DoMore! is a five-year-long project within cancer research, developing new diagnostic methods based on artificial intelligence. The purpose has been to utilise new technology to reduce over-and undertreatment of cancer by providing a new system for cancer prognosis. The lack of objective and precise methods for cancer prognosis is the most crucial cause behind the prevalence of over-and under treatment of cancer patients worldwide.
Explore the results of the DoMore!-project: Results
The DoMore! project is split into 7 work packages defining the steps necessary for completion of the project. Each Work package (WP) has a dedicated team reporting to the project administration. This approach has contributed to the projects success.
Production of images that will be used in methods development (WP-2-6) and in the test and validation studies in WP7.
Read more about Work Package 1 here >>
Fully automatisation of the analysis pipeline for the use of deep learning and big data to develop new grading systems for colorectal and lung cancer.
See details of the Histotyping Work Package
Developing reliable, fully automatic methods for tumor delineation and nuclear segmentation in tissue sections.
Read more about the Work Package here >>
Finding biomarkers based on analysis of nuclei in tissue sections. Further developed research tools for practical use in routine pathology.
Establishing solutions that ensure efficient data flow of quality-controlled images.
Developing a decision support system for prognostication of cancer patients using machine learning techniques applied to large datasets. Health economy and commercialisation.
Link to a page describing the WP
Test and validation studies, scientific publications and dissemination.
The DoMore! project was in 2016 selected as one of the Norwegian Research Council’s Lighthouse projects aiming to solve large societal challenges using cutting-edge technology. Our team of national and international experts work within many different fields, including digital image analysis, tumor pathology, cancer surgery, and oncology.
Heterogeneity is one of the key research interests for the DoMore!-project.
Why do more? As the amount of cancer patients rises, we need more advanced tools to give patients more reliable diagnostics and prognostics.
Using deep convolutional neural networks has enabled more accurate prognostication of cancer.
We need better tools to accurately assess the patient's outcome. Many cancer patients receive more treatment than they need. Overtreated patients are susceptible to severe side effects and reduced quality of life. This is a burden to patients and expensive to society. The shortage of pathologists only adds to the increasing difficulty to keep up with patient demand. The manual laboratory procedures needed to render a prognosis are time-consuming and subjective. We are using digital tools, to increase efficiency in pathology and to provide a more accurate prognosis to cancer patients. This will make the process behind the choice of treatment faster and safer.
We need better tools to accurately assess the patient's outcome. Many cancer patients receive more treatment than they need.
Overtreated patients are susceptible to severe side effects and reduced quality of life. This is a burden to patients and expensive to society.
The shortage of pathologists only adds to the increasing difficulty to keep up with patient demand. The manual laboratory procedures needed to render a prognosis are time-consuming and subjective. We are using digital tools, to increase efficiency in pathology and to provide a more accurate prognosis to cancer patients. This will make the process behind the choice of treatment faster and safer.
The interactive graphics linked above show prostate cancer in 3D and demonstrate heterogeneity as it is observed in sections.
A tumor can contain several regions of different and unevenly distributed aberrations that may or may not lead to cancer. This is called heterogeneity, and is a great challenge for prognosis of cancer. Prognostication and sampling are also challenges being explored in the DoMore! project.
We don't know enough about how to divide between a dangerous and less dangerous tumor. Cancers can follow multiple paths. Predicting the further development of the disease is complicated. Some will progress to metastases and death while others will remain indolent.
The pathologists who are currently doing the work need several years of specialisation. Health systems around the world experience a major shortage of this specialist group. Today´s prognostication is based on subjective assessments. If the conclusion is incorrect, it may cause additional costs, unnecessary side effects and, at worst, the death of the patient.
A single tumour can contain multiple regions of different and unevenly distributed aberrations that might or might not lead to cancer.
We seek a broader understanding of cancer heterogeneity in the enormous reserves of patient data stored in hospital data basis. Patterns in this data can tell us a lot about how a patient's cancer develops.
Sampling for most markers of cancer is performed on a too small fraction of a single block of tissue. Sample size plays a significant role in accurate cancer prognostics, especially when we take cancer heterogeneity into account.
Using prostate cancer as an example, the sample examined is a mere 1:1000 of a tumour. This small sample size means there is a higher risk of missing the cells that go on to kill the patient. Grading the sample and giving a correct prognosis depends heavily on the pathologist’s expertise. Pathologists assessing the same sample can arrive at vastly different results.
We have a team of national and international experts within many different fields, including digital image analysis, tumor pathology, cancer surgery and oncology.
The DoMore! project is funded by the Norwegian Research Council's ICT Lighthouse project grant