1. Introduction
Polycystic ovary syndrome (PCOS) is an endocrine and metabolic disease commonly seen in women of reproductive age, often accompanied by excessive androgen, ovulation dysfunction and polycystic ovary, and is the main cause of female reproductive and metabolic dysfunction.[1] Global epidemiological studies have shown that the prevalence of PCOS in women of reproductive age is 5 to 10%, whereas data based on the Rotterdam criteria indicate a prevalence of more than 20%.[1,2] As an endocrine disorder, PCOS is usually characterized by irregular menstruation (amenorrhea, oligomenorrhea, and anovulation) and hyperandrogenemia (hirsutism, alopecia, and acne), and is often accompanied by complications such as obesity, insulin resistance, and cardiovascular disease.[3–5] In addition, obesity, hirsutism, alopecia and acne can increase the risk of mental health problems, as well as lead to reduced quality of life.[6–8]
According to the Rotterdam diagnosis standard, patients with PCOS have the presence of ≥ 12 follicles 2 to 9 mm in diameter in at least 1 ovary and an ovarian volume >10 cm3.[9] In 2004, European Society of Human Reproduction and Embryology (ESHRE) and American Society for Reproductive Medicine (ASRM) defined the ultrasonic characteristics of PCOS.[10] The specific ultrasound signs of PCOS include an enlarged ovary surrounded by multiple small follicles, an enlarged ovarian stroma, and an enhanced stromal echogenicity.[11] Ultrasound is considered to be an effective diagnostic tool for PCOS.
PCOS is a metabolic and reproductive disorder characterized by ultrasonic, clinical and biochemical changes.[12] In addition to transabdominal ultrasound, more recently, uterine changes and ovarian conditions in PCOS patients have been evaluated by transvaginal Doppler ultrasound. The purpose of this study was to quantify total ovarian volume, ovarian medullary area, peak systolic velocity (PSV), and end-diastolic blood flow velocity (EDV) using Doppler ultrasound, calculate the resistance index (RI), and compare these data with healthy subjects to quantify the ultrasound characteristics of PCOS.
2. Materials and methods
2.1. Participants
This study has been approved by the ethics committee of the Third Affiliated Hospital of Qiqihar Medical University (approval no. 2024LL-37). This study was designed as a case-control study.
A total of 100 PCOS patients (PCOS group) from the Third Affiliated Hospital of Qiqihar Medical University from January 2024 to December 2025 and 100 healthy subjects (control group) who visited our hospital during the same period were selected as the study objects. PCOS was diagnosed according to the 2003 revised Rotterdam criteria (Rotterdam ESHRE/ASRM-Sponsored PCOS Consensus Workshop Group), requiring at least 2 of the following 3 features: oligo-ovulation or anovulation (manifested as menstrual cycle length > 35 days or amenorrhea > 6 months); clinical and/or biochemical hyperandrogenism (hirsutism, acne, and/or elevated serum total testosterone); and polycystic ovary morphology on transvaginal ultrasound, defined as ≥ 12 follicles measuring 2 to 9 mm in diameter in at least 1 ovary and/or ovarian volume > 10 mL. All patients met at least 2 of these 3 criteria for diagnosis.
PCOS patients aged between 28 and 40 years old, with the main clinical manifestations of infertility, hypertrichosis, acne, obesity, increased total blood testosterone levels, and abnormal menstrual cycles. Infertility was defined as the inability to conceive after 12 months of regular unprotected intercourse. Obesity was defined as a body mass index ≥ 30 kg/m2.
Abnormal menstrual cycles were characterized by oligomenorrhea (cycle length > 35 days) or amenorrhea (absence of menstruation for > 6 months).
Inclusion criteria were: all patients had symptoms of infertility, obesity, acne and hirsutism; diagnosed in our hospital; voluntary participation in the experiment; meet the requirements of the Medical Ethics Committee; sexual life history.
The exclusion criteria were: exclusion of other endocrine diseases; exclusion of abnormal follicular development caused by congenital malformations of the ovary; exclusion of patients with mental disorders and cognitive disorders; exclusion of patients with a history of sex hormone therapy within 3 months; malignant tumors of other organs; presence of ovarian neoplasms, functional ovarian cysts, or endometriosis that could mimic PCOS sonographically.
2.2. Doppler ultrasound
Resona R9 made by Shenzhen Mindray Bio-Medical Electronics Co., Ltd. (Shenzhen, China) and Voluson E8 made by GE HealthCare (General Electric Company, Chicago) were selected for transvaginal and transabdominal examination on the 3rd day after menstruation was cleared. During the transabdominal ultrasound examination, the patient was instructed to properly fill the bladder and to be in a supine position, and the bilateral ovaries were examined by multi-section scanning.
All transvaginal ultrasound examinations were performed after obtaining explicit informed consent from the patients and in the presence of a female chaperone.
The transvaginal ultrasound was performed with a probe with a diffusion angle of 120 degrees, a frequency of 5.0 to 7.5 MHz, and a Doppler sampling volume of 1 to 2 mm. Before the examination, the patient was instructed to empty her bladder, lie in the supine position, apply a small amount of coupling agent and then put a condom over the probe and place it at the posterior vaginal dome, and then rotate the probe appropriately to adjust the direction of the examination to observe the ovaries.
Morphological changes and echo characteristics of bilateral ovaries were observed in both tests, and ovarian volume was measured, total volume and medullary volume were recorded.
For the color Doppler evaluation, the settings were optimized for low-velocity flow: the pulse repetition frequency (scale) was set at 0.3 to 0.6 kHz, wall filter at 50 Hz, and color gain was adjusted to minimize artifacts while maintaining sensitivity. The angle of insonation was kept below 60 degrees. The vascular study focused on arterial flow, specifically targeting the medullary branches of the ovarian artery.
Color Doppler sampling frame was used to find bilateral ovarian arteries for blood flow spectrum measurement, and hemodynamic parameters were recorded. These included EDV, PSV, and RI. After gynecological treatment of PCOS patients, the ovaries were reviewed every 3 months until the patient ovulated normally, and the corresponding data were recorded again and the results were analyzed.
2.3. Statistical analysis
The data were analyzed using SPSS 26.0 software. Measurement data were expressed as mean +/- standard deviation. Independent sample t-tests were used for comparisons between groups, while paired t-tests were employed for comparisons within groups.
Counting data were presented as cases or percentages (%) and compared using the chi-squared test. The Mann–Whitney U test was used for continuous variables that did not follow a normal distribution. Receiver operating characteristic (ROC) curves were constructed to evaluate the diagnostic performance of the ultrasound parameters.
The area under the curve (AUC), optimal cutoff values (determined by the Youden index), sensitivity, specificity, and accuracy were calculated for each parameter. A P value < .05 was considered statistically significant.
3. Results
3.1. Comparison of transvaginal and transabdominal ultrasound image characteristics
Transabdominal 2-dimensional ultrasound, which makes it more difficult to show medullary and cortical blood flow characteristics, cannot provide an accurate basis for PCOS (Fig. 1A). In contrast, transvaginal ultrasound can more accurately observe the internal structure of the ovary (Fig.
1B), and transvaginal color Doppler ultrasound further enables the assessment of ovarian vascular hemodynamics (Fig. 1C), thus improving the diagnostic accuracy of PCOS.
3.2. Comparison of ultrasound characteristics of PCOS patients before and after treatment
Compared with the control group, the total ovarian area, ovarian medullary area, PSV, and EDV were significantly higher in the PCOS group, whereas the RI was significantly lower (P < .05; Table 1). After gynecological treatment and the resumption of normal ovulation, these changes were reversed within the PCOS group: total ovarian area, ovarian medullary area, PSV, and EDV decreased significantly, and RI increased significantly compared with the corresponding pretreatment values (P < .05; Table 2).
3.3. ROC curve analysis of transvaginal ultrasonography parameters for diagnosis of PCOS
Table 3 details the true positives, true negatives, false positives, false negatives, sensitivity, specificity, positive predictive value, and negative predictive value for total ovarian area, medullary area, PSV, EDV, and RI.
The study included 100 women with PCOS and investigated the sensitivity and accuracy of total ovarian area, medullary ovarian area, PSV, EDV, and RI in diagnosing PCOS in PCOS patients by ROC curve analysis. Consistent with the expected results, total ovarian area and medullary ovarian area were of high value in the diagnosis of PCOS. EDV and RI are more sensitive to the diagnostic value in PCOS patients than the PSV.
See Figure 2. Since transvaginal ultrasound parameters were used for PCOS classification, potential incorporation bias exists.
4. Discussion
PCOS is a chronic endocrine disorder, the main manifestations of which include menstrual disorders, high androgen levels, polycystic ovarian changes, etc. PCOS patients have many complications such as infertility, obesity, acne, cardiovascular disease, and type 2 diabetes. Most patients only exhibit a few of these pathological features, and clinical heterogeneity is the main reason for the lack of unified diagnostic criteria for PCOS.
So far, 3 sets of diagnostic criteria have been proposed by international consensus groups, namely, the National Institutes of Health criteria from the National Institutes of Health, the Rotterdam criteria proposed by the ESHRE and the ASRM, and the Androgen Excess Society criteria proposed by the American Androgen Excess Society. Among them, the Rotterdam criteria include polycystic ovary morphology, defined as the presence of 12 or more follicles measuring 2 to 9 mm in diameter in at least 1 ovary and/or an ovarian volume > 10 mL.[13,14] Studies have suggested expanding the follicle count to 20 or more follicles of any size, but there is no international consensus to revise this standard.[15–18] Figure 3 shows the typical appearance of PCOS on a 2-dimensional ultrasound.
Emekci Ozay et al‘s study pointed out the correlation between ovarian stromal blood flow and stromal thickness and PCOS.[19] Teede et al have shown that many studies and diagnostic criteria for PCOS are based on weak evidence,[20] so it is necessary to improve their accuracy. Considering the limitations of transabdominal ultrasound in displaying medullary and cortical blood flow characteristics, our study aimed to explore the feasibility of transvaginal 2-dimensional ultrasound in improving the diagnostic accuracy of PCOS.
The ultrasonic effects on PCOS clearly demonstrate the intrinsic characteristics of polycystic ovaries, confirming an increase in the number of follicles and an increase in the total ovarian area. In addition, several studies have shown that PCOS also causes changes in patients’ vascularization parameters, and the ovarian stromal blood flow index, vascularization index, and vascularization blood flow index of women with PCOS are significantly higher than those of normal reproductive women.[21–23]
Our findings regarding the increased PSV and EDV, along with decreased RI in PCOS patients, align with the results of Emekci Ozay et al,[19] further supporting the hypothesis of increased stromal vascularization in PCOS. Notably, these hemodynamic changes were reversed after the resumption of normal ovulation in our cohort, which suggests that the altered ovarian stromal perfusion in PCOS is a functional and potentially reversible feature rather than a fixed structural change.
Our study showed that when measured using transvaginal ultrasound, patients with PCOS had significantly higher PSV as well as EDV, while RI was smaller than in the control group. Consistent with the expected results, the trend in these measures was significantly reversed after treatment. The offline area analysis of the ROC curve showed that the total ovarian area and ovarian medullary area captured by transvaginal color Doppler ultrasound were of high diagnostic value.
In addition, RI showed high sensitivity and specificity, whereas EDV and particularly PSV were less discriminative.
In the present study, the diagnostic performance of each ultrasound parameter was assessed individually. The 2-dimensional parameters showed the highest discriminative ability (total ovarian area, AUC = 0.999; ovarian medullary area, AUC = 1.000), followed by RI (AUC = 0.997), EDV (AUC = 0.877), and PSV (AUC = 0.631). As each parameter was assessed individually, the present data do not establish that the addition of Doppler indices to 2-dimensional morphological assessment improves diagnostic accuracy.
Instead, the Doppler indices provide additional hemodynamic information on ovarian stromal perfusion, which may help to characterize the functional state of the ovary and to monitor the response to treatment. Whether the formal integration of morphological and hemodynamic parameters into a multivariable model confers incremental diagnostic value remains to be determined in future studies.
It is noteworthy that the application of artificial neural networks (ANNs) in medical image analysis offers a novel approach for establishing multiparameter diagnostic models. According to a systematic review by Hafizovic et al, AI can significantly improve the sensitivity and specificity of PCOS diagnosis by automatically quantifying imaging features such as ovarian volume and blood flow distribution.[24] Similarly, Badnjevic team validated the diagnostic advantages of multiparameter integrated analysis for complex syndromes in a study on the classification of chronic obstructive diseases, suggesting that combinations of hemodynamic indicators in this study could achieve more precise threshold optimization through ANNs.[25] Furthermore, Hrvat predictive model for medical equipment performance based on ANNs indicates that combining ultrasound equipment performance parameters with biomarkers can enhance the robustness of diagnostic systems, providing valuable insights for dynamic monitoring of PCOS.[26] However, there are still limitations in applying existing AI models to ovarian microvascular imaging, as the deep learning methods employed by Veljovic et al in drug design have not been widely used in the field of ultrasound image feature extraction.[27] Future research could explore the automatic identification of ovarian stromal microvascular patterns using convolutional neural networks, combined with hemodynamic parameters, to construct an intelligent diagnostic system, thereby advancing early screening and personalized treatment evaluation for PCOS.
This study has several limitations. First, as a single-center case-control study, the sample size is relatively small, which may limit the generalizability of the findings. Multicenter studies with larger cohorts are needed to validate these results.
Second, ultrasound examinations, particularly Doppler assessments, are highly operator-dependent. Although all scans in this study were performed by experienced sonographers using standardized settings, inter-observer variability cannot be entirely ruled out. Third, the PCOS group represented a clinically selected population in which most patients presented with prominent clinical features such as infertility, obesity, hirsutism, acne, and overt menstrual disturbance, and was compared with healthy volunteers.
Such selection is likely to have increased the degree of separation between the 2 groups and may therefore have inflated the estimates of diagnostic performance, so that the very high AUC values obtained here should be interpreted with caution and may not be generalizable to unselected populations or to women with mild or early disease. Fourth, because ultrasound parameters including ovarian morphological features were incorporated into the diagnostic workup for PCOS, a potential incorporation bias (also referred to as work-flow bias) exists. That means the diagnostic performance of these ultrasound parameters, particularly the 2-dimensional morphological parameters (total ovarian area and ovarian medullary area), may be overestimated, and the near-perfect AUC values should not be interpreted as independent diagnostic validation.
These findings are most applicable to discrimination within this highly selected PCOS-versus-healthy-control cohort. Fifth, no combined multivariable diagnostic model was constructed and no formal statistical comparison of AUCs was carried out, so the incremental diagnostic value of the Doppler indices relative to the 2-dimensional parameters could not be quantified in this study.
5. Conclusion
In conclusion, transvaginal color Doppler ultrasound is a valuable method for the ultrasonographic assessment of PCOS. In this highly selected case-control cohort, the 2-dimensional morphological parameters (total ovarian area and ovarian medullary area) showed the highest discriminative performance, whereas the Doppler hemodynamic indices (PSV, EDV, and RI) provided additional information on ovarian stromal perfusion. However, due to the selected nature of the study population and the potential for incorporation bias, these findings should be interpreted as reflecting discrimination within this specific PCOS-versus-healthy-control cohort and require confirmation in larger, multicenter, and unselected clinical populations before broader application.
Author contributions
Conceptualization: Yanling Bu, Wenlong Yu, Wei Zhao, Xiao Liang, Fengyu Na, Jingjing Wu.