Artificial intelligence, game-based learning, and DNA-profiling to differentiate benign from malignant ovarian tumours, and to predict the prognosis in women with ovarian cancer.

The overall aim of this study is to find new strategies to differentiate benign from malignant ovarian tumours, and to predict the long-time prognosis in women with ovarian cancer - through artificial intelligence (AI), computerized image analysis (CIA), deep/convoluted neural networks (DNN/CNN), game-based learning, and the analysis of circulating tumour DNA (ctDNA) and tumour immune-profile.

We aim to;1/ Develop mobile app SonoQZ with a game-based training program, including several hundred ultrasound cases of ovarian tumours.

We will evaluate if this tool improves the ability of the examiner to discriminate benign from malignant ovarian tumours.2/ Explore, validate and prospectively test if computerized image analysis/machine-learning using DNN can discriminate benign form malignant ovarian tumours and thus be used in the triage of women with ovarian lesions.3/ Create a translational, personalized medicine platform including high-resolution ultrasound images and biological specimens (blood, tumour tissue) from women with ovarian tumours.

One project will focus on creating multimodal approaches through artificial neural networks including computerized image analysis, ctDNA and immunological profile to discriminate benign from malignant ovarian tumours, to predict prognosis and identify new pathways for individualized treatment in women with ovarian cancer.