Comparison of CNN-Learned vs. Handcrafted Features for Detection of Parkinson’s Disease Dysgraphia in a Multilingual Dataset

Simple item page
dc.contributor.authorGaláž, Zoltáncs
dc.contributor.authorDrotár, Petercs
dc.contributor.authorMekyska, Jiřícs
dc.contributor.authorGazda, Matejcs
dc.contributor.authorMucha, Jáncs
dc.contributor.authorZvončák, Vojtěchcs
dc.contributor.authorSmékal, Zdeněkcs
dc.contributor.authorFaúndez Zanuy, Marcoscs
dc.contributor.authorCastrillon, Reinelcs
dc.contributor.authorOrozco-Arroyave, Juan Rafaelcs
dc.contributor.authorRapcsak, Steven Z.cs
dc.contributor.authorKincses, Tamáscs
dc.contributor.authorBrabenec, Lubošcs
dc.contributor.authorRektorová, Irenacs
dc.coverage.issue1cs
dc.coverage.volume16cs
dc.date.issued2022-05-30cs
dc.description.abstractParkinson’s disease dysgraphia (PDYS), one of the earliest signs of Parkinson’s disease (PD), has been researched as a promising biomarker of PD and as the target of a noninvasive and inexpensive approach to monitoring the progress of the disease. However, although several approaches to supportive PDYS diagnosis have been proposed (mainly based on handcrafted features (HF) extracted from online handwriting or the utilization of deep neural networks), it remains unclear which approach provides the highest discrimination power and how these approaches can be transferred between different datasets and languages. This study aims to compare classification performance based on two types of features: features automatically extracted by a pretrained convolutional neural network (CNN) and HF designed by human experts. Both approaches are evaluated on a multilingual dataset collected from 143 PD patients and 151 healthy controls in the Czech Republic, United States, Colombia, and Hungary. The subjects performed the spiral drawing task (SDT; a language-independent task) and the sentence writing task (SWT; a language-dependent task). Models based on logistic regression and gradient boosting were trained in several scenarios, specifically single language (SL), leave one language out (LOLO), and all languages combined (ALC). We found that the HF slightly outperformed the CNN-extracted features in all considered evaluation scenarios for the SWT. In detail, the following balanced accuracy (BACC) scores were achieved: SL—0.65 (HF), 0.58 (CNN); LOLO—0.65 (HF), 0.57 (CNN); and ALC—0.69 (HF), 0.66 (CNN). However, in the case of the SDT, features extracted by a CNN provided competitive results: SL—0.66 (HF), 0.62 (CNN); LOLO—0.56 (HF), 0.54 (CNN); and ALC—0.60 (HF), 0.60 (CNN). In summary, regarding the SWT, the HF outperformed the CNN-extracted features over 6%(mean BACC of 0.66 for HF, and 0.60 for CNN). In the case of the SDT, both feature sets provided almost identical classification performance (mean BACC of 0.60 for HF, and 0.58 for CNN).en
dc.formattextcs
dc.format.extent1-18cs
dc.format.mimetypeapplication/pdfcs
dc.identifier.citationFrontiers in Neuroinformatics. 2022, vol. 16, issue 1, p. 1-18.en
dc.identifier.doi10.3389/fninf.2022.877139cs
dc.identifier.issn1662-5196cs
dc.identifier.orcid0000-0002-8978-351Xcs
dc.identifier.orcid0000-0002-6195-193Xcs
dc.identifier.orcid0000-0001-5126-440Xcs
dc.identifier.orcid0000-0002-1948-4653cs
dc.identifier.orcid0000-0002-8483-5448cs
dc.identifier.other177994cs
dc.identifier.researcheridT-8761-2019cs
dc.identifier.researcheridK-4001-2015cs
dc.identifier.researcheridT-9091-2019cs
dc.identifier.scopus56888706700cs
dc.identifier.scopus35746344400cs
dc.identifier.scopus57201029686cs
dc.identifier.scopus36855362600cs
dc.identifier.urihttp://hdl.handle.net/11012/204669
dc.language.isoencs
dc.publisherFrontierscs
dc.relation.ispartofFrontiers in Neuroinformaticscs
dc.relation.urihttps://www.frontiersin.org/articles/10.3389/fninf.2022.877139/fullcs
dc.rightsCreative Commons Attribution 4.0 Internationalcs
dc.rights.accessopenAccesscs
dc.rights.sherpahttp://www.sherpa.ac.uk/romeo/issn/1662-5196/cs
dc.rights.urihttp://creativecommons.org/licenses/by/4.0/cs
dc.subjectmachine learningen
dc.subjectdeep learningen
dc.subjectfeature extractionen
dc.subjectParkinson’s disease dysgraphiaen
dc.subjecthandwriting analysisen
dc.titleComparison of CNN-Learned vs. Handcrafted Features for Detection of Parkinson’s Disease Dysgraphia in a Multilingual Dataseten
dc.type.driverarticleen
dc.type.statusPeer-revieweden
dc.type.versionpublishedVersionen
sync.item.dbidVAV-177994en
sync.item.dbtypeVAVen
sync.item.insts2025.02.03 15:42:26en
sync.item.modts2025.01.17 15:21:30en
thesis.grantorVysoké učení technické v Brně. Fakulta elektrotechniky a komunikačních technologií. Ústav telekomunikacícs
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
fninf16877139.pdf
Size:
2.24 MB
Format:
Adobe Portable Document Format
Description:
fninf16877139.pdf
Download
Collections
Ústav telekomunikací