View in EDS

Hand-Drawn Image (HDI) dataset: Deep approach for essential tremor recognition.

Saved in:
Bibliographic Details
Title: Hand-Drawn Image (HDI) dataset: Deep approach for essential tremor recognition.
Authors: Alwaer, Thiheebah1,2 (AUTHOR) thiheebah.alwaer@regim.usf.tn, Jarraya, Islem1 (AUTHOR) islem.jarraya@regim.usf.tn, Dhieb, Thameur1,3 (AUTHOR) thameur.dhieb@regim.usf.tn, Neji, Mohamed1,2 (AUTHOR) mohamed.neji@regim.usf.tn, Farhat, Nouha4,5 (AUTHOR) nouha.farhat15@gmail.com, Sellami, Sirine4,5 (AUTHOR) sirinesellami16@gmail.com, Hamdani, Tarek M.1,6 (AUTHOR) tarek.hamdani@regim.usf.tn, Damak, Mariem4,5 (AUTHOR) mariem_benamar@yahoo.fr, Mhiri, Chokri4,5 (AUTHOR) mhiri.chokri@gmail.com, Alimi, Adel M.1,7 (AUTHOR) adel.alimi@regim.usf.tn
Source: Multimedia Tools & Applications. Mar2026, Vol. 85 Issue 3, p1-29. 29p.
Abstract: Essential Tremor (ET) is a common neurological disorder characterized by involuntary hand tremors that worsen during voluntary movement and diminish at rest. Hand-drawing tasks have proven effective for assessing ET progression. However, most existing studies focus on digital devices or task-limited approaches, such as spiral drawing alone, overlooking the simplicity and accessibility of paper-based methods. Moreover, no publicly available dataset currently includes multiple drawing tasks per individual to enable comprehensive ET severity evaluation. This paper introduces a novel hand-drawn image dataset specifically designed for ET analysis, comprising images from 28 ET patients and 28 healthy controls collected at Habib Bourguiba Hospital in Sfax, Tunisia. Participants performed three distinct hand-drawing tasks, spirals at two distances and straight lines, with both hands, aligned with Tasks 11, 12, and 13 of the Fahn–Tolosa–Marin Tremor Rating Scale. To classify ET from these images, this work employed various deep learning models and introduced, for the first time in this context, the TinySiamese network. TinySiamese was adapted to fuse features across all six hand-drawing tasks (three per hand) performed by each individual, significantly improving classification performance. This fusion approach enables a more accurate and holistic tremor assessment by capturing bilateral motor patterns and task-specific variations that single-task models may overlook. The proposed dataset and deep learning framework offer a practical, low-cost solution for ET screening and monitoring, particularly suited for resource-limited settings. By combining paper-based tasks with smartphone digitization and lightweight deep learning models, this work provides a scalable and accessible tool for early diagnosis and disease progression tracking. [ABSTRACT FROM AUTHOR]
Copyright of Multimedia Tools & Applications is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
Description
Abstract:Essential Tremor (ET) is a common neurological disorder characterized by involuntary hand tremors that worsen during voluntary movement and diminish at rest. Hand-drawing tasks have proven effective for assessing ET progression. However, most existing studies focus on digital devices or task-limited approaches, such as spiral drawing alone, overlooking the simplicity and accessibility of paper-based methods. Moreover, no publicly available dataset currently includes multiple drawing tasks per individual to enable comprehensive ET severity evaluation. This paper introduces a novel hand-drawn image dataset specifically designed for ET analysis, comprising images from 28 ET patients and 28 healthy controls collected at Habib Bourguiba Hospital in Sfax, Tunisia. Participants performed three distinct hand-drawing tasks, spirals at two distances and straight lines, with both hands, aligned with Tasks 11, 12, and 13 of the Fahn–Tolosa–Marin Tremor Rating Scale. To classify ET from these images, this work employed various deep learning models and introduced, for the first time in this context, the TinySiamese network. TinySiamese was adapted to fuse features across all six hand-drawing tasks (three per hand) performed by each individual, significantly improving classification performance. This fusion approach enables a more accurate and holistic tremor assessment by capturing bilateral motor patterns and task-specific variations that single-task models may overlook. The proposed dataset and deep learning framework offer a practical, low-cost solution for ET screening and monitoring, particularly suited for resource-limited settings. By combining paper-based tasks with smartphone digitization and lightweight deep learning models, this work provides a scalable and accessible tool for early diagnosis and disease progression tracking. [ABSTRACT FROM AUTHOR]
ISSN:13807501
DOI:10.1007/s11042-026-21422-3