What Ovarian Cancer Looks Like on Ultrasound Images

Ovarian cancer on ultrasound typically appears as a complex mass containing both solid and cystic components, often with irregular internal walls, thick septations, and abnormal blood flow detected on Doppler imaging. A purely fluid-filled, thin-walled cyst is almost always benign, while a mass that mixes solid tissue with fluid-filled spaces and shows rich blood supply raises serious concern. The specifics of what sonographers look for, and what separates a worrisome image from a reassuring one, involve a surprisingly detailed visual vocabulary that has been formalized into scoring systems used worldwide.

The Features That Raise a Red Flag

When a sonographer evaluates an ovarian mass, they are looking at a cluster of features rather than a single finding. The broad categories include the internal architecture of the mass, its outer walls, and any growths projecting into fluid-filled spaces. A benign cyst tends to be smooth, thin-walled, and filled with uniform fluid that appears dark on ultrasound. Cancer, by contrast, tends to produce masses that are “complex,” meaning they have a mix of solid tissue (which appears brighter or gray) and fluid-filled spaces (which appear dark).

One of the most telling features is the presence of papillary projections, which are finger-like or nodular growths of solid tissue that extend from the wall of a cyst into the fluid inside it. In a study comparing benign, borderline, and malignant ovarian tumors, malignant papillary projections averaged about 35 mm in size compared to roughly 10 mm in benign ones. Malignant projections also had an irregular surface about 88% of the time, versus 23% for benign projections, and they tended to attach to the cyst wall at a wide, obtuse angle rather than the narrow base seen in benign growths.¹ Those distinctions are subtle enough that they require an experienced eye, but they are among the strongest visual clues available on a standard ultrasound.

Other features associated with malignancy include thick internal septations (the walls dividing the mass into compartments), solid areas within a predominantly cystic mass, and an overall irregular shape. The more solid tissue a mass contains relative to fluid, the more suspicious it becomes. A mass that is entirely solid is not automatically cancerous, but solid components with irregular borders push the risk estimate upward.

What Blood Flow Patterns Tell the Sonographer

Beyond the shape and architecture of a mass, Doppler ultrasound reveals how blood moves through it. Cancer needs blood vessels to grow, and the vessels it builds are structurally abnormal. They tend to be leaky, chaotic, and have low resistance to blood flow. On a color Doppler image, this shows up as prominent blood flow within the solid parts of a mass, and the flow has a distinctive pattern when measured.

Two numbers that sonographers use to characterize this flow are the resistance index (RI) and pulsatility index (PI). In benign tumors, these values tend to be higher, reflecting the normal resistance that healthy blood vessels offer. In malignant tumors, both numbers drop. One study found that about 88% of malignant tumors had a PI below 0.8, compared to fewer than 5% of benign tumors, and roughly 83% of malignant tumors had an RI below 0.6 versus about 7% of benign ones.² Meanwhile, nearly all malignant tumors showed detectable vascularity on color Doppler, compared to about two-thirds of benign ones.³

The location of blood flow matters too. Blood flow at the center of a mass, within solid components or along the base of papillary projections, is more concerning than flow only at the periphery. In the study of papillary projections mentioned earlier, every malignant projection measuring 10 mm or larger showed blood flow on color Doppler, while benign projections of the same size showed none.¹ A highly vascular solid component, in other words, adds substantially to the suspicion that a mass is malignant.

Standardized Scoring Systems

Because the interpretation of ovarian masses depends so heavily on the examiner’s experience, researchers have developed structured scoring systems that attempt to standardize how ultrasound findings are categorized and communicated. Two of the most widely used are the IOTA Simple Rules and the O-RADS classification.

The IOTA Simple Rules work by checking a mass against a short list of features associated with malignancy (called “M-rules”) and a short list associated with benign disease (called “B-rules”). If only M-rules apply, the mass is classified as likely malignant; if only B-rules apply, it is classified as likely benign; if both or neither apply, the result is inconclusive. In practice, the rules can be applied in the vast majority of cases and achieve strong accuracy. A prospective study found that simple rules could be applied in about 94% of tumors, reaching a sensitivity around 85% and specificity near 99%.⁴ Sensitivity was higher in postmenopausal women (about 92%) and specificity was higher in premenopausal women (about 99%), a pattern that makes intuitive sense since the types of masses encountered differ by age.

The O-RADS system, developed by the American College of Radiology, uses a scale from 0 to 5 that assigns a risk of malignancy to each mass. An O-RADS score of 1 means the finding is physiologically normal, while a score of 5 indicates a high risk of malignancy. Each score comes with a management recommendation, from no follow-up at the low end to surgical evaluation at the high end.⁵ In a U.S. validation study, using a cutoff of O-RADS 4 or higher to flag possible cancer yielded a sensitivity of about 91%, a specificity of about 82%, and notably high negative predictive value near 99%, meaning that masses scored below that threshold were very rarely malignant.⁶

Both systems perform similarly in head-to-head comparisons. A study comparing IOTA Simple Rules, O-RADS, and the blood marker CA-125 found sensitivities above 94% for both imaging-based systems and overall areas under the curve in the range of 0.80 to 0.83.⁷ The practical takeaway is that either system substantially reduces the risk of a subjective or ambiguous ultrasound report.

Benign Masses That Can Mimic Cancer

Not every complex or solid-looking ovarian mass is malignant, and one of the trickiest aspects of ovarian ultrasound is telling cancer apart from the benign masses that sometimes look alarming. Dermoid cysts, also called mature teratomas, are among the most common benign ovarian tumors in younger women. They can contain fat, hair, and even calcified material like teeth, which produces a distinctive ultrasound appearance: a bright echogenic focus called the “tip of the iceberg” sign, fat-fluid levels, or a disorganized internal pattern called “dermoid mesh.” When two or more of these classic features appear together, the diagnosis is usually straightforward.⁸ But dermoids that lack these typical features can look worryingly solid.

Fibromas and fibrothecomas are another source of confusion. These are benign solid tumors of the ovary that can appear as a uniformly solid mass, sometimes accompanied by fluid in the pelvis (ascites) and even elevated CA-125 blood levels, both of which are associated with cancer. When all of those features line up, the clinical picture can closely mimic advanced malignancy, leading to more aggressive surgery than the patient actually needs.⁹

Endometriomas deserve special mention. These are cysts formed by endometriosis, and their classic ultrasound appearance is a “ground-glass” pattern, a hazy, low-level internal echo that makes the cyst look like frosted glass. That appearance is reassuring. The danger arises when an ovarian clear cell carcinoma, an aggressive type of cancer, develops from endometriosis. In one study, the loss of the ground-glass pattern was an independent indicator that an endometriosis-related cyst had undergone malignant transformation, along with larger cyst size, older patient age, and the appearance of solid components.¹⁰ When a clear cell cancer does develop from endometriosis, ground-glass echogenicity of the cyst fluid is more common than in clear cell cancers that arise independently, which can make early detection even more challenging.¹¹

Looking Beyond the Ovary

Once a mass is identified as suspicious, ultrasound does not stop at the ovary itself. Sonographers also evaluate the surrounding abdomen and pelvis for signs that cancer may have already spread, which directly influences treatment planning and prognosis.

Ascites, the accumulation of fluid in the abdominal cavity, is one of the most readily detected signs of advanced disease. Ultrasound picks up ascites with high reliability, achieving pooled sensitivity around 95% and specificity around 91% in a meta-analysis of studies evaluating tumor spread.¹² The relationship between ascites and disease stage is striking. In early-stage ovarian cancer (stages I and II), ascites appears in only about 17% of cases, and the volume is small. In advanced-stage disease (stages III and IV), nearly 90% of cases involve ascites, and most patients have more than half a liter of fluid at surgery.¹³

Ultrasound can also identify omental metastases, deposits of cancer in the fatty apron of tissue that drapes over the intestines. Metastatic omentum appears on ultrasound as solid, immobile tissue or discrete nodules, and in one prospective study it was identified with an accuracy above 90% and a positive predictive value near 95%.¹⁴ For some sites, ultrasound actually outperforms CT scanning. In one comparison, detection rates for metastases in the pelvic area and on the bowel surface were significantly higher with ultrasound than with CT, though performance was similar for other locations like the diaphragm and liver surface.¹⁵

Transvaginal Versus Transabdominal Approach

Ovarian ultrasound is performed using two approaches, and each has its strengths. The transvaginal probe, inserted into the vagina, sits much closer to the ovaries and produces higher-resolution images of small structures. Comparative studies found that transvaginal image quality was superior in roughly 80-87% of scans.¹⁶ This matters for identifying fine details like small papillary projections, thin septations, and the internal architecture of complex cysts. In practice, transvaginal ultrasound is the primary tool used to characterize an ovarian mass and assign risk scores.

The transabdominal approach, using a probe on the lower abdomen with a full bladder, sacrifices detail for a wider field of view. It is better at surveying the entire pelvis and upper abdomen, making it useful for checking for ascites, large masses that extend beyond the pelvis, and omental or peritoneal disease. Most thorough evaluations use both approaches together, starting with the transabdominal scan for overall orientation and following with the transvaginal probe for detailed characterization of the mass itself.

Where Ultrasound Falls Short

For all its strengths in characterizing known masses, ultrasound has genuine limitations as a screening tool for ovarian cancer. Many of the most aggressive ovarian cancers are now believed to originate not in the ovary itself but in the fallopian tubes, specifically from cells on the fimbriae. These tiny structures are not reliably imaged by ultrasound, and by the time a tumor originating there becomes large enough to see, it may have already spread. Researchers have pointed out that only a fraction of metastatic tumors may reach a detectable size before they metastasize, meaning annual screening with ultrasound may miss a large share of early-stage cancers.¹⁷

Large screening trials have reflected this challenge. The U.S. PLCO trial, which randomized tens of thousands of women to ultrasound screening or usual care, found no reduction in ovarian cancer mortality among those screened.¹⁸ This does not mean ultrasound is useless; it means that finding ovarian cancer early through population-wide screening is a fundamentally harder problem than screening for cancers that grow more predictably or in more accessible locations. Ultrasound remains excellent at evaluating a mass once it is found, but the hope that routine screening scans would catch ovarian cancer at a curable stage has not been borne out by the evidence so far.

When MRI Comes Into the Picture

If ultrasound leaves a mass in the “indeterminate” category, MRI is the usual next step. A systematic review concluded that MRI with intravenous contrast provides the highest confidence in detecting ovarian cancer, but its chief advantage is actually its specificity, its ability to identify benign masses and spare patients unnecessary surgery.¹⁹ In one comparison, transvaginal ultrasound was more sensitive than MRI (92% versus 83%), but MRI was substantially more specific (84% versus 59%), giving it a better overall diagnostic accuracy.²⁰

In practice, this means ultrasound is better at catching cancer when it exists, while MRI is better at confirming that something suspicious is actually benign. The two modalities complement each other rather than compete. MRI also offers higher sensitivity for specific benign conditions like endometriotic cysts and hemorrhagic cysts, where its overall sensitivity was found to be about 95% compared to roughly 78% for transvaginal ultrasound.²¹ Practically, if your ultrasound shows something concerning, getting a second look with MRI can either escalate the urgency or provide reassurance that surgery is not needed.

Artificial Intelligence and the Future of Ovarian Ultrasound

One of the biggest limitations of ovarian ultrasound is that it depends heavily on the skill and experience of the person holding the probe. Expert examiners at high-volume centers often outperform the scoring systems described above, but most patients are not scanned by those experts. This is where artificial intelligence is starting to make a tangible difference.

A meta-analysis pooling data from over 15,000 ultrasound images found that AI models achieved an overall sensitivity of 81% and specificity of 92% for distinguishing malignant from benign ovarian masses.²² Those numbers are solid but represent an average across many different AI approaches. The most advanced models are doing better. A large international study using transformer-based neural networks, trained on over 17,000 images from 20 centers across eight countries, found that the AI significantly outperformed both expert and non-expert human examiners on every metric tested. In a simulated triage scenario, the AI-driven support reduced referrals to experts by 63% while actually improving diagnostic accuracy compared to current practice.²³

A separate Chinese multicenter study showed similar promise, with a deep learning model outperforming radiologists at detecting ovarian cancer in multiple validation datasets. When radiologists were given the AI’s output as a second opinion, their accuracy and sensitivity both increased significantly.²⁴ Newer systems are also attempting to automate the extraction of features used in scoring systems like O-RADS and IOTA, segmenting solid tissue, locules, and papillary projections directly from the image.²⁵

These tools are not yet a replacement for a trained clinician, but they are closing the gap between what expert centers achieve and what community hospitals can offer. For a disease where the quality of the initial ultrasound interpretation can determine whether a patient gets timely surgery or months of watchful waiting, that gap has real consequences.