Can an Ultrasound Miss Endometriosis? What a Normal Scan Does and Doesn’t Rule Out

Endometriosis Surgical Specialists International — Reviewed September 2026

In one paragraph: A normal workup and an absence of disease are not the same finding, and they are routinely reported to you as if they were. The phenotype of endometriosis most strongly associated with miscarriage in the published data is also the phenotype that scans are least able to find or rule out. This post is about what imaging can and cannot see, what “unexplained” actually means after two or more losses, and what a proper evaluation would do differently.

“Normal” Is a Statement About the Tests, Not About You

You have had two or more losses. Karyotypes, thyroid, antiphospholipid antibodies, a saline sonogram or hysteroscopy, maybe a thrombophilia panel ordered without much conviction. All normal. You were told you have unexplained recurrent pregnancy loss, and offered another cycle.

Here is the distinction nobody made for you.

“We did not find anything” is not the same sentence as “there is nothing there.” The first describes the performance of a set of tests. The second describes your body. Turning one into the other is a translation error, and it happens constantly — not through carelessness, but because “unexplained” is the only code available when the standard panel comes back clean.

A recurrent-loss workup finds the things it is built to find. It is not built to find superficial peritoneal endometriosis. Nothing in it is.

The Finding That Should Have Changed the Conversation

If endometriosis mattered to pregnancy loss only when severe, a clear ultrasound would be reassuring. That is not what the stage-by-stage data show.

The study that looked at this directly compared 143 women with surgically confirmed endometriosis against 143 controls, across 508 prior pregnancies. The miscarriage rates:

• rASRM stage I/II (minimal to mild): 42.1 in 100

• rASRM stage III/IV (moderate to severe): 30.8 in 100

• Controls: 22.0 in 100

Those are absolute rates, not relative increases. And the gradient runs the wrong way. The mildest disease carried the highest loss rate.

One qualifier belongs with those figures and is almost always left off: the excess was confined to subfertile women. Among fertile women in the same study the rates were 24.5% versus 21.5% — not significantly different. Whatever is driving the gradient, it shows up in women who were already having difficulty conceiving.

There are two honest explanations, and nobody has separated them.

The first is biological. Superficial peritoneal lesions are inflammatory, active and scattered across the pelvic lining. A large endometrioma is, in a sense, walled off. It is plausible that widely distributed active disease does more to the early pregnancy environment than one contained cyst. Reduced implantation specifically in mild disease appears in the pooled data as well.

The second is ascertainment bias, and it is the more deflating of the two. Mild disease is disproportionately found by accident — during surgery performed because a woman was struggling to conceive or had already lost pregnancies. So the stage I/II cohorts are loaded with women already in trouble, while the stage III/IV cohorts include women found because of a cyst or pain, with no fertility history at all. That alone could generate the entire gradient.

We will not resolve this for you, because it has not been resolved. Both may be partly right.

What follows clinically does not depend on which one wins: the absence of severe disease on a scan tells you very little about your risk.

Why the Imaging Did Not Find It

Transvaginal ultrasound in expert hands is genuinely good at deep, infiltrating disease. In the international consensus statement on non-invasive imaging, sensitivity reaches around 89% for rectal involvement.

Then it drops.

• Uterosacral ligaments: about 60% sensitivity on ultrasound (95% specificity). MRI does better here, at about 81% sensitivity (83% specificity).

• Rectovaginal septum: about 57%.

• Bladder: about 55%.

• Parametrium: the weakest compartment for both ultrasound and MRI.

These are referral-centre numbers, produced by operators specifically trained in this. A routine pelvic ultrasound to check your ovaries and lining is not the same examination. And 60% sensitivity means that even in a trained centre, roughly four in ten women with uterosacral disease are missed on ultrasound. If that compartment is the question, it is reasonable to ask whether MRI would add anything.

Then the category that matters most here.

A normal scan does not exclude endometriosis, and this is most true of superficial peritoneal disease: no imaging method has an established sensitivity for it. The international consensus statement on non-invasive imaging covers deep endometriosis only; superficial disease is explicitly outside its scope. Operator skill improves what a scan finds in deep disease. It has not been shown to close this gap. Emerging biomarker and AI-assisted tools show promising preliminary performance but lack external validation and are not ready for clinical use.

That is the same phenotype that carried the 42.1% loss rate.

You Do Not Need Surgery to Be Believed — and a Clear Scan Still Excludes Nothing

ACOG’s 2026 clinical practice guideline recommends transvaginal ultrasound as the first imaging test and no longer requires surgical confirmation — diagnosis may be made, and treatment begun, on symptoms, history and examination. NICE has moved in the same direction on imaging, recommending early transvaginal ultrasound in everyone with suspected endometriosis even when the pelvic examination is normal; it does not go on to endorse diagnosis without surgical confirmation. ESHRE says it plainly: use imaging in the workup, but a negative finding does not exclude endometriosis.

Hold both halves of that. The barrier to being taken seriously has come down — you should not have to consent to an operation in order to have your history believed.

But the shift cuts the other way too, and this is the part that gets lost. If a positive scan is now enough to diagnose, it becomes very easy for a negative scan to be treated as enough to exclude. It is not. A normal scan means the scan was normal. Nothing more was established that day.

The Symptoms That Get Normalised

Almost nobody with endometriosis has no symptoms. Far more often, the symptoms were never counted as symptoms — by you or by anyone examining you.

We do not think of this as a lesion in isolation. We have argued that endometriosis and recurrent pregnancy loss may frequently be two expressions of one underlying dysfunction rather than two diseases sharing a patient — a question we posed in Fertility and Sterility in 2023.

A separate idea of ours explains why the symptoms get scattered. In some patients the illness predates, exceeds and outlasts the pelvic disease eventually found, and we have proposed a name for that pattern: Pelvic Overlap Spectrum Syndrome (POSS) — endometriosis or adenomyosis, confirmed or strongly suspected, plus features in other systems such as gut motility, bladder and pelvic floor, nerve sensitisation, autonomic, vascular, immune, connective-tissue and mood, typically spread across many specialists. Reproductive outcomes are recorded but deliberately do not count toward the definition. It is a proposal with provisional research criteria, not a validated diagnosis, and it describes symptoms that occur together without claiming they share one biology. One practical consequence is that surgery is aimed at structural disease and is not offered as treatment of the whole illness.

Seen that way, “no symptoms” usually means one of these was normalised:

• Spotting for two or three days before your period properly starts. You were told this is normal. It is common. Different words.

• Pain with ovulation, mid-cycle, every month, on one side or alternating.

• Bowel symptoms that track your cycle — looser stools or worse cramping around menstruation, often already labelled irritable bowel syndrome.

• Bladder urgency or frequency without infection.

• Deep pain with intercourse, worse in certain positions, which you adapted around rather than reported.

• Fatigue that is not explained by sleep.

• A pelvic examination that hurt more than expected, and was not documented as a finding.

• Ovaries that were hard to see on a previous scan, or described as “stuck,” immobile, or requiring abdominal pressure to visualise.

That last one is a physical sign, not a symptom. It is frequently written into a report and then ignored. Reduced ovarian mobility is information.

None of these prove you have endometriosis. All of them are reasons to look properly — and each is routinely dismissed, one at a time, by a different clinician.

The Delay Is Structural, and It Has Not Improved

An umbrella review published in 2026, covering 52 systematic reviews and more than six million participants, found the median time from symptom onset to diagnosis ranging from 0.3 to 12 years, with the longest delays in Europe at six to ten years. NICE cites eight to ten years.

The important sentence in that review is the one about the trend: this delay has not meaningfully decreased in twenty years. The authors attribute it to structural barriers — patient delay, clinical delay, a shortage of specialised centres — rather than to any gap in technology.

What a Proper Evaluation Would Actually Do

Some of this may already have been done. Ask for the specifics, not the summary.

• Start with the tissue, if there is any: The 2026 ASRM committee opinion on recurrent pregnancy loss names chromosomal analysis of miscarriage tissue by array-based technology as the first evaluation step. It also defines recurrent loss as two or more losses, counts biochemical (hCG-confirmed) losses, and states the losses need not be consecutive. If you were told you did not qualify for a workup, check that definition against what you were told.

• Have someone look at the junctional zone: The inner layer of the myometrium behaves abnormally in endometriosis and adenomyosis, and junctional-zone thickening has been reported in up to 90% of endometriosis patients examined — though whether that abnormality precedes the disease or follows it is unproven. ESHRE’s recurrent-loss guideline now states that all women with recurrent loss could have 2D ultrasound to rule out adenomyosis. That is the only guideline-level endorsement of looking, and it is frequently not done.

Ask These Questions, in These Words

• Was my ultrasound performed by someone specifically trained in deep endometriosis assessment, using standardised terminology?

• Were the uterosacral ligaments, the pouch of Douglas and the parametrium specifically assessed and documented?

• Were my ovaries mobile on examination?

• Was my junctional zone measured?

• Given that no scan can reliably rule out superficial disease, what would change your assessment?

And know what this does not buy you. Finding endometriosis would not establish that it explains your losses. ESHRE’s endometriosis guideline goes only as far as asking clinicians to “be aware that there may be an increased risk of first trimester miscarriage and ectopic pregnancy” — a Strong recommendation with the hedge written into it. And no trial has ever shown that removing it reduces miscarriage — the Cochrane review reports no live-birth data at all, ASRM’s 2026 opinion lists treatment of endometriosis and adenomyosis among interventions without evidence of benefit, and the association itself is weak or absent in assisted reproduction cohorts. Does endometriosis cause miscarriage? A surgeon’s reading of the evidence sets all of that out in full.

What it does buy you is an accurate description of your own pelvis. That is not a small thing when the alternative is another year of “unexplained.”

How ESSI Approaches This

At ESSI the scan is not outsourced. Dr. Vidali performs his own transvaginal ultrasound, Doppler studies and pelvic mapping. The person reading the images is the person who would operate, so what the scan shows and what the surgery would find are held in one head.

What that looks like in practice:

• A compartment-by-compartment map: We look at the ovaries and their mobility (“stuck” ovaries), the rectum and rectovaginal septum, the uterosacral ligaments, the bladder, and the uterine wall and junctional zone for adenomyosis. The standardised IDEA approach guides it.

• MRI where it adds something: Particularly for uterosacral ligament and parametrial disease, where MRI outperforms ultrasound.

• An honest statement of what the scan cannot see: Superficial peritoneal disease has no reliable imaging test. If your symptoms point there and the scan is clear, we tell you so, rather than calling you normal.

• A decision that follows the map, not the other way round: Sometimes the map points to surgery. Sometimes it points to medical treatment, pelvic-floor therapy, or protecting your eggs first. The first time you meet your surgeon should never be the day of your surgery.

What a First Consultation Involves

Not a surgical booking. A diagnostic conversation — and specifically, a second opinion on the word “unexplained”.

We read what you already have: every loss, every result, every operative report and every scan, including the ones that were called normal. Where indicated, detailed pelvic ultrasound with attention to the junctional zone, ovarian mobility and deep disease, performed by the surgeon rather than delegated. Then a plain statement of what we think is going on and what we still cannot see.

Keep Reading

• Does endometriosis cause miscarriage? A surgeon’s reading of the evidence — the full evidence review this post sits under.

• Miscarriage after a euploid embryo: what to check next — what PGT-A tests, what it does not, and the question to ask about your miscarriage tissue.

• Surgery before your next transfer: questions to ask before you consent — the case against reflexive operating, told by a surgeon.

References

1. Kohl Schwartz AS, Wölfler MM, Mitter V, et al. Endometriosis, especially mild disease: a risk factor for miscarriages. Fertil Steril. 2017;108(5):806–814.e2.

2. Horton J, Sterrenburg M, Lane S, Maheshwari A, Li TC, Cheong Y. Reproductive, obstetric and perinatal outcomes in women with adenomyosis and endometriosis. Hum Reprod Update. 2019;25(5):593–633.

3. Condous G, Gerges B, Thomassin-Naggara I, et al. International consensus statement on non-invasive imaging for pelvic deep endometriosis and endometriosis classification systems. J Minim Invasive Gynecol. 2024;31(7):557–573.

4. Nurmakova A, et al. J Clin Med. 2026;15(12):4583.

5. ACOG Committee on Clinical Practice Guidelines–Gynecology. Diagnosis of Endometriosis. Clinical Practice Guideline No. 11. Obstet Gynecol. 2026;147(3):432–448.

6. National Institute for Health and Care Excellence. Endometriosis: diagnosis and management. NG73. Last reviewed September 2025.

7. ESHRE Endometriosis Guideline Group (Becker CM, Bokor A, Heikinheimo O, et al.). Hum Reprod Open. 2022;2022(2):hoac009.

8. Vidali A, Riccio LGC, Abrao MS. Endometriosis and recurrent pregnancy loss: two manifestations of the same underlying dysfunction? Fertil Steril. 2023;119(5):836–837.

9. ASRM Practice Committee. Recurrent pregnancy loss: a committee opinion. Fertil Steril. 2026;125:1023–1041.

10. Vercellini P, Conca B, Salmeri N, et al. Arch Gynecol Obstet. 2026;313:62.

11. ESHRE Guideline Group on Recurrent Pregnancy Loss. Hum Reprod Open. 2023;2023(1):hoad002.

12. Bafort C, Beebeejaun Y, Tomassetti C, Bosteels J, Duffy JMN. Laparoscopic surgery for endometriosis. Cochrane Database Syst Rev. 2020;10:CD011031.pub3.

13. Wei Y, Xiao X, Wu X, Xie C, Li T. Front Med. 2026;13:1630529.

Endometriosis Surgical Specialists International (ESSI)