New AI-based tech helps to find and diagnose lung cancer

AI is making great strides in the medical field. A new AI-powered digital pathology platform will analyze lung tissue to help diagnose cancer.

AI technologies have quickly gained traction in medicine, achieving success in automating tasks and analyzing medical images.

New research by the University of Cologne's Faculty of Medicine and University Hospital Cologne, led by physician Dr. Yuri Tolkach and professor Dr. Reinhard Büttner, has contributed to yet another AI breakthrough. The team built a digital pathology platform that uses algorithms to automatically analyze tissue sections from lung cancer patients.

Lung cancer is one of the most common cancers in humans and has a very high death rate. Particularly non-small cell lung cancer (NSCLC) is the second most common and the deadliest type of epithelial cancer. It makes up over 80% of all lung cancers.

Successful treatment depends heavily on precise pathological examination when pathologists analyze biopsies and resection specimens. This process has undergone intense digitalization, allowing healthcare providers to analyze specimens on the screen. The adoption of AI can further optimize this process.

"The new tools can not only improve the quality of diagnosis but also provide new types of information about the patient's disease, such as how the patient is responding to treatment," explained Dr. Tolkach.

Scientists trained AI on the largest high-quality training dataset available, enabling the technology to quickly analyze biopsy samples, precisely segmenting 11 types of tumor and benign tissue at the pixel level.

In a research paper published in Cell Reports Medicine, they showed how their tool is useful in two key ways: first, by creating a precise model to identify types of NSCLC, which was tested and confirmed with patient data from multiple hospitals.

Second, they identified four measurable markers in tissue samples that can help predict how the cancer might progress and how long a patient might survive. Additionally, they publicly shared three datasets to support global lung cancer research and algorithm development.

The researchers believe their platform will be useful for diagnosing diseases, predicting how they will progress, and helping with future predictions. According to the university's press release, the team will continue to conduct validation studies with five pathological institutes in Germany, Austria, and Japan.

AI is not always accurate with its diagnoses

While AI has been highly effective in automating healthcare tasks and analyzing medical images, commercial AI tools like ChatGPT are also being increasingly used by patients for self-diagnosis.

Research indicates that ChatGPT effectively provides information and support across a range of scenarios, including mental health assessments, counseling, medication management, and patient education. Nonetheless, some studies highlighted that ChatGPT's accuracy in diagnosing conditions in children was only 17%.

However, research into popular AI tools like Llama-2-chat, Vicuna, Medllama2, Bard/Gemini, Claude, ChatGPT-3.5, and ChatGPT-4 showed that they were not very accurate in diagnosing genetic diseases.