion techniques may be less suitable for the characterization of ovarian endometriotic lesions.
