Artificial intelligence may soon play a major role in catching pancreatic cancer earlier, according to new research showing that an advanced imaging tool outperformed dozens of radiologists in a large international study.
The authors of the study, which appeared in The Lancet Oncology, wrote that AI may support radiologists during busy clinical workloads. They added that the system could help identify cancer earlier in people with no symptoms. However, they also stressed that more research must explore the use of AI in real-world screening settings.
They cautioned that false positives carry clinical and financial consequences, especially when AI tools operate at scale. Consequently, future investigations will focus on reducing unnecessary alerts while preserving high detection rates. In addition, the researchers plan to examine how the tool performs in pre-diagnostic groups, where early detection may offer the greatest benefit.
Pancreatic ductal adenocarcinoma remains one of the deadliest cancers in the world. It causes more than 467,000 deaths every year and often goes undetected until it reaches an advanced stage. Patients usually show few symptoms early on, and the cancer spreads quickly. Researchers noted that this late discovery severely limits treatment options. They added that patients who receive surgery for early-stage disease tend to live far longer, with median survival rising to about 32 months.
A team from Radboud University Medical Center in the Netherlands examined whether AI could push those numbers higher by spotting tumours sooner. The group, led by Natalia Alves, MSc, assessed the performance of an AI system against 68 radiologists in an observational comparative study.
Read more: Breath Diagnostics onboards new president and closes critical financing
Read more: Breath Diagnostics leader speaks at lung cancer education event in Louisville
The AI system reduces false positives by 38%
They used a dataset of 3,340 CT scans, of which about 32 per cent showed pancreatic cancer. The researchers established the reference standard using tissue samples and at least three years of clinical follow-up.
The AI tool delivered an area under the receiver operating characteristic curve of 0.92. The radiologists produced an average AUROC of 0.88. The AI system also posted a sensitivity of 85.7 per cent and a specificity of 83.5 per cent. Furthermore, it reduced false positives by roughly 38 per cent compared with human readers. The team explained that this reduction could ease unnecessary anxiety, cut follow-up procedures and trim costs.
Alves said the technology processed contrast-enhanced CT scans consistently and detected patterns that even experienced clinicians sometimes miss. She noted that patients might benefit if the tool is used during routine imaging for other conditions. Additionally, the researchers pointed out that contrast-enhanced CT scans are already common in hospitals, so AI-driven screening could occur without extra radiation or additional appointments.
Artificial intelligence is gaining ground in several other areas of cancer care. Researchers are building systems that analyse mammograms, lung scans and skin images with growing accuracy. These tools read thousands of images at high speed and offer another set of eyes for clinicians. Furthermore, developers design them to detect subtle patterns that may signal the earliest signs of disease.
In breast cancer, AI models can spot small calcifications that radiologists sometimes overlook. Some systems also compare current scans with past images to track tiny changes. This approach can shorten the time to diagnosis and reduce unnecessary biopsies.
Read more: Breath Diagnostics pioneers novel lung cancer breath test
Read more: Breath Diagnostics takes aim at lung cancer with One Breath
Hospitals are adopting artificial intelligence systems
In lung cancer, AI tools review CT scans for faint nodules that may grow into dangerous tumours. They also measure those nodules with precision, which helps doctors decide when to act.
AI is also moving into pathology labs. Digital pathology platforms examine tissue slides and count cell types. They flag unusual structures and predict how aggressive a tumour may be. Additionally, researchers use AI to study genetic data. This work can identify mutations that drive cancer growth and guide targeted therapies.
Hospitals are adopting AI systems that forecast treatment responses as well. These models analyse clinical records and imaging results to estimate how patients will react to chemotherapy or immunotherapy. In addition, they help teams choose the most effective options for each case.
Several companies are using artificial intelligence to strengthen cancer detection tools. One example is Breath Diagnostics, Inc., a Kentucky-based firm developing a non-invasive breath-analysis test. Its platform captures biomarkers from a single exhaled breath and uses machine-learning models to identify volatile organic compounds linked to lung cancer. The system aims to give doctors a fast, simple method to detect disease without imaging or invasive procedures.
Another major player is GRAIL, Inc. (NASDAQ: GRAL), which created the Galleri multi-cancer early detection test. The company uses a methylation-based sequencing method and advanced algorithms to search for DNA patterns associated with dozens of cancers. The test only requires a blood draw and can detect signals that may appear long before symptoms.
A third company, Dxcover Limited, combines a blood-based “liquid biopsy” with AI-driven spectral analysis. The technology studies subtle molecular changes within a patient’s blood sample. It aims to identify cancer at early stages, when treatment choices are wider and survival odds improve.
.
joseph@mugglehead.com
