Designing deep learning studies in cancer diagnostics
If artificial intelligence-based technology for cancer diagnostics exists, why is it still not in use? How can new knowledge in deep learning and artificial intelligence in diagnostics benefit cancer patients in the fastest and safest way?
Researchers at our institute have authored a perspective advocating performance estimation in external cohorts. They strongly advise that a primary analysis is predefined in a standardized protocol, preferentially stored in an online repository. They also recommend more careful and rigorous research to facilitate the successful use of AI in the clinic.
Read the article here
The article has received attention worldwide, and a selection of citations is listed below.
Podcast about Artificial Intelligence and Medicine
Listen to our Institute Director Håvard E. Danielsen explaining Domore! and the use of artificial intelligence in cancer prognostication and diagnostics.
NORA (Norwegian Artificial Intelligence Research Consortium) aims to strengthen Norwegian research, education and innovation within artificial intelligence, machine learning and robotics, as well as other relevant research that supports the development of artificial intelligence applications.
Listen to their podcast here.
ICGI's video presentation
Watch the video made by our colleagues at the section for Dissemination and Visualisation, presenting the Institute for Cancer Genetics and Informatics
ICGI Strategy Document for the upcoming years
Congratulations to Karolina Cyll
Karolina Cyll came to the ICGI as a MSc student in 2011, where her project entailed reviewing and increasing the efficiency of the ploidy preparation lab procedure.
In December 2020 she defended her thesis for the PhD degree, titled “High-resolution, high-throughput nuclear analysis as a prognostic marker in prostate cancer.” The study was dedicated to finding prognostic markers for patients with early and intermediate stage of prostate cancer. Investigated material comprised of formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomies as well as biopsy materials from patients under surveillance. The main focus was genomic instability, resistance to cell death and immortality of cancer cells. The employed methods were DNA ploidy, nucleotyping, immunohistochemistry and FISH.
We are happy Karolina will continue at our Institute as Post.doc
Andreas Kleppe appointed Associate Professor at University of Oslo
We are so proud of the institutes' many young researchers and pleased to see everything they achieve. Congratulations to Post Doc Andreas Kleppe on his position as Associate Professor at the University of Oslo, Department of Informatics.
The position will involve teaching in image processing, image analysis and deep learning, as well as supervision of master's degree students and doctoral students. Andreas defended his dissertation in 2017 and was the year after the first author of the widely acclaimed article Chromatin organization and cancer prognosis: a pan-cancer study https://www.thelancet.com/.../PIIS1470-2045(17.../fulltext. We wish Andreas good luck with his new, exciting tasks!
The amount of stroma - a new prognostic marker for cancer?
Solid tumours contain many different cellular components in addition to tumour cells. The stroma is the supportive framework of an organ, usually composed of connective tissue, distinguished from the cells or tissues performing the special function of the organ. The manual assessment of the amount of reactive stroma has been shown, by us and others, to be a prognostic marker in both colorectal cancer and prostate cancer. This video describes how to perform an automated analysis of tumour stroma content using routine diagnostic H&E stained tissue sections. Stroma can be combined with DNA ploidy to identify patients at increased risk for poor outcome. The DNA Ploidy and Stroma biomarker are based on a research collaboration between ICGI and professor David Kerr and the University of Oxford.
AI benefits colorectal cancer patients
Professor David Kerr at Oxford University talks about the possible benefits for cancer patients, following the article "Deep learning for prediction of colorectal cancer outcome: a discovery and validation study" , published in "The Lancet" on February 1st 2020.
Marco Novelli on the ICGI Lancet article
Our research colleague based at the University of London, Professor Marco Novelli, talks about his excitement for the results presented in the Lancet article published in February 2020. The article was authored by researchers from ICGI, University of Oslo and colleagues in the UK.
ICGI's Lancet article presented at Oslo Life Science 2020
The ICGI recently developed a clinically useful prognostic marker using deep learning and digital scanning of conventional hematoxylin and eosin-stained tumor tissue sections. The assay has been extensively evaluated in large, independent patient populations, correlates with and outperforms established molecular and morphological prognostic markers, and gives consistent results across tumor and nodal stage.
The article "Deep learning for prediction of colorectal cancer outcome: a discovery and validation study" was published in "The Lancet" on February 1st 2020, and has since been presented on several occasions. One of them was at the conference "Oslo Life Science 2020", held in the amazing "Aula" of, and by, the University of Oslo, where among others the newly appointed Director of Oslo University Hospital; Bjørn-Atle Bjørnbeth, attended. He congratulated Håvard Danielsen, Ole-Johan Skrede, Sepp De Raedt and the other co-authors with their excellent achievement.
Please click here to download and read the article in The Lancet (free).
We call the method Histotyping or the CRC-v1-Network, to learn more about the Histotyping project, click here.