The analysis of medical image data can be improved by artificial intelligence (AI) algorithms, latest research has claimed.
A new study presented at the international competition AutoPET revealed that deep learning algorithms can better detect the location and size of tumours.
Positron emission tomography (PET) and computer tomography (CT) imaging techniques can successfully identify cancer, experts have said.
- Patients prefer ChatGPT responses over doctors’ responses
- W8Buddy app paves the way for enhanced weight management in the NHS
- Earlier type 2 diabetes diagnosis linked to increased dementia risk
PET scans produce detailed 3-dimensional images of the inside of the body by using radionuclides to visualise metabolic processes.
The images can clearly show the part of the body being investigated, including any abnormal areas, and can highlight how well certain functions of the body are working.
A CT scan is a test that also takes detailed pictures of the inside of your body layer by layer using an X-ray tube to visualise the anatomy and localise tumours. It’s usually used to diagnose conditions or check how well treatment is working.
PET scans are often combined with CT scans to produce even more detailed images. This is known as a PET-CT scan.
To determine the size of a cancerous tumour lesions, doctors have to manually marking 2D slice images – a task which is extremely time-consuming.
Professor Rainer Stiefelhagen said: “Automated evaluation using an algorithm would save an enormous amount of time and improve the results.”
- Study highlights problems associated with AI
- Natural weight loss drug as effective as Ozempic and without side effects, academics claim
- Brain function enhanced by good sleep and exercise
Researchers from Karlsruher Institut für Technologie have found that the automated analysis of medical image data could transform cancer care.
The authors said: “An ensemble of the top-rated algorithms proved to be superior to individual algorithms. The ensemble of algorithms is able to detect tumour lesions efficiently and precisely.”
Professor Stiefelhagen concluded: “While the performance of the algorithms in image data evaluation partly depends indeed on the quantity and quality of the data, the algorithm design is another crucial factor, for example with regard to the decisions made in the post-processing of the predicted segmentation.”
The study has been published in the journal Nature Machine Intelligence.