Developing methods using artificial intelligence (AI) to give cancer patients a more precise prognosis and counteract overtreatment.

With the success of the automated methods and better prognostic markers developed within the DoMore!-project in the period 2016 - 2021, we have been able to and adequately address the challenge of cancer heterogeneity and to “DoMore!” with the same amount of resources. Better prognostic markers will result in more personalised medicine with less over- and under-treatment. For patients to benefit from our research, we have established the company DoMore Diagnostics AS and provided the Norwegian eHealth Company DIPS with the necessary tools to integrate our products.

We are proud to have contributed to the digitalisation of pathology and paved the way for the transition from digital pathology to in silico pathology, in addition to introducing AI and Deep Learning into tissue diagnostics. This transition to in silico pathology will not only change pathology as we know it, but also compensate for the shortage of pathologists and the uneven distribution of best practices. In this way, we fulfilled the ultimate goal as a Lighthouse project, to shine a light on the way forward for further knowledge, development and value creation for the sake of the Norwegian healthcare system and, most importantly, the patient.

In 2016, following a comprehensive multistep-evaluation by national and international experts in ICT and E-Health, the Research Council of Norway selected DoMore! as one of three projects to receive funding as part of its ambitious Lighthouse Project scheme. As the name suggests, Lighthouse Projects were intended to be beacons, guiding and inspiring future projects to address large societal challenges using cutting-edge technology.

Projects were funded by IKTPLUSS as part of an initiative with BIA and HELSEVEL, which called for the development of new innovative ICT-based products and solutions in response to key societal challenges to be deployed in Norway’s public and private sectors. Assessed on the basis of Excellence, Impact, and Implementation, projects were expected to have long-term goals and challenge-driven aims to create value in health, healthcare, and welfare. The DoMore! project was awarded 60 million Norwegian Kroner on the basis of its fulfilment of these criteria.


Background and challenges to address

Cancer poses enormous personal, clinical and societal challenges, and the number of global annual cancer cases keeps increasing at a high rate. Successful cancer treatment relies on a correct diagnosis, preferably
at an early stage of the disease. However, cancers are heterogeneous and can follow multiple paths, not all of which progress to metastases and death; some cancers are indolent, causing little or no harm during the patient’s lifetime. The ability to assess the disease’s likely outcome is equally important, and the importance of prognostic methods and markers cannot be overstated.

Before starting primary treatment, a patient’s disease is clinically staged, a process to determine the extent to which the cancer has progressed, and graded, assessing the aggressiveness of the disease and the patient’s likely outcome. The prognostic value of histological grading varies both by cancer type and between pathologists, as this is a subjective process. Grading depends heavily on the pathologist’s expertise, and inter-observer and
intra-observer agreements are moderate only. The grading performed by specialist pathologists working at a tertiary centre is prognostically better than grading by general pathologists. The inter-observer agreement among specialist pathologists is better than that of general pathologists, so a patient’s care can also be impacted by
their location and ability to access higher-tier hospitals. Ultimately, patients pay the price through under- or overtreatment.

The variability of grading systems’ abilities to prognosticate, and the subjectivity in using these systems, pose significant challenges for accurate diagnosis and prognosis. Furthermore, pathologists are scarce, and the resources in the health service are limited, necessitating efficiency, productivity, and quality improvements
without using more resources. To adequately address the heterogeneity issue, one needs to analyse more samples, DoMore!, with the same number of resources. As pathology has demonstrated, critical prognostic information may be deduced from a tumour’s growth pattern through subjective comparisons of such patterns
and patients outcome. By applying tools using AI and Deep Learning on Big Data, DoMore! sought to establish more robust grading systems, developing generic and objective digital prognostic markers for prostate, colon, rectum, and lung cancers.


All the materials used in DoMore! were collected from surgically resected tumours from the prostate (Pca), colorectal (CRC), and lung cancer patients provided by our collaborators (see figure). These three cancer types were selected for the degree to which they represent the wide-ranging spectrum of cancer in heterogeneity, aggressiveness, and incidence. The use of machine learning and deep learning demands “big data”, and as digital pathology is in an early phase, most of the “big data” used in DoMore! needed to be produced within the project. Initially, we planned to include 18 162 tumour blocks from a total of 7055 patients divided between the three cancer types. Devoting roughly 1/3 of the total budget to data production along with the increased interest from new collaborators during the project, we were able to do more than initially planned, with a total inclusion of
12 370 patients and digitalisation of 81 173 hematoxylin and eosin (H&E) stained sections (virtual slides).



Under the guidance of its Director, Professor Håvard E. G. Danielsen, ICGI initiated the DoMore! Project in 2016 with an interdisciplinary consortium of international experts in digital image analysis, tumour pathology, cancer surgery, and oncology. We built a strong leadership structure with a steering committee and executive board that oversaw DoMore! progress, managing both risks and innovation over its five years of activity. We carefully selected partners based on their unique competence and perspectives to ensure that we had the best possible
chance to achieve our goals. Two-thirds of the partners work directly and daily with the issues we addressed in this project.The other third are companies working with and investing in health services. By partnering with key opinion leaders within their fields, we strove to ensure the knowledge was converted into clinical application.

Page visits: 342