Modern medicine is witnessing a major transformation with the increasing reliance on artificial intelligence to improve diagnostic and treatment techniques, especially in complex diseases such as cancer. Medical imaging technologies, such as positron emission tomography (PET) and computed tomography (CT), are prominent examples of how artificial intelligence is used to improve diagnostic accuracy and the speed of treatment decisions.
PET and CT technologies: the basis for accurate diagnosis
PET and CT technologies combine the diagnostic power of monitoring metabolic activity and imaging anatomical structure, respectively. While PET is used to track high metabolic activity in malignant tumors using radioactive glucose molecules such as fluorine-18-deoxyglucose (FDG), CT provides precise anatomical images showing tumor locations and structural details. When these two technologies are combined, doctors get a comprehensive view that helps determine the optimal treatment plan.
Challenges of traditional evaluation
Doctors often face challenges in evaluating tumors due to the large number of lesions that appear in cancer patients. The process of manually sizing and counting these tumors requires arduous effort and a long time, in addition to the possibility of human error. These complexities highlight the need for more sophisticated solutions, prompting researchers to turn to automated systems powered by artificial intelligence.
Artificial intelligence changes the rules of the game
Professor Rainer Stiefelhagen from the Karlsruhe Institute of Technology (KIT) leads a specialized team to develop algorithms based on artificial intelligence with the aim of improving the accuracy and speed of tumor evaluation. Stiefelhagen points out that these automated systems are capable of significantly reducing the time required to analyze images while enhancing the accuracy of results.
The autoPET challenge: a step towards the future
In 2022, Stiefelhagen’s team, in cooperation with the Institute for Artificial Intelligence in Medicine (IKIM), participated in the international autoPET challenge. This challenge brought together researchers from around the world to develop algorithms capable of automatically segmenting active tumors in PET/CT images.
The challenge focused on using deep learning techniques, where algorithms were trained on massive, annotated datasets. The results showed that using an ensemble of algorithms was the most efficient in detecting tumors compared to individual algorithms, which enhances the performance of automated systems in clinical environments.
The future of artificial intelligence in medicine
Despite the tremendous developments, there are still challenges related to the adaptation of algorithms to various clinical conditions. Researchers are working to improve automated systems to become more accurate and flexible in dealing with diverse real-world scenarios. The goal is to reach a level of automated analysis that makes tumor diagnosis faster and more reliable.
The impact of artificial intelligence on cancer care
The integration of artificial intelligence in medicine is expected to lead to a significant improvement in cancer diagnosis and treatment plans. This technology can enable early tumor detection and provide personalized treatment plans, contributing to improved survival rates and quality of life for patients. Recent research published by Stiefelhagen’s team in the journal Nature Machine Intelligence points to tremendous potential awaiting practical application in the medical field.
Collaboration between science and technology
Progress in artificial intelligence in medicine depends on interdisciplinary collaboration. The autoPET challenge is a vivid example of how medical knowledge can be combined with computer science to develop innovative solutions that make a difference in patients’ lives.
