CT scans are among the most powerful tools in cancer detection, but they are not infallible. Tumors can be missed because they are too small, blend into surrounding tissue, sit in an area the scan covers poorly, or simply because the scan was timed in a way that made the lesion invisible. In a lung cancer screening study, 32 cancers were missed across 39 CT scans, split roughly between the radiologist not spotting the abnormality and the radiologist seeing it but not recognizing it as cancer. The reasons behind these misses are worth understanding, because knowing the limitations of a CT scan can shape what you and your doctor do next.
Why Tumor Size and Tissue Density Matter
The most intuitive reason a CT scan misses cancer is that the tumor is simply too small to see clearly. CT imaging works by measuring how much X-ray energy different tissues absorb. A tiny cluster of cancer cells, just a few millimeters across, may not absorb enough differently from its surroundings to stand out. In one study of lung cancers missed during low-dose CT screening, the average diameter of the missed cancers was about 5 mm at the time of the initial scan, growing to roughly 11 mm by the time they were eventually caught on a follow-up scan.1PubMed. Low-Dose CT Screening for Lung Cancer: Computer-aided Detection of Missed Lung Cancers A 5 mm lesion is about the size of a small pea, and distinguishing it from normal tissue or a benign finding is genuinely difficult.
But size alone does not explain every miss. Some tumors are “isoattenuating,” meaning they absorb X-rays at the same rate as the organ they are growing in, making them effectively invisible on a standard scan. This is a particular problem for pancreatic cancer. Isoattenuating tumors account for roughly 11 to 14 percent of all pancreatic cancers, and most of these are small.2PubMed Central. Suspicious findings observed retrospectively on CT imaging performed before the diagnosis of pancreatic cancer Because the pancreas is a soft organ and pancreatic tumors often have a similar density to normal pancreatic tissue, these cancers can hide in plain sight even on well-performed scans. In one study, all of the mass-like lesions that were retrospectively identified had a median size of just 1.2 cm, a size where isoattenuation makes detection extremely challenging.
Radiologists sometimes spot indirect clues instead of the tumor itself. A pancreatic duct that is dilated or cut off abruptly, a patch of focal tissue thinning, or unexplained inflammation near the pancreas can all suggest a hidden malignancy. These subtle signs are among the most commonly missed early indicators of pancreatic cancer on CT.3PubMed. Hidden in plain sight: commonly missed early signs of pancreatic cancer on CT The challenge is that these indirect findings also show up in benign conditions, so interpreting them correctly requires considerable expertise and clinical context.
How Tumor Composition Affects Visibility
Not all cancers are made of the same stuff, and the physical makeup of a tumor changes how it appears on CT. Mucinous tumors, for example, produce large amounts of mucin, a gel-like substance. In colorectal cancer, mucinous carcinomas show up differently from their non-mucinous counterparts: they tend to have more areas of low density on CT, and the solid portions enhance less with contrast dye.4PubMed. CT differentiation of mucinous and nonmucinous colorectal carcinoma That lower enhancement can make them harder to distinguish from fluid or benign tissue, particularly if the radiologist is not specifically looking for this pattern.
Cancers that spread as thin sheets rather than forming distinct lumps present an even tougher problem. Peritoneal carcinomatosis, where cancer seeds itself across the lining of the abdominal cavity, is a notorious blind spot for CT. The tumor deposits are often individually tiny, and they can take the form of flat plaques or sheets only a few cells thick. Early disease may be microscopic and confined to abdominal fluid, while more advanced disease can coat organ surfaces or thicken the omentum.5Journal of Nuclear Medicine. Peritoneal Carcinomatosis: Role of 18F-FDG PET CT is built to spot masses, not films of tumor, and this fundamental mismatch helps explain why peritoneal disease frequently goes undetected until surgery.
Anatomic Blind Spots on CT
Every imaging technique has areas it covers well and areas where visibility drops. On chest CT, the regions where the lungs meet the heart, the major blood vessels, and the spine are common blind spots. Tumors near the lung hilum (where the airways and blood vessels enter the lung) or tucked against the mediastinum (the central compartment of the chest) are harder to see because surrounding structures create visual clutter. A review of clinical blind spots on chest CT catalogued a long list of these trouble zones, from the perihilar and paramediastinal regions to the thyroid, the bones, the breast tissue, and even the upper abdomen at the bottom edge of the scan.6PubMed Central. The blind spots on chest computed tomography: what do we miss
Similar blind spots exist in body CT. In oncology patients specifically, studies comparing standard CT with PET/CT have identified findings in the breast, lung, colon, ovaries, and even blood vessels that are commonly overlooked on CT alone.7PubMed Central. Blind spots at oncological CT: lessons learned from PET/CT These are not equipment failures. They are limitations of how two-dimensional cross-sectional images represent three-dimensional anatomy. A lymph node tucked behind a large blood vessel, a small tumor nestled against the diaphragm, or a lesion at the very edge of the scan field are all examples of findings that are physically present in the images but easy to overlook.
When Contrast Timing Makes or Breaks the Scan
Many CT scans involve injecting a contrast dye that temporarily makes blood vessels and well-supplied tissues brighter on the images. The timing of the scan relative to that injection is critical, because different tumors show up best at different moments. Some cancers, like hepatocellular carcinoma (the most common liver cancer), are fed by arteries and light up brightly in the first few seconds after contrast reaches the liver, a window called the arterial phase. If the scan is taken too early or too late, the tumor may not stand out.8PubMed Central. Computed Tomography Techniques, Protocols, Advancements, and Future Directions in Liver Diseases
The practical challenge is that people’s circulatory systems vary. In one study, the time it took for contrast to reach peak concentration in the aorta ranged from 18 to 32 seconds across patients. When scan timing was tailored to each patient rather than using a fixed delay, the proportion of scans rated as optimally timed jumped significantly, and the visibility of liver tumors improved substantially.9PubMed. Patient-tailored scan delay for multiphase liver CT: improved scan quality and lesion conspicuity with a novel timing bolus method Older research on liver tumors in patients with cirrhosis confirmed the same principle: certain small hepatocellular carcinomas were visible only during the arterial phase and would have been missed entirely if the scan captured only the portal venous phase.10PubMed. Multiple-phase helical CT of the liver for detecting small hepatomas in patients with liver cirrhosis: contrast-injection protocol and optimal timing Using a one-size-fits-all timing protocol leaves some patients with suboptimal scans, and suboptimal scans miss more cancers.
Human Factors and Cognitive Bias
Even when the tumor is visible on the images, a radiologist may not recognize it. Radiology is fundamentally a pattern recognition task performed under time pressure, and errors happen. In the lung cancer screening study mentioned earlier, missed cancers fell into two categories: detection errors, where the radiologist simply did not notice the abnormality, and interpretation errors, where the radiologist saw something but classified it as benign.11PubMed. Lung cancers missed at low-dose helical CT screening in a general population: comparison of clinical, histopathologic, and imaging findings Detection errors were more common, which suggests that the challenge is often about attention and search patterns rather than knowledge.
Cognitive biases play a real role. “Satisfaction of search” is a well-documented phenomenon where finding one abnormality makes the radiologist less likely to keep looking for a second one. “Anchoring bias” occurs when a radiologist fixates on the first possible diagnosis that comes to mind and interprets subsequent evidence in light of that initial impression. These are not signs of incompetence; they are features of how human cognition works under conditions of complexity and uncertainty.12PubMed. Cognitive and system factors contributing to diagnostic errors in radiology System-level factors compound the problem: inadequate clinical history provided to the radiologist, poor imaging technique, excessive workload, and suboptimal reading conditions all increase the risk of a miss.13PubMed. Minimising the impact of errors in the interpretation of CT images for surveillance and evaluation of therapy in cancer
Awareness of these biases has led to practical countermeasures. Structured reporting templates force radiologists to systematically evaluate each anatomic region rather than relying on a free-form search. Double reading, where a second radiologist reviews the same scan, catches additional findings. And increasingly, artificial intelligence is being deployed as a safety net, a topic covered further below.
When Other Imaging Tools Fill the Gaps
If CT has known blind spots, the logical question is whether other imaging methods can cover them. The answer is often yes, but the right tool depends on the cancer type and body region.
PET/CT combines a standard CT scan with a metabolic tracer, usually a radioactive sugar that cancer cells absorb more readily than normal cells. For lung cancer staging, PET/CT has higher sensitivity and accuracy for characterizing lung nodules than CT alone, and it is particularly good at finding unexpected metastases outside the chest.14PubMed. PET/CT versus MRI for diagnosis, staging, and follow-up of lung cancer However, PET/CT has its own limitations. Very small, slow-growing, or low-metabolism tumors may not take up enough tracer to be visible. Ground-glass opacities in the lung, which can represent early or indolent cancers, are one example where PET/CT often adds little to what CT already shows.
MRI uses magnetic fields rather than X-rays and excels at soft tissue contrast. For detecting liver metastases from colorectal cancer, MRI significantly outperforms both CT and PET/CT, especially for small lesions. In one study, CT detected only about 16 percent of colorectal liver metastases smaller than 10 mm, while MRI caught about 74 percent of the same small lesions.15PubMed. Diagnostic performance of CT, MRI and PET/CT in patients with suspected colorectal liver metastases: the superiority of MRI That is a dramatic difference, and it explains why guidelines for colorectal cancer patients often call for liver MRI rather than relying on CT alone for staging.
The takeaway is not that CT is inferior, but that no single imaging modality catches everything. The best diagnostic approach often involves layering multiple tools, each covering the other’s weaknesses.
AI as a Second Set of Eyes
Artificial intelligence systems trained on thousands of CT scans are increasingly being tested as decision-support tools for radiologists. In one large study, an AI system detected primary lung cancers on non-screening chest CT with about 97 percent sensitivity and spotted metastases with about 92 percent sensitivity, performance that was comparable to or slightly better than radiologists reading the same scans.16Communications Medicine. Deep learning for the detection of benign and malignant pulmonary nodules in non-screening chest CT scans The tradeoff was a slightly higher rate of false positives, about 0.6 extra false alarms per scan compared to the radiologists.
Where AI may prove most useful is not in replacing radiologists but in catching what they miss. A study comparing radiologists reading lung CT scans with and without AI assistance found that the number of missed pulmonary nodules dropped significantly when AI flagged suspicious areas for the radiologist to review.17PubMed Central. Impact of AI Assistance on Radiologist Accuracy for Lung Nodule Detection on Chest CT AI does not get tired, does not suffer from satisfaction of search, and processes every voxel of every image with the same level of attention. Used alongside a human reader, it addresses some of the cognitive and fatigue-related vulnerabilities that contribute to missed findings.
That said, current AI systems are best validated for specific tasks like lung nodule detection. Their reliability for other cancer types, in different body regions, or in patients with complex anatomy is still being established. Regulatory approval, integration into clinical workflows, and radiologist trust remain practical hurdles.
Newer CT Hardware That Improves Detection
The CT scanners themselves are evolving in ways that directly address detection gaps. Dual-energy CT scans the body at two different X-ray energy levels simultaneously. Because different tissues absorb low-energy and high-energy X-rays differently, dual-energy CT can distinguish materials that look identical on a conventional scan. This capability is useful for identifying iodine uptake in tissues (which helps differentiate enhancing tumors from cysts), characterizing bone marrow for metastatic disease, and improving lesion visibility overall.18PubMed Central. Dual-Energy CT in Oncologic Imaging The technology generates specialized image reconstructions, including virtual images at specific energy levels that can boost the contrast between a tumor and its background, making lesions that would be invisible on a standard scan become detectable.19PubMed. Dual-energy (spectral) CT: applications in abdominal imaging
Early research into new contrast agents designed specifically for dual-energy CT is also showing promise. In a preclinical study, a tantalum oxide-based contrast agent provided higher tumor visibility scores and detection rates than conventional iodine-based contrast across all scan time points.20PubMed. Evaluation of a Novel Tantalum Oxide-Based Contrast Agent for Liver Imaging and Tumor Detection in Dual-Energy CT: A Preclinical Proof-of-Concept Study This is still experimental and years from clinical use, but it illustrates how hardware and chemistry are co-evolving to close the detection gap.
Photon-counting detector CT is the next generation beyond dual-energy. Conventional CT detectors integrate all the photon energies that hit them into a single signal, losing spectral information in the process. Photon-counting detectors register each individual X-ray photon and measure its energy, enabling sharper images, better contrast between tissues, and intrinsic spectral analysis without the dose penalty of scanning twice. These capabilities directly address many of the limitations of conventional CT in oncology, from resolving small lesions to better characterizing tissue composition.21PubMed Central. Photon-counting detector CT in oncology: a new era of cancer imaging Photon-counting CT systems are now entering clinical use at major medical centers, though widespread availability will take time.
Liquid Biopsy and the Move Beyond Imaging Alone
One of the most significant shifts in cancer detection strategy is the recognition that imaging does not have to work alone. Liquid biopsy, which analyzes blood samples for fragments of tumor DNA, circulating tumor cells, or other cancer-derived molecules, offers a fundamentally different approach to finding cancer. Because tumor material enters the bloodstream from all sites of disease, liquid biopsy can capture information about cancers that imaging cannot see, whether because of size, location, or tissue density.22PubMed Central. Liquid biopsies: Potential and challenges
Liquid biopsy is already in clinical use for monitoring known cancers and detecting recurrence. Its role in primary screening is still being refined. Multicancer early detection tests, which look for signals from multiple cancer types in a single blood draw, are in large-scale clinical trials. These tests are not replacements for imaging; they are better understood as complementary tools. A blood test might flag that cancer is present somewhere in the body, and imaging would then localize it. The combination could be especially valuable for cancers that CT tends to miss, like early pancreatic cancer or peritoneal spread, where imaging alone has well-documented limitations.
What to Do If You Are Worried About a Missed Cancer
If you have symptoms that persist despite a clean CT scan, it is reasonable to ask your doctor about follow-up. A single negative CT scan does not guarantee that cancer is absent, particularly for the cancer types and anatomic regions discussed above. Several practical steps can make a difference:
- Ask about the protocol: Was contrast used? Was the scan timed appropriately for the organ in question? A non-contrast CT of the liver, for example, will miss many cancers that a properly timed multiphase scan would catch.
- Request a complementary study: If your doctor suspects a specific cancer type where CT has known weaknesses, MRI or PET/CT may be appropriate as a next step.
- Provide full clinical history: Radiologists interpret images in context. Knowing your symptoms, risk factors, and family history helps them look in the right places and raise the right level of suspicion.
- Consider a second opinion: Having another radiologist review the same images can catch findings that were overlooked the first time. This is especially worthwhile when clinical suspicion remains high despite a negative report.
- Follow-up imaging: For indeterminate or borderline findings, short-interval follow-up scans (often at three to six months) can reveal growth that confirms or rules out cancer. Some cancers are simply too small to characterize on a first scan but become diagnosable once they have grown slightly.
Ground-glass opacities in the lung illustrate the value of patience and follow-up. These hazy areas on CT can represent infection, inflammation, or early cancer. Many are managed with periodic low-dose CT surveillance rather than immediate biopsy, because only a subset will progress, and the risk of an invasive procedure may outweigh the benefit when the lesion is tiny and stable. If the solid portion of such a nodule grows over time, that becomes a clearer signal to investigate further.
The broader message is that cancer detection is increasingly a process rather than a single event. No imaging study is perfect, and the most effective approach combines the right scan, performed with the right technique, read by an attentive radiologist, and situated within a clinical context that includes your symptoms, your history, and sometimes additional diagnostic tools like blood-based biomarkers. The technology is getting better year by year, with AI, dual-energy CT, and photon-counting detectors all chipping away at the gaps. But understanding that those gaps exist is the first step toward making sure they do not have the final word.