Artificial intelligence and radiation effects on brain tissue in glioblastoma patient: preliminary data using a quantitative tool

Donatella Franco; Vincenza Granata; Roberta Fusco; Roberta Grassi; Valerio Nardone; Laura Lombardi; Salvatore Cappabianca; Renata Conforti; Francesco Briganti; Roberto Grassi; Ferdinando Caranci

doi:10.1007/s11547-023-01655-0

Artificial intelligence and radiation effects on brain tissue in glioblastoma patient: preliminary data using a quantitative tool

Radiol Med. 2023 Jul;128(7):813-827. doi: 10.1007/s11547-023-01655-0. Epub 2023 Jun 8.

Affiliations

¹ Division of Radiology, Department of Precision Medicine, "Università degli Studi della Campania Luigi Vanvitelli", Naples, Italy.
² Division of Radiology, "Istituto Nazionale Tumori IRCCS Fondazione Pascale - IRCCS di Napoli", Naples, Italy. v.granata@istitutotumori.na.it.
³ Research & Development and Medical Oncology Division, Igea SpA, Naples, Italy.
⁴ Italian Society of Medical and Interventional Radiology (SIRM), SIRM Foundation, Via della Signora 2, 20122, Milan, Italy.
⁵ Advanced Biomedical Sciences Department, Federico II University, Naples, Italy.

PMID: 37289266
DOI: 10.1007/s11547-023-01655-0

Abstract

Purpose: The quantification of radiotherapy (RT)-induced functional and morphological brain alterations is fundamental to guide therapeutic decisions in patients with brain tumors. The magnetic resonance imaging (MRI) allows to define structural RT-brain changes, but it is unable to evaluate early injuries and to objectively quantify the volume tissue loss. Artificial intelligence (AI) tools extract accurate measurements that permit an objective brain different region quantification. In this study, we assessed the consistency between an AI software (Quibim Precision^® 2.9) and qualitative neruroradiologist evaluation, and its ability to quantify the brain tissue changes during RT treatment in patients with glioblastoma multiforme (GBM).

Methods: GBM patients treated with RT and subjected to MRI assessment were enrolled. Each patient, pre- and post-RT, undergoes to a qualitative evaluation with global cerebral atrophy (GCA) and medial temporal lobe atrophy (MTA) and a quantitative assessment with Quibim Brain screening and hippocampal atrophy and asymmetry modules on 19 extracted brain structures features.

Results: A statistically significant strong negative association between the percentage value of the left temporal lobe and the GCA score and the left temporal lobe and the MTA score was found, while a moderate negative association between the percentage value of the right hippocampus and the GCA score and the right hippocampus and the MTA score was assessed. A statistically significant strong positive association between the CSF percentage value and the GCA score and a moderate positive association between the CSF percentage value and the MTA score was found. Finally, quantitative feature values showed that the percentage value of the cerebro-spinal fluid (CSF) statistically differences between pre- and post-RT.

Conclusions: AI tools can support a correct evaluation of RT-induced brain injuries, allowing an objective and earlier assessment of the brain tissue modifications.

Keywords: Artificial intelligence; Brain tissue changes; Glioblastoma; Radiotherapy.

MeSH terms

Artificial Intelligence
Atrophy / pathology
Brain / diagnostic imaging
Brain / pathology
Glioblastoma* / diagnostic imaging
Glioblastoma* / pathology
Glioblastoma* / radiotherapy
Humans
Magnetic Resonance Imaging / methods
Preliminary Data
Radiation Injuries* / diagnostic imaging
Radiation Injuries* / pathology