Thesis on the potential of artificial intelligence in accurate diagnosis of tumor tissue

Here various technologies ranging from machine learning, image analysis and statistics are used to help pathologists or other doctors with disease diagnosis and prognosis. Developing methods in computational pathology has two main goals; Automating routine flows that would otherwise be performed by pathologists and developing methods to extract new information from tissue images. In his thesis he focuses Philip WeitzPhD student in Department of Medical Epidemiology and Biostatisticsabout the development, application and evaluation of new methods for accurate diagnosis.

What did you research in your thesis?

The first two studies in my dissertation focus on developing and evaluating methods for predicting gene expression from histopathological images. In my other three studies, we focus on recording pathological images. One of my posts describes the ACROBAT dataset, which we published for a histopathology image analysis competition.

Why did you become interested in the topic?

– Advances in large-scale data analysis methods and the availability of large data sets have allowed the development of models in machine learning and artificial intelligence for medical diagnosis in the context of research. I believe that these approaches have the potential to improve current clinical practice, and I want to contribute to this development. It is very motivating for me that the results of my research can benefit patients.

What do you think should be done in future research?

– Going forward, we must conduct more and more validation studies of the methods and solutions developed by us and others. Currently, many studies rely on in-house test datasets to evaluate algorithm performance, providing insufficient support for broader adoption of these solutions in clinical pathology.

thesis

“Artificial Intelligence in Histopathological Image Analysis for Precision Cancer Medicine”

Philip Weitz. Karolinska Institutet (2023), ISBN: 978-91-8017-148-9