Can Ultrasound Detect Stomach Cancer?

Ultrasound can detect stomach cancer, but how well it does so depends heavily on the type of ultrasound used and how far the cancer has progressed. A standard transabdominal scan picks up advanced gastric tumors with high reliability, yet catches early-stage cancers less than two-thirds of the time. Endoscopic ultrasound, which places the probe inside the stomach itself, offers much finer detail for staging. The gap between these approaches, and the situations where each one matters, is wider than most people expect.

Transabdominal Ultrasound and Advanced Versus Early Cancer

A transabdominal ultrasound is the type most people picture: a technician presses a probe against your abdomen and reads the image on a screen. For stomach cancer, a meta-analysis pooling data from multiple studies found an overall pooled accuracy of about 65%, but that number hides a striking split. For advanced gastric cancers, the sensitivity reached roughly 99%, meaning it almost never missed a large, deeply invasive tumor. For early-stage cancers, the sensitivity dropped to around 61%.¹ That is a huge gap. If you already have a bulky tumor thickening the stomach wall, a belly scan stands a good chance of flagging it. If the cancer is shallow and small, the same scan will miss it close to four times out of ten.

A separate study looking at preoperative staging accuracy painted a similar picture. Transabdominal ultrasound correctly staged T3 tumors (those invading through much of the stomach wall) about 87% of the time but managed only around 56% for T1 tumors, the earliest stage. The overall staging accuracy was roughly 78%.² These numbers tell the same story: the thicker and more invasive the tumor, the easier it is for a standard ultrasound to see it.

Improving the Image With Oral Contrast Agents

One of the biggest problems with scanning the stomach through the abdominal wall is that the stomach is usually full of air, folds, and variable amounts of food. Air scatters ultrasound waves, which is why sonographers typically ask you to drink water before a scan. Researchers have pushed this idea further by having patients swallow an echoic cellulose-based contrast agent, a thick liquid that fills the stomach and creates a uniform background against which tumors stand out more clearly.

In one study using this approach, the detection rate for advanced gastric cancers hit 100%. For shallower T1b tumors it was about 77%, and for the shallowest T1a cancers it was 67%. The overall accuracy for determining how deep the tumor had invaded was around 77%, with the best results again in more advanced disease.³ Most of the patients whose tumors were missed were obese, with BMIs in the high 20s to low 30s, reinforcing a persistent limitation of scanning through the belly wall.

A Chinese screening program tested oral contrast transabdominal ultrasound in rural areas where endoscopy was not widely available. Using a wall-thickness cutoff of 7 mm, they achieved about 81% sensitivity and 69% specificity for detecting high-risk disease. When they combined multiple ultrasound parameters into a risk score, sensitivity climbed to about 94%.⁴ This is not as precise as endoscopy, but as a first-pass screening tool in settings where endoscopes and trained endoscopists are scarce, it fills a real gap. A 2025 Chinese expert consensus statement endorsed oral contrast ultrasound as a promising, non-invasive initial screening tool, particularly given the difficulty of rolling out nationwide endoscopy programs for large at-risk populations.⁵

Endoscopic Ultrasound for Staging

Endoscopic ultrasound, or EUS, is a fundamentally different beast. A tiny ultrasound transducer is mounted on the tip of an endoscope and placed directly against the stomach wall from the inside. Because there is no skin, fat, or bowel gas in the way, the image resolution is far higher. EUS can distinguish the individual layers of the stomach wall, which is essential for determining how deeply a tumor has invaded.

An older but well-regarded comparison found that EUS matched the final pathology in 92% of cases for depth of tumor penetration, compared to just 42% for dynamic CT.⁶ That wide margin has narrowed somewhat as CT technology has improved. A more recent study comparing multi-detector CT with virtual gastroscopy against EUS found staging accuracies of about 82% and 84%, respectively, with no statistically significant difference.⁷ Modern CT has clearly closed the gap, but EUS retains a particular advantage for distinguishing between T1 and T2 tumors, a distinction that often determines whether a patient can be treated with endoscopic resection or needs surgery.

That said, EUS staging is not perfect. One retrospective study of over 200 patients found an overall T staging accuracy of about 64%, with accuracy as low as 33% for T1 tumors and highest at 75% for T3 tumors. Overstaging occurred in about 13% of cases and understaging in about 23%.⁸ This variability is one reason surgeons rarely rely on any single imaging method alone.

Checking for Lymph Node Spread

Whether cancer has reached nearby lymph nodes is one of the most important questions in treatment planning. EUS can visualize the lymph nodes surrounding the stomach and assess their size, shape, and internal echo pattern. A meta-analysis of EUS accuracy for lymph node staging found pooled sensitivity of about 82% and specificity of about 68%.⁹ That specificity figure is worth pausing on: roughly a third of the time, EUS identified a node as cancerous when it was not, or vice versa. Lymph nodes can be enlarged from inflammation rather than cancer, and small nodes can harbor microscopic tumor deposits that ultrasound simply cannot resolve.

Transabdominal ultrasound performs somewhat similarly for lymph node assessment. One study reported sensitivity of about 78% and specificity of about 64% for detecting lymph node metastasis.¹⁰ Neither approach is definitive, which is why the final lymph node status is often confirmed only after surgery, when a pathologist examines the removed nodes under a microscope.

Why Certain Patients Get Less Reliable Results

Obesity is the single biggest technical obstacle for transabdominal stomach ultrasound. More tissue between the probe and the stomach means the ultrasound beam scatters and weakens. A study of ultrasound image quality found that BMI was the strongest predictor of poor imaging, explaining a large share of the variation in quality scores.¹¹ This aligns with the oral contrast study mentioned earlier, where undetected tumors clustered in patients with BMIs above 28.

Gas inside the stomach and intestines is another persistent problem. Unlike water or tissue, gas reflects ultrasound waves almost completely, creating bright white patches that block the view. Operator skill compounds both issues. A review on gastric ultrasound assessment noted that experienced operators produce highly reproducible measurements, while less experienced ones need structured training before results become reliable.¹² Ultrasound is inherently operator-dependent in a way that CT and MRI are not. Two sonographers scanning the same patient can come away with meaningfully different images depending on probe angle, pressure, and technique.

Contrast-Enhanced Ultrasound and Tumor Blood Flow

Beyond simply looking at the stomach wall’s structure, researchers have explored using microbubble contrast agents injected into the bloodstream. These tiny bubbles light up on ultrasound as they flow through blood vessels, allowing clinicians to see the tumor’s blood supply in real time. Because cancers generally recruit new blood vessels to fuel their growth, contrast-enhanced ultrasound can highlight areas of abnormal vascularity.

One study combining intravenous microbubbles with an oral contrast agent found a strong positive correlation between the contrast-enhanced signal intensity and the density of microscopic blood vessels within gastric tumors. Both the signal intensity and the vessel density were much higher in cancerous tissue than in normal stomach wall.¹³ Another study found that contrast-enhanced ultrasound parameters correlated significantly with perfusion CT measurements and differed between well-differentiated and poorly differentiated tumors.¹⁴ This kind of information could eventually help predict how aggressive a tumor is before any tissue is removed, though it remains largely a research tool rather than routine clinical practice.

Tissue Sampling Through Endoscopic Ultrasound

One advantage unique to EUS is that it can guide a needle directly into a suspicious lesion while the operator watches on the ultrasound screen. This matters most for submucosal tumors, the kind that grow under the stomach’s inner lining and are often invisible on a standard endoscopy. These lumps could be anything from a benign muscle tumor to a gastrointestinal stromal tumor (GIST) that needs treatment.

Diagnostic yields from EUS-guided needle sampling vary. A study using a core biopsy needle reported diagnostic results in 86% of cases.¹⁵ An older study using a different needle type achieved adequate tissue in about 63% of patients, with high agreement between the needle sample and the final surgical specimen when tissue was obtained.¹⁶ Location within the stomach also affects success: one study found that sampling tumors in the gastric antrum (the lower part of the stomach near the exit) yielded diagnostic tissue in none of the five attempts, while tumors in the body and upper stomach were successfully sampled in over 80% of cases.¹⁷

Tracking Treatment Response

For patients receiving chemotherapy before surgery, doctors need to know whether the treatment is shrinking the tumor. CT scans are the standard tool for this, but they measure the tumor’s size, not its blood supply. A tumor can stop growing without shrinking much, because dead cancer cells, scar tissue, and swelling fill the space the living cancer vacated. Contrast-enhanced ultrasound can potentially catch this by showing decreased blood flow in a responding tumor even when the size on CT has not changed. One study reported that contrast-enhanced ultrasound was slightly more accurate than CT for predicting the degree of pathological response to chemotherapy, though the difference was not statistically significant in that small sample.¹⁸

EUS has also been tested for restaging after chemotherapy. The absolute staging accuracy tends to drop after treatment, likely because chemotherapy-induced inflammation and fibrosis distort the wall layers that EUS relies on to judge depth. However, seeing a downshift of two or more stages on sequential EUS scans was significantly associated with a better long-term outcome, suggesting the trend is more informative than the exact stage call.¹⁹

Ultrasound During Surgery

There is one more setting where ultrasound plays a role that often surprises people: inside the operating room, during the surgery itself. When a surgeon is performing a minimally invasive (laparoscopic) gastrectomy for an early gastric cancer, one of the trickiest challenges is figuring out exactly where the tumor is. Early cancers are small, and the surgeon is working with camera-guided instruments, unable to feel the stomach wall directly. Metal clips placed during a prior endoscopy can mark the spot, but they are not always easy to find.

Laparoscopic ultrasound solves this by scanning the stomach wall from the outside during surgery. In one study, the ultrasound probe detected the endoscopically placed clips in every patient, with a mean detection time under five minutes. All resection margins came back tumor-free.²⁰ A more recent comparative study found that laparoscopic ultrasound-guided localization significantly reduced operative time compared to intraoperative gastroscopy (inserting a second endoscope during the operation) while maintaining equivalent safety and cancer outcomes.²¹ It is a niche application, but for the patient on the table, it can be the difference between preserving more of the stomach or losing too much.

Artificial Intelligence and Where the Technology Is Heading

The most operator-dependent imaging modality is naturally one of the first to attract AI assistance. Several research groups have developed deep learning algorithms designed to improve ultrasound interpretation for gastric cancer. One study using a deep convolutional neural network applied to contrast-enhanced gastrointestinal filling ultrasound found that the AI-processed images improved the display of tissue lesions around the stomach and helped judge the depth of tumor invasion and lymph node involvement.²² Another study comparing AI-assisted ultrasound combined with endoscopy against endoscopy alone found that the AI-enhanced approach performed closer to expert-level recognition, with substantially faster processing times.²³

These are early-stage results, and none of these AI systems has entered widespread clinical use. But the appeal is obvious: if an algorithm can compensate for operator inexperience and flag suspicious regions that a less-trained sonographer might miss, ultrasound-based screening could become practical in exactly the resource-limited settings where it is most needed. The 2025 Chinese expert consensus on ultrasound screening already reflects this direction, advocating for structured training programs and standardized protocols that could eventually integrate algorithmic support.²⁴

Cost and Accessibility

One reason ultrasound keeps resurfacing in gastric cancer research despite its limitations is cost. An ultrasound machine is a fraction of the price of a CT scanner, let alone an MRI. It is portable, does not use radiation, and the consumables are minimal. For initial screening of large populations, those economics matter. A cost-effectiveness analysis of diagnostic strategies for upper gastrointestinal malignancies found that strategies combining CT with standard ultrasound alone were not cost-effective, but adding endoscopic ultrasound to the mix improved cost-effectiveness significantly because of its high diagnostic accuracy relative to its procedure cost.²⁵

In countries with high gastric cancer rates and uneven access to endoscopy centers, the tradeoff between sensitivity and accessibility is not academic. Missing some early cancers with an ultrasound screen is bad, but missing all of them because no screen was offered at all is worse. That calculus explains why oral contrast ultrasound screening programs have gained traction in parts of China and why the research pipeline for AI-enhanced ultrasound targets exactly these underserved populations. For patients in well-resourced healthcare systems, ultrasound is more likely to appear as part of a staging workup alongside endoscopy and CT than as a standalone detection tool.