Lora Liu, MD, on closing the gaps in endometriosis care

Lora Liu, MD, an endometriosis excision specialist based in New York, New York, discusses the evolving landscape of endometriosis diagnosis, surgical management, and emerging technologies in this in-depth interview. Despite significant advances in high-resolution ultrasound and MRI, Liu emphasizes that surgical visualization remains essential, as imaging is unlikely to reliably detect lesions smaller than 2 mm or capture the full extent of surrounding inflammation and fibrosis. Laparoscopy, she notes, allows surgeons not only to see, but also to feel disease—a dimension imaging cannot replicate.

Liu highlights how preoperative surgical mapping is transforming patient counseling and operating room (OR) preparation, enabling more realistic expectations around procedures such as bowel resection or ureteral reimplantation. She also addresses the growing role of artificial intelligence (AI)–assisted imaging, arguing that AI could standardize endometriosis detection across institutions and help surgeons anticipate operative complexity before entering the OR.

On the question of recurrence, Liu distinguishes between true biological recurrence and residual disease left behind by incomplete excision—a distinction she finds frequently overlooked in the literature. She identifies key gaps in the evidence base, including the lack of long-term comparative data and the difficulty of designing apples-to-apples studies given the disease's complexity.

Liu also discusses emerging surgical technologies, including robotics, indocyanine green (ICG) imaging for ureter, nerve and other structure visualization, and the future potential of augmented reality in the OR. She closes with a compelling vision for a tracer-based diagnostic tool—analogous to a PET scan—that could illuminate endometriosis lesions both preoperatively and intraoperatively.

For general ob-gyns, she offers practical referral guidance, encouraging early specialist involvement when imaging findings, failed medical management, or exam findings raise concern.

Contemporary OB/GYN:What clinical questions still require surgical visualization despite improvements in imaging and diagnostics?

Lora Liu, MD: We’ve come a long way with imaging. It’s very operator dependent, but ultrasound is remarkably advanced these days. We've made huge advances with detecting soft tissue disease and deep disease. However, you still need a surgical diagnosis first because there are some things that imaging misses. MRI takes slices throughout the entire body and they're at 2-4 mm, which is incredibly small, very thin slices of the body. However, a lot of times, endometriosis lesions are less than 2 mm, and so they're going to get missed on those slices. We see this a lot, especially with peritoneal disease or bowel lesions; sometimes they're just like tiny studs on the bowel, and they're clear, and they're maybe less than a millimeter, but it's all endo and it's going to be completely missed. Even the most skilled sonographer or the highest-resolution MRI will miss them. These lesions simply fall below the resolution of imaging.

Another thing is that sometimes you can see with ultrasound or MRI that there's a bowel nodule. You see the core of it, but what you may not see is the surrounding tissue, whether it's inflammation or fibrosis. Many times, you are unable to see all the collateral damage. You just see that little pebble of endo. With laparoscopy, you're able to see everything. You’re also able to touch the tissue—to feel it—and think, “Oh my gosh, this is rock hard. This is like rubber cement.” You hear the phrase “frozen pelvis.” It's because, literally, the organs are completely tethered. They're completely encased in scar tissue and inflammation. With ultrasound, you can interrogate tissues with the probe and it’s a dynamic exam. With MRI, you can't. So there's the element with surgery that you're actually able to not only see, but feel the disease, as well as the surrounding tissue. You're able to find normal tissue planes with surgery. Unless you are a very specialized ultrasonographer or radiologist, that is tough to see with imaging alone. Imaging has come a long way, but it will never replace the surgical diagnosis and treatment.

High-resolution ultrasound and MRI protocols are improving detection of deep, infiltrating disease. How are these tools changing preoperative planning for excision surgeons?

I think we're beginning to shift from “Let's just put a camera inside and see what we see.” Don't get me wrong, we're still doing a diagnostic laparoscopy because, again, imaging is not 100% accurate. So you still always have to take a look. But I think with appropriate preoperative surgical mapping, you are going to be able to approach surgery with better preparation and set patients up for realistic expectations. For instance, if we get an imaging study before and we see this big bowel lesion, we can tell the patient “The likelihood of us having to do a bowel resection is high. But I do not think we're going to have to do a colostomy because the lesions is far enough away from the anal verge. I'm fairly confident that we will be able to put you back together.” Or if the lesion is very low, you're going to say, “We're going to have to see with surgery. But there is a chance that you might wake up with a temporary colostomy.” We had to do that one time in a patient where we looked at her MRI and we said, “This is no good. We are not going to be able to put you back together.” And so she knew about it. She understood she would likely wake up with an ostomy bag, require about 3 months of recovery, and then have it reversed.” Patient counseling is huge. Because before, if we didn't have that imaging, we'd just go in and be like, oh, no—the patient has no idea she's going to need a bowel resection or she may wake up with a colostomy. That’s a big thing. [It’s] the same thing with the ureter. If you see significant hydroureter—which means that it’s very dilated or that it's being squished by all the deep endometriosis which can compromise the kidney -- if we see that on ultrasound or MRI, we'll say, “There's a likelihood that we may have to resect a little part of your ureter and then put it back in the bladder. That's called a ureteral reimplantation. You’re going to have a stent for 6 weeks. You're going to have to go home with a Foley catheter in your bladder for 2 weeks while it heals.” It’s just better for the patient, right? Instead of saying, “We'll see what we see. We'll handle it.” Also, from a surgical standpoint and an operating room logistics standpoint, you can go into the OR at the beginning of the case and say, “I need X, Y, and Z because the likelihood of doing a bowel resection is high. We're going to need you to pull all the equipment.” This makes the OR much more efficient - no scrambling for equipment mid-case. Without that preparation, needless delays prolong both anesthesia and operative time.

How might AI-assisted imaging improve identification of complex disease involving bowel, bladder, or nerves?

As I mentioned before, endometriosis imaging is very operator dependent. There are sonographers out there who are extremely good at mapping. Highly skilled technicians can see everything that they need to see with an ultrasound, whereas a regularly trained ob-gyn is not going to be able to do that kind of dynamic assessment. The same is true with MRIs. I usually like to have my patients send me their MRI images so I can look at them. I'm not a radiologist, but I look for things a general radiologist may not be trained to recognize. A general radiologist is reading brain scans, foot scans, bone scans, oncology studies—a wide range of conditions, not necessarily with specific expertise in endometriosis. So it's very operator dependent. With AI, when you can get these top-notch ultrasound technicians who can really train AI and tell it, “This is what you look for”—or MRIs, where you can just plug in your MRI images and have AI read it specifically for endometriosis, for adenomyosis, for X, Y, and Z—it's just going to be huge. I think AI in health care is always a little bit slow to adopt—and I understand why—but it will be a game changer.

Is recurrence primarily a surgical issue, a biologic disease process, or a combination—and how should this influence postoperative management?

Recurrence is tricky. There is surgical excision and surgical ablation. Ablation is where you burn the disease. Excision is where you resect the disease. Even if you do excision, which is the gold standard, there's still a lot of variability among surgeons in terms of technique and whether they really get all the disease. You can never be sure that you get all the disease, but I think incomplete excision is often labeled as a recurrence, even though it's not a recurrence—the disease was left behind the first time or the second time or the third time. So it's not really a recurrence. It was just never resected or excised to begin with. It's more residual.

However, there are patients who truly do have a recurrence. You get all the endo out, and then in a few years, you go back in and you think, “How did this happen? We did a complete resection.”

There's a strong biological component. Endometriosis lesions are heterogeneous. The disease involves inflammatory, hormonal, and neuroimmune factors. Recurrence reflects that same complexity—surgical and biological forces are both at play. Surgery is one part of treatment for endometriosis. Before, we used to think, “Just have surgery, get rid of it, you're good for the rest of your life.” I don't think people think like that anymore. Because we now understand it's far more complex. Surgery is one pillar of treatment; biological, neuroimmunologic and hormonal factors are the other pillars.

What are the biggest gaps in evidence comparing excision surgery, ablation, and long-term medical therapy?

The biggest challenge is that most of the studies are not truly apples-to-apples comparisons. Like I said before, the terms excision and ablation can mean very different things depending on the surgeon's training [and] type of disease--whether it's deep disease, whether it's peritoneal, whether their uterus is involved. We haven't even talked about adenomyosis, but adenomyosis is essentially endometriosis of the uterus. That can throw in a whole other curveball for these comparison studies. So if someone has adenomyosis but they have a complete excision, they still have disease left behind in the uterus. It’s not necessarily a full excision because they still have the uterus.

Many trials focus on short-term outcomes—6 months, 12 months—which is great. But what about 3 years? 5 years? 10 years? We don't have many of those studies. Endometriosis is very complex. There's peritoneal disease, there's deep disease. There are patients with adenomyosis, there are patients with endometriomas—chocolate cysts representing deep ovarian disease. There are so many different factors. Designing a rigorous study that meaningfully compares these approaches across all those variables is an enormous challenge.

Are registries or large data sets beginning to clarify which patients benefit most from excision surgery?

There are some registry sets out there, and they're starting to collect data. And I think what they suggest—we don't know for sure—is that patients with deep disease do better long term, in terms of symptom relief, than patients with just peritoneal disease. They also have higher rate of recurrence, but their symptoms do seem to respond better to surgery. [It’s the] same with patients who've had multiple surgeries. Patients who have a very good surgery the first time around will do better than someone who has multiple surgeries over and over again with incomplete excision.. But again, these are small data sets.

There are quite a few registries now. There is one in Europe and one in Australia, as well as others, which are aggregating data across multicenter academic institutions to identify meaningful patterns. I'm very encouraged by what those efforts may ultimately reveal.

Do you see emerging technologies—robotics, augmented reality, or intraoperative imaging—meaningfully changing excision surgery in the next decade?

I use robotics. My father, who was one of the pioneers in gynecologic laparoscopy never did robotics because it was just emerging towards the end of his career. I will say from an ergonomic standpoint, robotics is much better for the surgeon, plus you have instrument articulation and wristed movement. Also, you’re able to have a 3D view of the surgical field. You can really get to see the depth of the disease. Robotics has been a transformative step forward for endometriosis surgery and minimally invasive gynecology more broadly.

In terms of intraoperative imaging, in my practice we use ICG to visualize the ureter. So instead of having to dissect out the ureter—which we do anyway—at least you'll be able to do a quick fluorescence check and say, “Oh my gosh, there's the ureter right there. I didn't think it was so close.” Some are also using ICG or fluorescence to try to visualize endometriosis, nerves, and bowel integrity, which is is an exciting development that could meaningfully improve visualization. Augmented reality holds similar promise—the idea of overlaying normal anatomy onto a frozen pelvis to guide restoration is a genuinely exciting frontier.

Could AI or predictive analytics eventually help surgeons anticipate operative complexity or complication risk before entering the OR?

Absolutely. This is probably where AI can help us the most in the more immediate future. As with imaging interpretation, AI can help us estimate the complexity of surgery. It can generate outputs like, “Based on this MRI, you have a 40% likelihood of needing a bowel resection.” Or “The surgery is going to take about 4 hours.” I think it could deliver that kind of preoperative risk stratification automatically. Obviously, AI is not perfect, and you can never ignore the human element when it comes to surgery and healthcare in general. But I think it really could guide us. For surgeons without a multidisciplinary team readily available, AI could flag something like, “This ureter looks really bad. You should have urology available in this upcoming case because the likelihood of needing a ureteral reimplantation is about 65%.” It could really help.

How should general ob-gyns decide when to manage medically vs referring for specialized excision surgery?

I have such respect for general ob-gyns. They are so busy, and they are doing remarkable work, and endometriosis is a complex subspecialty. When it comes to knowing when to refer, my first piece of advice is simple: if you're uncertain or uncomfortable managing it, refer. Don't reassure a patient with “You're fine” when something doesn't feel right. Trust that instinct and send her to a specialist.

Second, if imaging shows an endometrioma or adenomyosis, that is a clear signal of deep disease warranting referral. Similarly, if patients are not responding to standard first-line management—NSAIDs, oral contraceptives, or an IUD—and are still reporting significant pain at follow-up, that's reason to refer. Or if they're coming in asking for pain medication—if they're saying, “I need oxycodone to get through my period”—that's a red flag; this is not a normal period. Likewise, significant distress on pelvic exam should prompt specialist involvement.

If one scientific breakthrough could most improve surgical outcomes for patients with endometriosis, what would it be?

You know how they have PET scans for cancer? After you've had cancer, you go in every 6 or 12 months, you get a PET scan and make sure that nothing lights up. What if there were something similar for endometriosis—where a patient with suspected disease could swallow or inject a tracer, and then an hour later they get imaging, and any endometriosis lesions would show uptake—highlighted on imaging. This “tracer” could also guide surgeons in the OR, helping them to identify subtle lesions in real time. Endometriosis lesions can be tiny and very hard to detect, especially if you don't know what you're looking for. Equally important, let's say you think someone has endometriosis and they take this tracer, and nothing lights up—you could tell the patient, “Your pain is likely not from endometriosis, or not from recurrence.” The applications could span preoperative planning, intraoperative guidance, and postoperative surveillance. That would be a genuine breakthrough.