Artificial Intelligence Decision Tools for Tumour diagnosis

Spain trails far behind other countries in terms of translational and clinical initiatives for the development of technologies to assist the diagnosis and treatment of cancer patients. On the other hand, there is wide consensus on that the development, practical implementation, and usage of decision support systems by medical experts and clinicians holds promise for substantially improving the quality of medical practice, with the current agenda for personalized medicine.
For these systems to be useful, they must be based on techniques that behave robustly in a context pervaded with uncertainty.
The first goal of the AIDTumour project is the design and development of robust Soft Computing techniques for the reduction of the uncertainty that is inherent in medical diagnosis and prognosis in the field of clinical oncology. Novel Soft Computing techniques for data clustering, classification, feature selection (both for clustering and classification), graphical modelling, and rule extraction (both from supervised and unsupervised models), will be developed. Furthermore, novel latent models-based visualization techniques will be studied and implemented.
The second goal is the integration of these tools within a web-accesible clinical oncology decision support system, including flexible visualization methods integrated within a virtual reality environment.
We use original data supplied by the Grup d'Aplicacions Biomèdiques de l'Espectroscòpia de Ressonància Magnètica Nuclear (GABRMN) from Universitat Autònoma de Barcelona (UAB) to accomplished the aforementioned research goals. The available data consist of in vivo spectra obtained by nuclear magnetic resonance (a non-invasive technique), corresponding to diverse types and grades of human brain tumours.
The developed tool prototype will be evaluated and validated by several collaborating hospitals (both national and international) specialized in oncology.