Can an Ultrasound Detect Ovarian Cancer?

Ultrasound can detect ovarian cancer, and it is the first-line imaging tool used to evaluate suspicious masses on or near the ovaries. But “detect” covers a wide range of performance depending on the type of ultrasound, the scoring system applied, the experience of the person holding the probe, and whether the goal is diagnosing a known mass or screening someone with no symptoms. In research settings, modern ultrasound-based classification systems pick up malignant masses with sensitivity above 90%, yet the same technology has failed to reduce deaths when used to screen the general population. That gap between diagnostic power and screening effectiveness is the most important thing to understand about ultrasound and ovarian cancer.

What Ultrasound Actually Shows

Ultrasound uses sound waves to create images of the ovaries and surrounding structures. A transvaginal probe, placed inside the vagina, gets closer to the ovaries and produces sharper images than an abdominal scan. What the examiner sees is the shape, size, and internal structure of any mass: whether it is filled with fluid, solid, or a mix of both; whether it has smooth walls or irregular projections; and whether blood flow patterns suggest abnormal growth. These features help distinguish a harmless cyst from something that needs further investigation.

A simple fluid-filled cyst with thin, smooth walls is almost always benign. Warning signs that raise suspicion include irregular solid areas, thick internal walls or septa, papillary projections growing into the cyst cavity, and the presence of free fluid in the pelvis. None of these features alone proves cancer, but the more of them that appear together, the higher the risk. The challenge is that some benign conditions mimic these features, and some early cancers look deceptively simple.

How Doctors Score What They See

Radiologists and gynecologists do not just eyeball an ultrasound and guess. Several structured scoring systems exist to standardize how ovarian masses are classified, and the choice of system affects accuracy considerably.

The most widely adopted framework in the United States is the O-RADS (Ovarian-Adnexal Reporting and Data System), published by the American College of Radiology. It assigns masses to one of six risk categories, from O-RADS 0 (incomplete study) through O-RADS 5 (high risk of malignancy), with a management recommendation attached to each level.¹ Multiple validation studies have confirmed that O-RADS categories correlate well with actual malignancy rates, with most research using O-RADS 4 as the optimal cutoff for flagging cancer.² A large meta-analysis pooling data from 15 studies found that O-RADS ultrasound achieved a pooled sensitivity of 94% and specificity of 81%.³

A separate system developed by the International Ovarian Tumor Analysis (IOTA) group uses a set of ten “simple rules,” five suggesting malignancy and five suggesting a benign mass. Features flagging malignancy include an irregular solid tumor, fluid in the abdomen, four or more papillary structures, a large irregular multilocular-solid mass, and very high blood flow on Doppler. Features pointing to a benign mass include a simple unilocular cyst, small solid components, acoustic shadows, a smooth multilocular tumor under 100 mm, and no detectable blood flow. When prospectively tested, these rules applied to about three-quarters of tumors and achieved a sensitivity of 95% with a specificity of 91%.⁴

The IOTA group also developed a more sophisticated model called ADNEX, which uses a mathematical algorithm rather than simple yes-or-no rules. In a prospective study, ADNEX achieved higher sensitivity than the simple rules (80% versus about 67%) and slightly better specificity (94% versus 91%). Combining the ADNEX model with a blood marker called CA-125 pushed overall diagnostic accuracy to 91%.⁵

It is worth noting that performance varies across studies. One study evaluating O-RADS found a sensitivity of only 52% and specificity of 84%, substantially lower than the pooled estimates from larger meta-analyses.⁶ Results like these are a reminder that the impressive pooled numbers from meta-analyses represent averages across many centers, and that individual performance depends heavily on local factors.

What Color Doppler and Contrast Add

Standard grayscale ultrasound shows anatomy. Adding color Doppler shows blood flow inside and around a mass, which matters because cancerous tumors tend to recruit new, chaotic blood vessels. Combining grayscale appearance with Doppler flow patterns improves the ability to tell malignant from benign masses. One study found that this combination yielded a sensitivity of about 81% and specificity of about 94%.⁷ In malignant tumors, blood-flow resistance tends to be lower, reflecting the leaky, poorly formed vessels that cancers build; research has confirmed that virtually all malignant ovarian tumors show low resistance on Doppler, while most benign ones do not.⁸

A newer technique called contrast-enhanced ultrasound (CEUS) involves injecting a small amount of microbubble contrast into a vein, which lights up the blood supply of a mass in real time. A meta-analysis of seven studies found that CEUS achieved a pooled sensitivity of 93% and specificity of 91% for distinguishing malignant from benign ovarian tumors.³ Combining CEUS with the O-RADS scoring system pushes sensitivity to roughly 95% and specificity to about 86%, outperforming either method alone.⁹ CEUS is not yet standard in every clinic, but it is becoming more available and is particularly useful for masses that look ambiguous on regular ultrasound.

Why Screening Healthy Women Has Not Worked

Given these accuracy numbers, you might expect that routinely screening all women with ultrasound would catch ovarian cancer early and save lives. It has not. This is one of the most important and counterintuitive findings in ovarian cancer research.

The U.S. Preventive Services Task Force (USPSTF) recommends against screening for ovarian cancer in women who are not at known high risk. Their 2018 recommendation was based on evidence that screening with transvaginal ultrasound, the blood marker CA-125, or both together does not reduce ovarian cancer deaths. A major trial with over a decade of follow-up found essentially no difference in ovarian cancer mortality between screened and unscreened groups.¹⁰

The UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS), one of the largest randomized trials ever conducted on this topic, enrolled over 200,000 women and followed them for 14 years. The ultrasound-only screening group showed an 11% mortality reduction that was not statistically significant, meaning the observed difference could easily have been due to chance. A multimodal group that used CA-125 blood tests followed by ultrasound fared slightly better with a 15% reduction, but that result also fell short of statistical significance.¹¹

The reasons screening fails are multiple. Ovarian cancer is relatively rare, so even a test with 90% sensitivity will generate many more false alarms than true catches in a general population. When used as the sole screening test, transvaginal ultrasound is sensitive but has a low positive predictive value, meaning most women flagged by screening turn out not to have cancer.¹² Those false positives lead to unnecessary surgeries, each carrying its own risks. And unlike cervical or colon cancer, ovarian cancer does not have a long, reliably detectable precancerous stage; it can spread quickly between annual screens.

When Ultrasound Is Used and When It Works Best

The distinction between screening and diagnosis matters enormously. Ultrasound performs well when a doctor has a reason to look: pelvic pain, bloating, a mass felt during an exam, abnormal bleeding, or elevated tumor markers in a blood test. In these situations the prior probability of finding something meaningful is higher, which dramatically changes how useful the test is. A 94% sensitivity rate means something very different when one in five patients in the room actually has cancer compared with when one in 2,500 does.

For women already known to have an ovarian mass, the real question ultrasound answers is not “Is there something here?” but “How worried should we be about what is here?” The scoring systems described above were designed precisely for this purpose: helping decide whether a mass can be safely monitored, referred to a general gynecologist, or sent urgently to a gynecologic oncologist for surgery. Getting that triage right has real consequences for outcomes, because women whose ovarian cancer is operated on by a subspecialist gynecologic oncologist tend to do better than those treated by general surgeons.

The Experience of the Person Doing the Scan

Ultrasound is more operator-dependent than almost any other imaging modality. Unlike a CT or MRI, where the machine produces a standardized set of images that any radiologist can review, ultrasound image quality depends on who is holding the probe, what angles they choose, and how they interpret what they see in real time.

Research has quantified this gap. A study comparing ultrasound examiners with varying levels of training found that the most experienced investigator achieved 92% accuracy in classifying ovarian masses, while less experienced observers ranged from 82% to 87%. Agreement between two highly experienced examiners was excellent, but agreement among less experienced observers was only moderate.¹³ Another study looking at specific histological diagnoses found that even Level II examiners (trained but not expert) achieved good sensitivity for certain tumor types, including about 82% for the most common type of ovarian cancer (serous cystadenocarcinoma), but overall agreement with pathology was only moderate.¹⁴

The practical impact of operator skill extends beyond accuracy numbers. One analysis found that when ultrasound was performed by sonographers with more than ten years of experience, only about 22% of patients underwent major surgical staging procedures, compared with 37% when scans were done by less experienced technologists. The experienced group correctly identified the nature of abnormalities in nearly all cases, while the less experienced group did so in only about half.¹⁵ In other words, a skilled operator does not just read the image better; they spare patients unnecessary operations.

Why Seeing the Ovaries Gets Harder with Age

Ovarian cancer is most commonly diagnosed in women over 60, which creates an unfortunate irony: the patients who most need the test are often the hardest to scan. After menopause, ovaries shrink and lose their follicular activity, making them harder to distinguish from surrounding tissue on ultrasound. In the large UKCTOCS screening trial, researchers explicitly noted that ovarian visualization is difficult in older postmenopausal women because of decreasing ovarian size.¹⁶

A study specifically examining how often ovaries could be seen found that after age 50, the probability of visualizing both ovaries dropped steadily. By age 85, about a third of women had neither ovary visualized on ultrasound. Still, one or both ovaries could be seen in roughly two out of three women even past age 80.¹⁷ Body habitus also plays a role: higher body weight can push the ovaries further from the probe and degrade image quality. This is another area where operator experience makes a difference, as a skilled sonographer knows tricks for improving visualization, like having the patient change position or applying gentle pressure with the probe.

Combining Ultrasound with Blood Markers

Ultrasound rarely works in isolation for evaluating ovarian masses. Doctors frequently order blood tests for tumor markers alongside imaging, and combining the two can improve accuracy beyond what either achieves alone.

CA-125 is the most commonly used blood marker. It is elevated in many ovarian cancers, but also rises in endometriosis, pelvic infections, liver disease, and even normal menstruation, which limits its usefulness as a standalone test. A newer marker called HE4 (human epididymis protein 4) tends to be more specific. Recent research found that combining O-RADS ultrasound with HE4 produced a particularly strong combination, achieving 95% sensitivity and 92.5% specificity for distinguishing malignant from benign tumors.¹⁸

In premenopausal women, the added cost of blood markers on top of ultrasound-based assessment may not provide enough diagnostic benefit to justify routine use. A cost-effectiveness study found that adding biomarker panels to expert ultrasound assessment using the IOTA simple rules offered no diagnostic advantage over expert ultrasound alone, while roughly doubling the per-patient cost.¹⁹ The practical takeaway: blood markers are most useful in postmenopausal women or when the ultrasound findings are ambiguous, not as a blanket add-on for every scan.

What Happens When Ultrasound Cannot Give a Clear Answer

For masses that remain indeterminate after ultrasound, even with Doppler and scoring systems, pelvic MRI is the preferred next step. MRI excels at characterizing soft tissue and can often resolve what ultrasound cannot, particularly for complex masses with both solid and cystic components. A systematic review confirmed pelvic MRI’s role as the preferred advanced second imaging test for indeterminate adnexal masses, helping assign patients to the appropriate level of surgical care.²⁰ CT scans, while useful for staging cancer that has already been diagnosed (checking whether it has spread to lymph nodes or distant organs), are less helpful than MRI for characterizing the mass itself.

The referral pathway matters. A mass that an experienced ultrasound examiner calls clearly benign can usually be monitored with repeat imaging. A mass scored as high-risk on O-RADS or flagged by IOTA rules typically goes straight to a gynecologic oncologist. It is the middle ground, the indeterminate masses, where the pathway becomes most critical. Sending every ambiguous mass directly to surgery leads to unnecessary operations; watching too many of them risks delaying cancer diagnosis. MRI fills that gap by providing a second, independent assessment before a surgical decision is made.

High-Risk Women and Surveillance

Women who carry BRCA1 or BRCA2 gene mutations face a dramatically higher lifetime risk of ovarian cancer. You might assume that intensive surveillance with regular ultrasound and CA-125 testing would catch cancers early in these women, but the evidence is discouraging. A study of annual screening in women at increased familial risk found that surveillance with transvaginal ultrasound and CA-125 was ineffective at detecting tumors early enough to meaningfully improve survival in BRCA1/2 carriers.²¹ BRCA-related ovarian cancers tend to be high-grade serous tumors that grow and spread rapidly, often becoming advanced between annual scans. For this reason, current guidelines generally recommend risk-reducing surgery (removal of the ovaries and fallopian tubes) rather than surveillance for BRCA carriers who have completed childbearing.

Artificial Intelligence and Where Things Are Heading

AI tools trained to analyze ultrasound images of ovarian masses are an active area of development. Early results are promising, with representative studies reporting accuracies between 85% and 96%. However, most of this research relies on data from single centers and has not been validated across different hospitals, patient populations, and equipment.²² The appeal is obvious: if an algorithm could reliably classify masses from ultrasound images, it would reduce the dependence on operator expertise and potentially bring expert-level accuracy to smaller clinics. But the gap between a promising retrospective study and a tool that works reliably in day-to-day clinical practice is large, and no AI system has yet replaced human assessment for ovarian mass evaluation in routine care.

Ultrasound in Younger Patients

Ovarian cancer is rare in children and adolescents, but ovarian masses are not. In younger patients, ultrasound is performed transabdominally (through the abdominal wall) rather than transvaginally. A review of over 1,300 ultrasound examinations in pediatric and adolescent patients found that both ovaries could be seen in about 73% of premenarchal girls and roughly 80% of postmenarchal adolescents using a transabdominal approach. Ovarian cysts were identified on initial scans in about 10% of patients, with the vast majority being simple unilocular cysts.²³ Most ovarian masses in this age group turn out to be functional cysts or benign tumors like mature teratomas (also called dermoid cysts), which resolve on their own or are easily treated. The scoring systems designed for adult women have not been specifically validated in pediatric populations, so assessment in younger patients relies more heavily on the examiner’s experience and the mass’s basic imaging characteristics.