Skin cancers pose significant diagnostic challenges, particularly in telemedicine where physical examination cannot be performed. Dermatologists are compelled to practise defensive medicine due to the fear of overlooking potentially deadly melanomas. This defensive approach results in costly overtreatment of benign lesions and dangerous delays in diagnosing cancers. SynTIA seeks to develop an artificial intelligence (AI) product that provides synergistic, trustworthy and interpretable diagnostic feedback for tele-dermatologists. SynTIA will guide the dermatologist towards a cost-effective and safe treatment strategy. More specifically, SynTIA will suggest the likely diagnoses, accompanied by the reasoning and confidence behind this analysis. Additionally, SynTIA will show the tumor's potential evolution over the next 3-6 months if left untreated. Finally, SynTIA will offer an analysis of the clinician's bias and likelihood of misdiagnosis, supported by examples of previous similar cases. The development of SynTIA will rely on data from the AISC-database, which is a large repository of skin lesion images and over 400,000 annotations from the educational Dermloop Learn application. We will use both established and cutting-edge machine learning methodologies, such as Concept Bottleneck and Diffusion Models, to achieve the desired outcomes. The resulting solution will be commercialized within Dermloop, an existing and rapidly scaling telemedical platform for skin diagnostics.

Projektet har til formål at forbedre forædlingen af byg (Hordeum vulgare L.) ved at videreudvikle CRISPR-Cas9 genredigering og adressere udfordringerne med regenerationsrecalcitranse, som forhindrer anvendelsen i kommercielle sorter. Transformationseffektiviteten kan øges ved at modificere Agrobacterium tumefaciens med et Type III Sekretionssystem, samtidig med at regenerationsprocessen forbedres. Dette kan opnås ved anvendelsen af syntetiseret REF1-peptid til at inducere et regenerationssignal.