Grant to 3 Artificial Intelligence Projects for Rare Cancer

Projects use artificial intelligence methods to detect, diagnose and treat rare cancers

The Hanarth Fund awards grants to three Radboudumc projects. The projects use artificial intelligence methods to detect, diagnose and treat rare cancers.

Knowledge in oncology is expanding rapidly with constantly new treatments based on advances in immunology, genetics, and systems biology. This leads to an “explosion” in electronic patient records. Additionally, there is a demand for personalized cancer treatment. Each patient presents a huge data challenge, with vast amounts of information about current status and past pathways. All of this makes medical decision-making increasingly complex, and decisions are often not the best solution. Proper use of AI can often improve the detection, diagnosis, and treatment of rare cancers.

Granted projects:

Detecting invisible cancer: digital detection of metastatic gastric cancers – Sheila van der Post; Francesco Ciompic
This project will develop artificial intelligence methods to improve the diagnosis of metastatic gastric cancer (DGC). Since metastatic gastric cancer can easily be missed or difficult to find in biopsies, AI will aid the pathologist with the diagnostic examination. This improves the discovery of relevant cell types, with great potential for improving cancer prognosis. In addition, automation of cell detection by AI algorithms will allow quantitative and objective evaluation of DGC patterns in a large series of slides, which may lead to new insights into specific morphological features of DGC, such as spatial cell distribution patterns. Through this project, researchers aim to use artificial intelligence to better identify and categorize prospective (H)DGC patients in order to increase the discovery of individual patients and families. This may eventually result in better stratification of the patient for therapeutic options and clinical decisions.

Improving transparent AI methods for personalized prediction based on rare cancer data – Marian Juncker; Kate Royce Carla Van Herpen Tone Colen
The prognosis for patients with rare cancers is worse than that of patients with common cancers. This is partly because it is difficult to identify possible treatments for rare types of cancer, because usually only small amounts of patient data are available for research. Pooling data from different organizations can alleviate the situation, but this is challenging in practice due to organizational and logistical issues.

Two complementary pathways have been proposed to address the challenges of small data sets for rare cancers. The first is to focus on more powerful inference techniques that can better handle a small sample size. The second path is to design and improve machine learning algorithms that avoid the need to aggregate data in one place by “cycling” around medical institutions with small data sets (consolidated learning). Salivary gland cancer data will be analyzed by the suggested methods.

Multimodal data integration to guide the development of immunotherapies for ocular melanoma patients. Johannes Textor
In the Netherlands, melanoma of the eye occurs in about 200 patients annually. Up to half of patients with primary ocular melanoma develop metastases. Ocular melanoma is clinically and genetically distinct from melanoma and the response to immunotherapy is low. In this project, images and data from multiple groups of patients are analyzed to investigate tumor immune interactions. Data mining risks (a problem with data mining when random variations in data are categorized as master patterns) and potential data merging problems should be avoided as much as possible. To do this, an analysis approach is first developed and optimized using simulated data that is similar to real data. Ultimately, we want to gain a mechanistic understanding of immune processes, which should help improve treatment of uveal melanoma patients in the future.”

Hanarth Fonds
Hanarth Fund was established on September 28, 2018 and is co-owned by Arthur del Prado, founder and former CEO of ASM International. The fund aims to promote and improve the use of artificial intelligence and machine learning to improve diagnosis, treatment, and outcomes for cancer patients. In this context, the Hanarth Fund supports scientific research that focuses specifically, but not exclusively, on rare forms of cancer. Potential applicants are clinicians or scientists with a background in oncology with an interest in artificial intelligence and machine learning along with partners(s) who have a background in artificial intelligence, machine learning and an interest in oncology.