Abstract Details

Title Artificial Intelligence Applied on Conventional Magnetic Resonance Images Improves the Correct Diagnosis of CNS Diseases Mimicking Multiple Sclerosis

Topic Multiple Sclerosis

Presentation(s) S45 - Multiple Sclerosis: Imaging (4:54 PM-5:06 PM)

Poster/Presentation
Number 008

Objective To present a deep-learning approach for the automated classification of different CNS diseases presenting with brain white matter (WM) hyperintense lesions and to compare model performance with that of two expert neuroradiologists blinded to diagnosis.

Background The diagnostic work-up of patients with suspected multiple sclerosis (MS) may be challenging due to the frequency of WM hyperintensities in a variety of neurological conditions. Using MRI, a great effort is ongoing to allow an automatic and reliable diagnosis for patients with different diseases mimicking MS. The accuracy of deep-learning-based imaging diagnostics is reaching or is going beyond the level of clinical experts by using information directly driven by raw or minimally processed data.

Design/Methods 268 brain T1-weighted and T2-weighted MRI scans, acquired on a 1.5T and 3T MR scanners, were collected from patients with migraine (n=56), MS (n=70), neuromyelitis optica spectrum disorders (NMOSD) (n=91) and CNS vasculitis (n=51). The model architecture, trained on 178 images, was based on a cascade of four 3D convolutional neural network layers followed by a fully dense layer after features extraction. The ability of the final algorithm to correctly classify the diseases in an independent test set of 90 MRI was compared with that of 2 expert neuroradiologists.

Results

In the test set, the deep learning algorithm showed higher classification accuracy (92.2% vs 59% for migraine, 98.8% vs 78% for MS, 88.6% vs 4.4% for NMOSD, 92.1% vs 51% for vasculitis) and higher specificity (97.1% vs 88.4% for migraine, 98.4% vs 75.5% for MS, 92.9% vs 92% for NMOSD, 93.2% vs 72.6% for vasculitis) compared with the two neuroradiologists (p=0.01). The inter-rater agreement was 84.9% (Cohen’s kappa=0.78, p<0.001).

Conclusions

The classification performance of the deep-learning algorithm exceeded that of 2 expert neuroradiologists, suggesting that artificial intelligence may be a powerful paraclinical tool in the diagnostic work-up of diseases mimicking MS.

Authors/Disclosures
Massimo Filippi, MD, FAAN (Ospedale San Raffaele, Neuroimaging Research Unit) PRESENTER	Dr. Filippi has received personal compensation in the range of $5,000-$9,999 for serving as a Consultant for Alexion, Almirall, Biogen, Merck, Novartis, Roche, Sanofi. Dr. Filippi has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Alexion, Biogen, Bristol-Myers Squibb, Merck, Novartis, Roche, Sanofi, Sanofi-Aventis, Sanofi-Genzyme, Takeda. Dr. Filippi has received personal compensation in the range of $500-$4,999 for serving on a Speakers Bureau for Bayer, Biogen, Celgene, Chiesi Italia SpA, Eli Lilly, Genzyme, Janssen, Merck-Serono, Neopharmed Gentili, Novartis, Novo Nordisk, Roche, Sanofi, Takeda, and TEVA. Dr. Filippi has received personal compensation in the range of $10,000-$49,999 for serving as an Editor, Associate Editor, or Editorial Advisory Board Member for Springer Nature. The institution of Dr. Filippi has received research support from Biogen Idec, Merck-Serono, Novartis, Roche, the Italian Ministry of Health, the Italian Ministry of University and Research, and Fondazione Italiana Sclerosi Multipla.
Loredana Storelli	Loredana Storelli has nothing to disclose.
	No disclosure on file
	No disclosure on file
Paolo Preziosa (Ospedale San Raffaele)	Mr. Preziosa has received personal compensation in the range of $500-$4,999 for serving on a Speakers Bureau for Bristol Myers Squibb . Mr. Preziosa has received personal compensation in the range of $500-$4,999 for serving on a Speakers Bureau for Sanofi Genzyme. Mr. Preziosa has received personal compensation in the range of $500-$4,999 for serving on a Speakers Bureau for Novartis. Mr. Preziosa has received personal compensation in the range of $500-$4,999 for serving on a Speakers Bureau for Roche. Mr. Preziosa has received personal compensation in the range of $500-$4,999 for serving on a Speakers Bureau for Merck.
Laura Cacciaguerra, MD, PhD (Mayo Clinic)	Dr. Cacciaguerra has nothing to disclose.
Maria A. Rocca (Neuroimaging Research Unit)	Maria Assunta Rocca has received personal compensation in the range of $500-$4,999 for serving as a Consultant for Biogen, Bristol Myers Squibb, Eli Lilly, Janssen, Roche. Maria Assunta Rocca has received personal compensation in the range of $500-$4,999 for serving on a Speakers Bureau for AstraZaneca, Biogen, Bristol Myers Squibb, Bromatech, Celgene, Genzyme, Horizon Therapeutics Italy, Merck Serono SpA, Novartis, Roche, Sanofi and Teva. The institution of Maria Assunta Rocca has received research support from MS Society of Canada, the Italian Ministry of Health, the Italian Ministry of University and Research, and Fondazione Italiana Sclerosi Multipla.

��ɫ��

��ɫ��