Project »KIMPALUS«

Problem statement

Demographic change poses major challenges to our healthcare system: while treatment options are becoming more diverse, better, but also more expensive, the need for treatment in general is increasing due to the aging population, while at the same time the number of treating physicians is decreasing. In addition, the incidence of diseases is shifting; for example, cancer and its treatment are playing an increasingly important role.

These developments are particularly pronounced in rural areas. Studies have analyzed, for example, that breast cancer is detected less frequently in rural areas and the mortality rate is higher [1] and overall, the care of tumor patients in rural areas in Germany remains a major challenge [2, 3, 4].

In this area of ​​tension, it is necessary to improve patient care and at the same time relieve the burden on those treating the disease. In the KIMPALUS project, we are therefore focusing on the early detection of the three most common types of cancer: For breast, prostate and lung cancer detected early, the treatment options are significantly better, less invasive and less stressful for the healthcare system. Through the targeted use of artificial intelligence, enormous increases in efficiency are to be expected here. We therefore want to establish AI support in these early detection processes.

Due to the different levels of maturity of early detection in the above-mentioned application cases, the technologies to be developed also differ.

For breast cancer early detection, a nationwide mammography screening program has already been established; the challenge there is the integration of new AI diagnostic systems. To this end, we want to develop new concepts within the framework of KIMPALUS, taking into account the recent study situation [5, 6], and test those concepts together with clinical partners in the Brandenburg area.

For lung cancer early detection, the introduction of a systematic screening program based on low-dose CT is imminent; the use of software for computer-assisted detection will be mandatory [7]. In this project, we want to develop a proposal for criteria for quality assurance of such software systems and implement a prototype test system.

There are promising international concepts for early detection of prostate cancer [8], so that an introduction can be expected within the next few years. The challenge here is the development and validation of corresponding AI-based support systems, which require a sufficiently large and representative amount of data. These data sets are difficult to compile in a centralized manner in accordance with data protection guidelines; we therefore want to provide an infrastructure in KIMPALUS for distributed training and distributed validation of corresponding algorithms.

Contribution to the mission of the Fraunhofer-Center for Digital Diagnostics

Using the expertise of Fraunhofer Mevis, this project will unlock medical image data as a digital ecosystem for the digital solutions developed within the framework of the ZDD. With the strategic focus on early cancer detection, we are strengthening the prevention-oriented part of the ZDD. The solutions developed within the framework of KIMPALUS for the development and validation of AI algorithms for early cancer detection, methods for quality assurance of these algorithms in practical use and for supporting integration into routine practice aim to eliminate the current hurdles to the widespread use of artificial intelligence in diagnostic practice and thus to increase the efficiency of medical care, especially in rural areas.

Vision

Within KIMPALUS, we want to develop scalable technologies that, beyond the specific project focus, enable and accelerate the use of AI to interpret clinical image data in early cancer detection specifically, but also generally in clinical diagnostics. By focusing on the three application areas of prostate, lung and breast cancer, we cover three diseases with a high incidence that nevertheless have a good prognosis when diagnosed early and are therefore highly relevant in practice. On the other hand, these application areas represent different levels of maturity of AI support in early detection. In this way, we technically cover a large part of the various requirements for the introduction of AI in the respective stages, so that the technologies developed in KIMPALUS can also be used for other applications of AI for medical image data.

_{[1] Übersichtsartikel auf gis.cancer.gov 2024, bezieht sich auf mehrere Publikationen, u.a.
Anderson T et al. Geographical Variation in Social Determinants of Female Breast Cancer Mortality Across US Counties. JAMA Netw Open. 2023 Sep 5;6(9):e2333618.
doi: 10.1001/jamanetworkopen.2023.33618. PMID: 37707814; PMCID: PMC10502521.}

_{[2] Artikel „13,8 Millionen für Verbundprojekt ONCOnnect“, Netzwerk Onkologischer Spitzenzentren, 05.07.2024}

_{[3] Artikel “Onkologen fordern: Versorgungslücken schließen“, Healthcare-in-europe.com, 01.02.2022}

_{[4] Artikel “Tumorpatienten in ländlichen Regionen: Neue Versorgungskonzepte erforderlich“, Deutsches Ärzteblatt 2018; 115(7): [4];
DOI: 10.3238/PersOnko.2018.02.16.01}

_{[5] MASAI: Lång K et al. Artificial intelligence-supported screen reading versus standard double reading in the Mammography Screening with Artificial Intelligence trial (MASAI): a clinical safety analysis of a randomised, controlled, non-inferiority, single-blinded, screening accuracy study. Lancet Oncol. 2023 Aug;24(8):936-944.
doi: 10.1016/S1470-2045(23)00298-X. PMID: 37541274.}

_{[6] PRAIM: Eisemann, N et al. Nationwide real-world implementation of AI for cancer detection in population-based mammography screening. Nat Med (2025). https://doi.org/10.1038/s41591-024-03408-6}

_{[7] LuKrFrühErkV: Verordnung über die Zulässigkeit der Anwendung der Niedrigdosis-Computertomographie zur Früherkennung von Lungenkrebs bei rauchenden Personen (Lungenkrebs-Früherkennungs-Verordnung — LuKrFrühErkV);
BGBl. 2024 I Nr. 162 vom 17.05.2024}

_{[8] STHLM3: Nordström T et al. Repeated Prostate Cancer Screening Using Prostate-Specific Antigen Testing and Magnetic Resonance Imaging: A Secondary Analysis of the STHLM3-MRI Randomized Clinical Trial. JAMA Netw Open. 2024 Feb 5;7(2):e2354577.
doi: 10.1001/jamanetworkopen.2023.54577. PMID: 38324313; PMCID: PMC10851096.}