Artificial intelligence applications in medical imaging: A review of the medical physics research in Italy.

PURPOSE: To perform a systematic review on the research on the application of artificial intelligence (AI) to imaging published in Italy and identify its fields of application, methods and results. MATERIALS AND METHODS: A Pubmed search was conducted using terms Artificial Intelligence, Machine Learning, Deep learning, imaging, and Italy as affiliation, excluding reviews and papers outside time interval 2015-2020. In a second phase, participants of the working group AI4MP on Artificial Intelligence of the Italian Association of Physics in Medicine (AIFM) searched for papers on AI in imaging. RESULTS: The Pubmed search produced 794 results. 168 studies were selected, of which 122 were from Pubmed search and 46 from the working group. The most used imaging modality was MRI (44%) followed by CT(12%) ad radiography/mammography (11%). The most common clinical indication were neurological diseases (29%) and diagnosis of cancer (25%). Classification was the most common task for AI (57%) followed by segmentation (16%). 65% of studies used machine learning and 35% used deep learning. We observed a rapid increase of research in Italy on artificial intelligence in the last 5 years, peaking at 155% from 2018 to 2019. CONCLUSIONS: We are witnessing an unprecedented interest in AI applied to imaging in Italy, in a diversity of fields and imaging techniques. Further initiatives are needed to build common frameworks and databases, collaborations among different types of institutions, and guidelines for research on AI.

Dosimetric accuracy of tomotherapy dose calculation in thorax lesions.

BACKGROUND: To analyse limits and capabilities in dose calculation of collapsed-cone-convolution (CCC) algorithm implemented in helical tomotherapy (HT) treatment planning system for thorax lesions. METHODS: The agreement between measured and calculated dose was verified both in homogeneous (Cheese Phantom) and in a custom-made inhomogeneous phantom. The inhomogeneous phantom was employed to mimic a patient's thorax region with lung density encountered in extreme cases and acrylic inserts of various dimensions and positions inside the lung cavity. For both phantoms, different lung treatment plans (single or multiple metastases and targets in the mediastinum) using HT technique were simulated and verified. Point and planar dose measurements, both with radiographic extended-dose-range (EDR2) and radiochromic external-beam-therapy (EBT2) films, were performed. Absolute point dose measurements, dose profile comparisons and quantitative analysis of gamma function distributions were analyzed. RESULTS: An excellent agreement between measured and calculated dose distributions was found in homogeneous media, both for point and planar dose measurements. Absolute dose deviations <3% were found for all considered measurement points, both inside the PTV and in critical structures. Very good results were also found for planar dose distribution comparisons, where at least 96% of all points satisfied the gamma acceptance criteria (3%-3 mm), both for EDR2 and for EBT2 films. Acceptable results were also reported for the inhomogeneous phantom. Similar point dose deviations were found with slightly worse agreement for the planar dose distribution comparison: 96% of all points passed the gamma analysis test with acceptable levels of 4%-4 mm and 5%-4 mm, for EDR2 and EBT2 films respectively. Lower accuracy was observed in high dose/low density regions, where CCC seems to overestimate the measured dose around 4-5%. CONCLUSIONS: Very acceptable accuracy was found for complex lung treatment plans calculated with CCC algorithm implemented in the tomotherapy TPS even in the heterogeneous phantom with very low lung-density.