An ordinal CNN approach for the assessment of neurological damage in Parkinson's disease patients

Research areas:

• Applications - Health
• Methodology - Machine learning
• Methodology - Ordinal classification
• Methodology - Convolutional Neural Networks

Year:

2021

Type of Publication:

Article

Keywords:

Artificial Neural Networks, Ordinal Classification, Data augmentation Computer-Aided Diagnosis

Authors:

• Barbero-Gómez, Javier
• Gutiérrez, Pedro Antonio
• Vargas, Víctor Manuel
• Vallejo-Casas, Juan-Antonio
• Hervás-Martínez, César

Journal:

Expert Systems with Applications

Volume:

182

Pages:

115271

Month:

November

ISSN:

0957-4174

BibTex:

@article{BarberoCNNParkinsonESWA, author = "Javier Barbero-G{\'o}mez and Pedro Antonio Guti{\'e}rrez and V{\'i}ctor Manuel Vargas and Juan-Antonio Vallejo-Casas and C{\'e}sar Herv{\'a}s-Mart{\'i}nez", abstract = "3D image scans are an assessment tool for neurological damage in Parkinson{\&}#039;s disease (PD) patients. This diagnosis process can be automatized to help medical staff through Decision Support Systems (DSSs), and Convolutional Neural Networks (CNNs) are good candidates, because they are effective when applied to spatial data. This paper proposes a 3D CNN ordinal model for assessing the level or neurological damage in PD patients. Given that CNN need large datasets to achieve acceptable performance, a data augmentation method is adapted to work with spatial data. We consider the Ordinal Graph-based Oversampling via Shortest Paths (OGO-SP) OGO-SP method, which applies a gamma probability distribution for inter-class data generation. A modification of OGO-SP is proposed, the OGO-SP-beta algorithm, which applies the beta distribution for generating synthetic samples in the inter-class region, a better suited distribution when compared to gamma. The evaluation of the different methods is based on a novel 3D image dataset provided by the Hospital Universitario Reina Sof{\'i}a (C{\'o}rdoba, Spain). We show how the ordinal methodology improves the performance with respect to the nominal one, and how OGO-SP-beta yields better performance than OGO-SP.", awards = "JCR(2021): 8.665 Position: 21/144 (Q1) Category: COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE", comments = "JCR(2021): 8.665 Position: 21/144 (Q1) Category: COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE", doi = "10.1016/j.eswa.2021.115271", issn = "0957-4174", journal = "Expert Systems with Applications", keywords = "Artificial Neural Networks, Ordinal Classification, Data augmentation Computer-Aided Diagnosis", month = "November", note = "JCR(2021): 8.665 Position: 21/144 (Q1) Category: COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE", pages = "115271", title = "{A}n ordinal {CNN} approach for the assessment of neurological damage in {P}arkinson{\&}#039;s disease patients", url = "www.sciencedirect.com/science/article/pii/S0957417421007028", volume = "182", year = "2021", }

Note:

JCR(2021): 8.665 Position: 21/144 (Q1) Category: COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Abstract:

3D image scans are an assessment tool for neurological damage in Parkinson's disease (PD) patients. This diagnosis process can be automatized to help medical staff through Decision Support Systems (DSSs), and Convolutional Neural Networks (CNNs) are good candidates, because they are effective when applied to spatial data. This paper proposes a 3D CNN ordinal model for assessing the level or neurological damage in PD patients. Given that CNN need large datasets to achieve acceptable performance, a data augmentation method is adapted to work with spatial data. We consider the Ordinal Graph-based Oversampling via Shortest Paths (OGO-SP) OGO-SP method, which applies a gamma probability distribution for inter-class data generation. A modification of OGO-SP is proposed, the OGO-SP-beta algorithm, which applies the beta distribution for generating synthetic samples in the inter-class region, a better suited distribution when compared to gamma. The evaluation of the different methods is based on a novel 3D image dataset provided by the Hospital Universitario Reina Sofía (Córdoba, Spain). We show how the ordinal methodology improves the performance with respect to the nominal one, and how OGO-SP-beta yields better performance than OGO-SP.

Comments:

JCR(2021): 8.665 Position: 21/144 (Q1) Category: COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE