Artificial intelligence and diagnostic imaging: Humanitas present at the 2020 meeting of EuSoMii
On October 24th the annual meeting of EuSoMII (European Society of Medical Imaging Informatics), the non-profit health organization that connects data scientists, computer scientists and specialists in radiology and other disciplines where the use of diagnostic imaging is crucial, will be held online.
Humanitas will participate in the convention with two task force projects, in which the use of Artificial Intelligence is aimed at improving the clinical performance in the field of pancreas surgery and the isolation of patients with Covid-19 according to the severity of symptoms.
Covid-19: safeguarding patients with Computed Tomography
Thanks to the resources put in place by Humanitas’ specialists in collaboration with CERTH (Centre for Research & Technology, Hellas), an algorithm has been developed that is able to accelerate, thanks to the use of Artificial Intelligence, the diagnosis process of patients with Covid-19, identifying at great speed those who require hospitalization in intensive care.
Patient data is analyzed through Artificial Intelligence in order to improve the hospital’s internal resources and combat the spread of the virus, ensuring the adequate isolation of patients according to the severity of their state of health.
By analyzing the images extracted using computed tomography (CT) it is possible, indeed, to promptly subdivide patients with Covid-19 into three groups, from group 1, composed of subjects who can be discharged and continue isolation at home, to group 3, who require immediate hospitalization in intensive care.
The algorithm was developed by a team of physicians and data scientists from Humanitas and CERTH, analyzing the CT results of patients admitted to Humanitas with symptoms confirmed by Sars-Cov2 between March and May 2020. Artificial Intelligence, in the event of a further increase in Covid-19 cases and a high rate of virus transmission, is therefore a valuable support to speed up and improve the therapeutic journey of patients.
Applications of Artificial Intelligence in pancreatic surgery
In the field of pancreatic surgery, pancreatic fistula (PF) is the most common post-operative complication associated with various risk factors including pancreatic morphology and sarcopenic obesity (SO).
Thanks to a research study involving specialists from the Diagnostic Unit, the Pancreatic Surgery Unit and the AI Center of Humanitas, a risk model has been created which is able to predict in advance the probability of occurrence of pancreatic fistula in patients undergoing pancreatic surgery, through the use of Artificial Intelligence.
The research involved the analysis of clinical and diagnostic data of one hundred patients undergoing pancreatic surgery between 2010 and 2020 and the developed algorithm showed a strong predictive value of negative cases. The result is therefore particularly important: the possibility to identify in the preoperative phase, the subjects for whom the risk of development of pancreatic fistula is lower, can help specialists to significantly optimize the clinical organization at the benefit of the patient.