Early detection of cancer – cancer diagnosis 2.0: How Ki ensures the lead

How can artificial intelligence (AI) help recognize cancer in an early stage?

Artificial intelligence (AI) can help with cancer diagnosis primarily through pattern recognition. AI models analyze medical images, such as X-rays or tissue samples, to identify early signs of cancer that may be overlooked by human experts. A AI model such as Atlas, which was trained on millions of tissue samples, and more precisely 1.2 million tissue samples from 490,000 cancer cases, shows promising results by classifying tumors with high accuracy.

Such models could also accelerate the diagnostic process and, in some cases, be more precise than traditional methods. The accuracy of Atlas was compared in tests with six leading AI pathology models that were used in the classification of breast cancer images and the classification of tumors. Atlas exceeded the other models in six of nine tests, especially when diagnosing intestinal cancer tissue, where it agreed with the results of human pathologists in 97.1 percent of cases. In the prostate crab biopsy classification, however, Atlas achieved only 70.5 percent. In total, it was in 84.6 percent of the tests with human experts. However, the human view remains indispensable for many diagnoses.

Prof. Dr. Anabel Ternès is an entrepreneur, future researcher, author, radio and TV presenter. She is known for her work in the field of digital transformation, innovation and leadership. In addition, Ternès is President of the Club of Budapest Germany, board member of the Friends of Social Business and Club of Rome.

What role do pattern recognition and machine learning do when diagnosing cancer?

Pattern recognition and machine learning are key technologies in cancer diagnosis. The first picture recognition options came up about 15 years ago. AI algorithms learn from large data records of medical images and identify subtle patterns that are difficult to recognize for humans. This technology can classify tumors quickly and precisely and help to recognize cancer at an early stage.

Machine learning continuously improves the diagnostic tools by learning from new data and increasing the accuracy over time. “Our task is pattern recognition,” says Andrew Norgan, pathologist and medical director of the digital pathology platform of the Mayo Clinic. “We look at the object carrier and collect the information that has proven to be important.”

What are the challenges and problems when using AI in practice for cancer diagnosis?

The use of AI in the cancer diagnosis faces challenges such as the quality and variety of data required for the training of algorithms. AI models need comprehensive and diverse data records to make precise predictions. In addition, they have to be validated even further in practice, since errors or incorrect diagnoses in sensitive areas such as cancer diagnosis can have serious consequences. Data protection and integration into existing systems are also problematic.

How do you assess the suitability of AI-based tests compared to conventional methods for cancer detection?

AI-supported tests offer potentially greater accuracy and speed in cancer detection, especially when analyzing images and data that is difficult to interpret for humans. They could complement conventional methods, but their suitability requires comprehensive validation and integration into existing systems. AI could be better in some cases, but human expertise remains indispensable in order to correctly interpret the results and avoid misdiagnoses.

Can you tell us more about the cooperation between the Mayo Clinic and Google in the development of a digital pathology atlas?

The Mayo Clinic and Google have worked together to develop a digital pathology atlas that uses machine learning to improve the diagnosis of cancer. This atlas collects and analyzes tissue -based data to recognize patterns that are connected to tumors. The aim is to enable more precise and faster diagnoses and to provide a valuable basis for research.