=Paper=
{{Paper
|id=Vol-3742/paper8
|storemode=property
|title=Leveraging graphics tablet and JPen library to detect essential tremor
|pdfUrl=https://ceur-ws.org/Vol-3742/paper8.pdf
|volume=Vol-3742
|authors=Ivan Osiichuk,Vitaly Brevus,Dmytro Bishchak,Yaroslav Mashtaliar,Ivan Mudryk
|dblpUrl=https://dblp.org/rec/conf/citi2/OsiichukBBMM24
}}
==Leveraging graphics tablet and JPen library to detect essential tremor==
https://ceur-ws.org/Vol-3742/paper8.pdf
Leveraging graphics tablet and JPen library to detect
essential tremor
Ivan Osiichuk1,†, Vitaly Brevus1,†, Dmytro Bishchak1, Yaroslav Mashtaliar2, Ivan
Mudryk1,†
1
Ternopil Ivan Puluj National Technical University, 56 Ruska str., Ternopil 46001, Ukraine
2
Ivan Franko National University of Lviv, 1 Universytetska str., Lviv 79000, Ukraine
Abstract
This article delves into the topic of essential tremor detection through the integration of software
recording techniques with graphics tablets and the JPen library.
Essential tremor, a common neurological disorder characterized by involuntary rhythmic shaking,
presents challenges in accurate diagnosis and monitoring. Leveraging the precision and versatility
of graphics tablets, coupled with the functionalities offered by the JPen library, this research
introduces an approach for tremor detection.
Despite the promise of this approach, the article also addresses the challenges and limitations
encountered during implementation with the JPen library and graphics tablet, including issues
related to calibration and signal recording. By discussing these challenges, this research aims to
provide insights for overcoming technical hurdles and optimizing the use of the JPen library in
essential tremor detection.
Keywords
tremor detection, digitizer, graphics tablet, capturing library, spiral recording1
1. Introduction
Essential tremor (ET), also known as familial tremor, is a neurological disorder characterized
by involuntary shaking or trembling of part of the body, most commonly the hands. Still, it
can also affect other areas such as the head, voice, arms, or legs. The tremor typically occurs
with movement and can worsen with stress, fatigue, caffeine, or certain medications [1].
The primary symptom of essential tremor is trembling, which can range from mild to
severe and may worsen over time. The tremor usually affects both sides of the body
CITI’2024: 2nd International Workshop on Computer Information Technologies in Industry 4.0, June 12–14, 2024,
Ternopil, Ukraine
∗
Corresponding author.
†
These authors contributed equally.
osiichuk100@gmail.com (I. Osiichuk); v_brevus@tntu.edu.ua (V. Brevus); dmytro.bischak@gmail.com (D.
Bishchak); mashtaliaryaroslav@gmail.com (Y. Mashtaliar); i1mudryk@ukr.net (I. Mudryk)
0009-0003-6844-2374 (I. Osiichuk); 0000-0002-7055-9905 (V. Brevus); 0009-0005-9993-9302 (D. Bishchak); 0009-
0004-6797-5483 (Y. Mashtaliar); 0000-0002-4305-1911 (I. Mudryk)
© 2024 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR
ceur-ws.org
Workshop ISSN 1613-0073
Proceedings
symmetrically. In addition to the hands, it can involve the head (nodding or shaking), voice
(quavering or shaky speech), and other body parts.
The exact cause of essential tremor is unknown, but it is believed to involve abnormal
electrical brain activity that affects the pathways responsible for controlling movement.
Essential tremor often runs in families, implying the possibility of a genetic component. In
addition, environmental factors may also play a role in triggering or exacerbating the tremors.
Essential tremor is diagnosed based on the medical history, symptoms, and physical
examination of a patient. Doctors may order tests, such as blood tests or imaging scans, to
rule out other conditions causing the tremors. One way to assess tremor is to draw a spiral
[2].
While there is no cure for essential tremor, several treatment options can help manage
symptoms and improve quality of life. These include medications such as beta-blockers, anti-
seizure drugs, or tranquilizers to help reduce tremors.
Certain lifestyle modifications may also help reduce the severity of essential tremor
symptoms. Avoiding triggers such as caffeine, stress, and fatigue can be beneficial.
Occupational therapy or physical therapy may also help individuals with essential tremor
learn techniques to cope with daily activities affected by tremors.
Essential tremor can significantly impact a person's daily life, making tasks that require
precise and steady motility, such as writing, eating, or drinking, challenging. The tremors may
also affect self-esteem and confidence, leading to social withdrawal or the avoidance of certain
activities.
Essential tremor is not life-threatening, and the severity of symptoms can vary widely
among individuals. While the condition may worsen over time, especially in older adults,
some people experience periods of remission or find that their symptoms remain stable with
treatment.
Overall, essential tremor is a chronic condition that requires ongoing management and
support. With proper treatment and lifestyle adjustments, many individuals with essential
tremor can effectively manage their symptoms and lead fulfilling lives.
2. Using the spirals drawing for detection
The ability of individuals with ET to draw both spirals and straight lines could aid in the
clinical assessment process. Digital graphic boards facilitate the collection of data regarding
the location of the pen tip, pressure applied, and tilt angles [3]. Various analytical approaches
have been employed, encompassing both Cartesian representations (y- versus x-position) and
polar views (radius versus angle) of pen tip coordinates [4, 5, 6, 7]. Numerous metrics related
to spiral frequency and time have been documented, including alterations in radius per angle,
radial error, and deviation analysis from an ideal or pre-established spiral. Additionally,
deviations from the starting point have been explored [8]. Studies have examined the
variability in spiral width [9] and the primary tremor axis of the spiral [10, 11, 12].
While drawing straight lines with a pen or digital tool is a more commonplace daily
activity, these tasks (with or without tracing) have been comparatively underexplored in
previous studies and tremor severity assessments compared to the spiral task. Alternative
versions of line trajectory drawing have been examined, such as drawing a line between two
points.
Efforts have been made to ascertain whether visual assessments of spirals and digital
analyses can complement the clinical diagnostic process. Findings indicate that features
derived from digital assessments, such as peak amplitude from velocity spectral curves and
residual deviation and fluctuation characteristics from radial fitted curves, exhibit correlations
with visual rating scores when logarithmic relationships are applied [6]. Moreover,
researchers have employed spiral scores to monitor changes in tremor severity over time,
revealing an upward trend in spiral scores. However, there appears to be a dearth of published
data comparing digital spiral outcome measures between essential tremor (ET) and control
groups, as well as in comparison to other drawing tasks, such as lines.
Drawing tasks entail both fine and gross motor skills and may necessitate different hand
postures, such as with or without the forearm resting on a surface [7]. They may also involve
multiple directional features (e.g., spirals) or necessitate a more defined drawing direction
(e.g., straight lines). In writing, vertical movement is controlled by finger joints, and
horizontal movement by the wrist. Tremor manifestations in ET may vary across upper limb
joints and in different movement directions within each joint, for example, exhibiting more
flexion-extension than a supination-pronation postural tremor in the wrist. Furthermore,
tremors may influence grasping forces and prehension kinematics in precision grips. Previous
studies have identified a primary tremor axis in drawing a spiral and a correlation between
the tremor axis and joint movements [13].
3. Recording a drawing pattern using a graphics tablet
Using a graphics tablet for essential tremor detection presents a promising avenue for both
research and practical application in the medical field. This can improve the quality and
precision of diagnosing and monitoring ET progression.
One of the key components of movement that characterizes a specific disorder is
frequency. For example, for essential tremor, it is 4-10 Hz, and for Parkinson's disease, it is 4-6
Hz [1]. Therefore, the polling rate is an especially important characteristic of the tablet for
diagnosing movement disorders.
However, technological advancements, especially the integration of graphics tablets, offer
innovative solutions for detecting and tracking tremor symptoms. Using a graphics tablet has
certain advantages:
Graphics tablets provide a quantitative and objective method for measuring tremor
severity. Traditional assessments rely on subjective observations by clinicians,
whereas using a graphics tablet allows for precise recording and analysis of tremor
characteristics such as frequency, amplitude, and direction [14].
With the ability to capture tremor data in real-time, graphics tablets enable continuous
monitoring of tremor fluctuations over time. This approach offers insights into the
progression of the condition and the effectiveness of treatment interventions.
Graphics tablets are compact, portable devices that can be easily integrated into
clinical settings or used in patients' homes. This accessibility enhances the feasibility
of frequent monitoring, particularly for individuals who may have difficulty traveling
to medical facilities for assessments [15].
The process has its peculiarities:
The sensitivity of the graphics tablet's sensor can influence the accuracy of tremor
measurements. Calibration procedures are necessary to ensure reliable data collection
and minimize discrepancies caused by variations in tablet sensitivity.
Patients with ET may experience challenges in navigation through the graphical user
interface of the tablet, especially if tremor symptoms affect their fine motor skills.
User-friendly interfaces and intuitive design features are essential for optimizing
usability and minimizing user frustration.
External factors such as environmental vibrations or unintentional movements can
introduce noise or artifacts into tremor data collected via graphics tablets [15]. Robust
filtering algorithms and signal processing techniques are required to differentiate
pathological tremor from extraneous disturbances.
And there are problems:
There is a lack of standardized protocols and guidelines for utilizing graphics tablets in
essential tremor assessment. Establishing consensus on measurement techniques, data
analysis methods, and outcome metrics is crucial for ensuring consistency and
comparability across studies.
While preliminary studies have demonstrated the feasibility and validity of using
graphics tablets for tremor assessment [15], further validation is needed to corroborate
findings and establish the reliability of this approach. Large-scale clinical trials and
comparative studies are necessary to evaluate the accuracy and clinical utility of
graphics tablet-based tremor detection methods.
The cost of graphics tablets may pose barriers to widespread adoption, particularly in
resource-constrained healthcare settings or for individuals with limited financial
means [15]. Efforts to develop affordable and accessible solutions, such as open-source
software platforms or low-cost sensor technologies, could facilitate a broader
integration of graphics tablet-based tremor detection systems.
In summary, leveraging graphics tablets for essential tremor detection offers numerous
advantages in terms of objective measurement, real-time monitoring, and accessibility.
However, addressing peculiarities related to sensor sensitivity, user interface design, and
interference factors, as well as overcoming problems such as standardization, validation, and
cost, are essential for realizing the full potential of this technology in clinical practice.
4. The library for pattern recording
4.1. The JPen library
In addition to the hardware part, software is no less important, being a necessary layer
between the tablet's sensors and the image on the screen or a set of recorded data. The JPen
Java library stands out as a good option for developers seeking to integrate graphics tablet
functionality into their Java applications.
JPen, short for Java Pen, is an open-source Java library designed to simplify the integration
of graphics tablet support into Java applications. JPen provides an abstraction layer that
allows developers to interact with graphics tablets in a platform-independent manner. This
means that regardless of the underlying operating system or tablet hardware, developers can
leverage JPen to access pen input and pressure sensitivity data seamlessly [16].
One of the key features of JPen is its support for various types of graphics tablets,
including those from popular manufacturers like Wacom and Huion. This broad compatibility
ensures that developers can adopt a single, standardized approach and omit device-specific
implementation details, which greatly widens the range of targeted devices and eases the
development and maintenance efforts.
The library offers a straightforward API that abstracts the complexities of dealing with pen
input, making it easy for developers to incorporate tablet support into their Java applications.
With JPen, developers can capture pen events such as pressure, position, tilt, and rotation,
enabling precise and responsive interaction with digital canvas or other graphical elements.
JPen also provides utilities for handling common tasks associated with graphics tablet
integration, such as configuring pen sensitivity settings and calibrating tablet input.
Additionally, the library includes support for multi-touch gestures, allowing developers to
create rich, multi-modal user experiences that combine pen input with touch interactions.
In conclusion, JPen is a valuable tool for developers looking to incorporate graphics tablet
support into their Java applications. With its broad compatibility, intuitive API, and active
community support, JPen simplifies the process of integrating pen input and pressure
sensitivity into Java-based creative software, empowering developers to unleash the full
potential of graphics tablets in their applications.
4.2. Technical aspects of JPen
JPen detects connected graphics tablets using platform-specific APIs or drivers provided by
tablet manufacturers. Upon detection, it initializes communication channels with the tablet,
establishing a bidirectional data flow for sending commands and receiving pen input events.
It employs an event-driven architecture where developers register event listeners to handle
various pen input events. These events include pen-down, pen-up, movement, pressure
changes, tilt, rotation, and multi-touch gestures. When a pen input event occurs, JPen triggers
the corresponding event listener, passing relevant data such as pen coordinates, pressure
levels, and pen properties.
The library translates raw pen input data from the tablet into standardized Java objects or
data structures. For example, it might represent pen coordinates as (x, y) coordinates on the
canvas and pressure levels as floating-point values normalized within a specified range (e.g.,
0.0 to 1.0) [17].
The library uses providers for device access. JPen has providers for Linux (XInput),
Windows (Wintab), Mac OS X (Cocoa, OS X 10.5), and the Java system mouse (Java AWT
mouse).
XInput is an extension of the X Window System protocol used in Linux operating systems.
It provides a standardized interface for handling input devices such as keyboards, mice,
touchpads, and graphics tablets.
Cocoa is Apple's native application framework for macOS and iOS platforms. It provides a
set of APIs and libraries for developing applications that run on Apple's operating systems.
Wintab is an API on Windows for accessing graphics tablets and pen input devices on
Windows systems. It offers a standardized interface for communicating with Wintab-
compatible tablets and accessing features such as pressure sensitivity, tilt, and rotation. Since
the library was mainly tested on Windows, more Wintab aspects are explained in the next
section.
Java AWT is a platform-independent toolkit for building graphical user interfaces (GUIs) in
Java. The use of this provider is universal, but it can degrade the speed of data retrieval and its
completeness. Figure 1 depicts the functional diagram of JPen:
1. The unprocessed information coming from the drawing tablet to the USB allows the
operating system to communicate with the tablet [16]. Without a driver, the operating
system wouldn’t know how to interpret the raw data from the tablet. The data flows
either through the driver layer, if it is installed, or gets into a rawer form.
2. Providers are operating system-specific APIs that can be used to communicate with
the tablet driver and get the raw data. Wintab is for Windows; XInput is used in
Linux; and Cocoa is needed for macOS. If the system does not have a driver, a standard
Java AWT is used. With AWT, not all types of data from the tablet are available and
may not be as accurate. When using drivers, data is available in higher resolution.
3. JPen is a key player in this interaction and connects this raw data with custom-written
software. It converts this data from the providers into Java objects, which can then be
used to write programs. Since specific providers are native to their OS, the Java Native
Interface (JNI) is used to retrieve data from such environments.
4. Then, the data can be used in written programs, where it is stored in the form of Java
objects. Now the information can be displayed on the screen or recorded into a file.
Figure 1: Functional diagram of the JPen with custom-written software
4.3. Wintab provider
After installing the tablet driver on Windows, it also installs DLLs that support Wintab. These
DLLs enable applications to communicate with the driver via a private data context. This
allows the application to integrate the tablet into its own data space and receive pen data
packets and configuration from the driver. Due to the partitioning of Wintab support within
the driver, multiple applications can have their own separate Wintab pen data streams from
the driver (Figure 2).
Figure 2: The general pipeline with Wintab
The core components of the Wintab programming model revolve around Wintab contexts
and Wintab data acquisition. In this model, an application is responsible for creating,
configuring, and opening a Wintab context. Within the application's message loop, it either
reacts to Wintab data messages and retrieves data or responds to other events like mouse
actions and retrieves data accordingly. Subsequently, the acquired data can be processed
within the application's environment for various purposes, such as drawing or signature
capture, both of which are common examples.
The Wintab API has long supported multiple independent contexts per application and per
tablet, allowing apps to receive separate pen data streams from each attached tablet. Contexts
are crucial for managing per-application data, defining what data is sent to the app, and how
it maps to the system display. Each app has its own context, which can be customized without
affecting other apps or global tablet preferences. Contexts serve as the interface for tablet
usage in applications, containing information about the tablet area, event types, and delivery
methods. While apps can open multiple contexts, most require only one. Contexts are
managed in an overlapping container, with the topmost context processing tablet events
based on overlapping order and attributes. Wintab ensures that each application receives data
only when it has keyboard focus, preventing background apps from interfering with
foreground tasks. This approach was implemented due to many apps not handling their
contexts properly.
Wintab offers two approaches for capturing pen data: message-driven and polling. In
message-driven capture, pen data is delivered to the application via Windows messages
triggered by the driver when data is available. The application processes this data and awaits
the next message. In polling capture, the application prompts Wintab for data when certain
non-Wintab messages (e.g., mouse-down or move messages) are received. This allows
continuous data caching as the pen moves within the tablet's range. Regardless of the capture
method, Wintab applications should register for messages notifying them of key events like
driver status changes or tablet attachment or detachment to manage their state effectively.
Since each application's Wintab context relies on the tablet driver's support, it remains valid
only while the driver is active. If the driver restarts (e.g., after a system reboot) or the
application is relaunched, a new Wintab context is created, requiring the application to
reopen it. Wintab message notifications facilitate managing such situations [18].
4.4. The overview of available data
It is important to know what data is available to us to build an analysis based on known input
components. For a general overview, it’s possible to take a demo offered by JPen. Figure 3
shows the graphical interface of this program.
It provides various types of data that can be obtained from a graphics tablet. Here are the
key data types available:
First, X and Y coordinates represent the horizontal and vertical positions of the pen on
the tablet surface.
It measures the pressure exerted by the pen on the tablet.
Tilt-X and Tilt-Y indicate the angles at which the pen is tilted in the X and Y
directions.
Side Pressure is an additional pressure metric, indicating the pressure applied from the
side of the pen.
Also, the rotation of the pen around its axis is measured.
Figure 3: The GUI of the JPen demo
In addition to these characteristics, the sampling period is also shown, which is important
for further work with the data. The optimal period gives the least data loss and less noise [19].
4.5. Event-listener architecture of JPen
JPen uses the event-listener architecture for interaction, an example of which is shown in
Figure 4. Event/listener architecture is a mechanism for handling events generated by user
interactions or system actions within applications. This architecture allows developers to
define event listeners to respond to specific events, such as button clicks, mouse movements,
key presses, and more [20].
Figure 4: Sequence diagram of JPen’s event/listener architecture
In general, the library's work follows this architecture without any major peculiarities:
When an interaction occurs on the tablet, the JPen library creates an event object
encapsulating the details of the interaction. This object contains information such as
coordinates, pressure, tilt, etc., depending on the type of interaction and capabilities of
the device.
The JPen library defines a listener interface, which defines methods for processing
different types of events. To use it in an application, you need to create a class
implementing these methods and add it as a listener.
When a user interacts with a tablet, the library detects the event and dispatches it to
all registered listeners.
Custom event-handling logic is implemented within the listener methods. For instance,
developers might process pen-down, pen-move, or pen-up events by updating the UI,
capturing input data, triggering actions, etc.
5. Tablet input recording solution
5.1. The overview of the prototype
To use JPen, it is required to create a data-recording application. The basic implementation of
the library requires the use of at least one AWT component. Building a Swing application as
the base is one possibility. The general class diagram is shown in Figure 5.
Figure 5: The class diagram of the tablet input recorder app
The program is launched by SpiralRecorderApp, which also initializes the Swing
application's main window. In addition to representing the GUI's primary window,
WindowFrame houses the buttons that allow a user to begin recording or saving a file. The
DrawingCanvas, a canvas for showing the spiral pattern, is a component of this class. It shows
the points that the graphics tablet has provided. The DrawingLine object is used to create a
link between such points because they might be received at various frequencies. This ensures
smooth tracing by drawing a line between the previously registered point and the current one
instead of showing individual points.
The DataManager controls the recording's status and writes it to a file when it comes to
working with input data from the tablet. PointRecorder is a component that works directly
with JPen; it maintains a collection of received points and has event handlers.
PenManager is responsible for connecting components that can handle library events and is
used in two different places in the program: DrawingCanvas and PointRecorder, which are
the two main places that interact with the library. Both places use the device manager
singleton from the library. These two classes save individual registered points, together with
their coordinates and time, using an object called SavedPoint.
Additionally, utility classes are required at certain points throughout the application. The
tablet's configuration reads from a file and provides access to the configuration; both features
are necessary to calculate relative coordinates. With a clock, you can obtain the current time
in the required format. The view of the prototype window is shown in Figure 6.
Figure 6: The GUI of the prototype
5.2. The models of the tablet used for motion capturing
Hardware, especially graphics tablets, is an important component of motion detection
solutions. Wacom Bamboo Capture and Huion Kamvas Pro 16 are used in the experiments
(Table 1).
Table 1
Tablets specification
Wacom Bamboo
Property Huion Camvas Pro
Capture
Resolution (pixels) - (no screen) 1920x1080
Polling rate (Hz) 133 220
Working area (mm) 147 x 92 344.16 x 193.59
5.3. The timestamp duplication problem
One of the problems with the data is the repetition of timestamps for different coordinates. An
example of such repetition for scheduled time can be seen in Table 2.
Table 2
The example of scheduled time repetition for different coordinates
registered time scheduled time r delta s delta x y z
0:0:0.937 0:0:0.946 0 0 80.1 -63.4 0.0
0:0:0.943 0:0:0.946 6 0 80.3 -63.2 0.0
Potential reasons for this behavior may include:
JNI, which has a certain overhead.
The driver used by a tablet (Wintab).
Hardware issues, such as a peculiarity of the tablet or incorrect communication with a
driver.
To understand the overall picture, it is useful to analyze groups of coordinates with the
same timestamp. Figure 7 shows the chart of coordinates over registered time for Bamboo
(left) and Huion (right).
The right part shows a significant number of "steps" representing different coordinates for
the same timestamp. On the other hand, the left chart shows a smooth line, with each
coordinate having a unique time. This pattern is repeated along the entire time axis, but the
intensity varies.
Figure 7: Coordinates by time for Bamboo (left) and Huion (right)
For greater clarity, the distribution of the number of points for each unique registered time
value can be analyzed. Since there may be additional hand movements at the beginning and
end of the recording, the first and last seconds of the recording were skipped for each dataset
to remove this factor. Table 3 shows the statistics for both tablets. As can be seen from Table
4, all registered points from Bamboo have a unique timestamp. In turn, 1 unique time point for
Huion is rare, as there is significant duplication of time values for coordinates.
To disclaim the responsibility of JNI and the library, it is possible to record a report using
the Wintab diagnostic tool [21]. When used, a file is created that similarly contains data on
the position of the pen at certain points in time. Thus, the data will be obtained closest to the
source using the software. Statistics for this tool are presented in the columns labeled as
wtcapt and calculated for both tablets. In total, 15 experiments were conducted for the Java-
prototype data and 20 for wtcapt. The tables show aggregated statistics for these data, with
the median taken for each characteristic.
Table 3
The statistics for the number of points for each unique registered timestamp
Statistics Bamboo (Java) Bamboo(wtcapt) Huion (Java) Huion (wtcapt)
groups 2213 296 895 308
mean 1 1 1.8 3.4
std 0 0 1 0.4
min 1 1 1 3
25% 1 1 1 3
50% 1 1 1 3
75% 1 1 3 4
max 1 1 4 4
Table 4
The distribution of the points number in each time group
Number of such groups
Number of points in the
Bamboo Huion Huion
group Bamboo(wtcapt)
(Java) (Java) (wtcapt)
1 2213 296 494.5 -
2 - - 131.5 -
3 - - 232.5 174
4 - - 60 134
7 - - - 19
The frequency is calculated based on the first and last timestamps and the total number of
points:
𝑙𝑡 − 𝑓𝑡 1
𝑝𝑒𝑟𝑖𝑜𝑑 = , 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 = (1)
𝑝−1 𝑝𝑒𝑟𝑖𝑜𝑑
Where lt – the last timestamp (seconds), ft – first timestamp (seconds), p – the number of
points.
The results are presented in Table 5. The maximum period is obtained as the maximum
value from the registeredDelta column, which shows the period for each pair of points.
Table 5
Calculated data frequency
Property Bamboo (Java) Bamboo(wtcapt) Huion (Java) Huion (wtcapt)
period (s) 0.0075 0.0075 0.00459 0.00453
frequency (Hz) 133 133 217 220
frequency w/o duplicates
133 133 118 63
(Hz)
maximum period (s) 0.008 0.008 0.032 0.032
The Huion Kamvas tablet emits data more frequently, but in both cases, there are many
coordinates with the same timestamp. In turn, Wacom Bamboo has a unique time for each
coordinate with the lower rate.
The recording frequency with all the points coincides with the manufacturer's claim.
However, after removing duplicates from the Huion data, this frequency is much lower. When
deleting duplicates, the last registered value of the coordinate group was retained.
As a solution to this problem, it is possible to discard the earlier points, leaving only the
last one for this time, since the movement is linear. However, all the points may be useful for
further analysis, in which case it is possible to add a marker for such points, realizing that
these are sequential movements, and the next point has a slightly later time stamp.
Still, the period between two consecutive records is not uniform throughout the whole
captured dataset in the case of both tablets, which may complicate carrying out any further
analysis of the dataset. To convert the dataset into one with a homogeneous period, one of the
existing tabulated function interpolation methods could be adopted, such as B-spline. For this,
the dataset could be represented as two functions dependent on time, for X and Y coordinates,
respectively. For each function, an interpolation is built, and a new dataset is created by
combining the evaluated values over an evenly spaced time-series for the whole capture
period. However, such techniques often require the domain of an interpolated function to
contain unique values. Hence, a dataset with duplicated time points, like those obtained from
Huion, must be carefully preprocessed and cleaned before applying such methods.
5.4. Java Native Interface (JNI)
An additional delay may be imposed by JNI. It is a technology that enables Java code to
interact with native applications and libraries written in other languages, such as C, C++, and
assembly. JNI allows Java applications to leverage existing native code libraries or access
system-level resources that are not directly accessible from Java.
JNI starts with Java code running within the JVM. When Java code needs to interact with
native code, it uses special syntax and constructs provided by the JNI framework. The native
code, typically written in languages like C or C++, contains the implementations of the
functions that Java code wants to call or interact with. This code must be compiled into
platform-specific binary files, such as DLLs (Dynamic Link Libraries) on Windows or shared
objects (.so) on Unix-like systems.
To invoke native code from Java, JNI provides a set of native methods and functions that
serve as a bridge between Java and native code. Java code calls these native methods, which
are declared using the native keyword. The JVM then locates and invokes the corresponding
native functions defined in the native code.
Java and native code have different data representations, and JNI provides functions for
converting data between Java types (such as int, String, Array, etc.) and their corresponding
native types (such as jint, jstring, jarray, etc.).
Since Wintab is written in C++ and is native to Windows, the Java app can only
communicate with it using such interfaces. And this can be a weak point in the system.
Conclusion
In conclusion, this set of tools is well-suited for motion analysis. The JPen library makes it
possible to record coordinates from various devices and allows you to use tablets with or
without a screen. Since the library is written in Java, the solution is cross-platform and works
on different operating systems.
The article compares Huion and Bamboo graphics tablets using a prototype for motion
recording. Among the problems considered is the repetition of time for different coordinates.
The analysis of the resulting trajectory with this problem is presented, and both tablets are
compared in terms of their behavior concerning this problem. Bamboo shows no duplication
of points by registered time but does not have a display, which can make it somewhat difficult
to perform the drawing test. Huion, on the other hand, has a higher polling rate and display
but has a significant number of different coordinates at the same time.
This analysis also made it possible to obtain the nominal frequency of each graphics tablet,
which is important for understanding the range of movement disorders that can be assessed
using such hardware. Both tablets can be used for the task of diagnosing tremor because the
frequency (133 Hz for Bamboo), even when duplicated (60 Hz for Huion), is higher than that
defined for tremor as up to 10 Hz.
The solution is suitable for motion recording in movement disorders and allows for
efficient retrieval of this data for further analysis. The advantage of custom-written software
for such a task is the ability to customize the required data to be recorded. Also, it allows for
the preprocessing of this data at the capturing stage.
References
[1] Haubenberger, Dietrich, and Mark Hallett. "Essential tremor." New England Journal of
Medicine 378.19 (2018): 1802-1810.
[2] Performance tests, 2024. URL: https://www.mayoclinic.org/diseases-conditions/essential-
tremor/diagnosis-treatment/drc-20350539.
[3] Zajki-Zechmeister, Tibor, et al. "Quantification of tremor severity with a mobile tremor
pen." Heliyon 6.8 (2020).
[4] Szumilas, Mateusz, et al. "A multimodal approach to the quantification of kinetic tremor
in Parkinson’s disease." Sensors 20.1 (2019): 184.
[5] Yu, N.-Y.; Van Gemmert, A.W.A.; Chang, S.-H. Characterization of graphomotor
functions in individuals with Parkinson’s disease and essential tremor. Behav. Res.
Methods 2017, 49, 913–922.
[6] Lin, Po-Chieh, et al. "A digital assessment system for evaluating kinetic tremor in
essential tremor and Parkinson’s disease." BMC neurology 18 (2018): 1-8.
[7] Louis, Elan D. "Utility of the hand-drawn spiral as a tool in clinical-epidemiological
research on essential tremor: data from four essential tremor cohorts."
Neuroepidemiology 44.1 (2015): 45-50.
[8] Ishii, Nobuyuki, et al. "Spiral drawing: Quantitative analysis and artificial-intelligence-
based diagnosis using a smartphone." Journal of the neurological sciences 411 (2020):
116723.
[9] Michalec, M.; Hernandez, N.; Clark, L.N.; Louis, E.D. The spiral axis as a clinical tool to
distinguish essential tremor from dystonia cases. Park. Relat. Disord. 2014, 20, 541–544.
[10] Roth, Navit, Orit Braun-Benyamin, and Sara Rosenblum. "Drawing Direction Effect on a
Task’s Performance Characteristics among People with Essential Tremor." Sensors 21.17
(2021): 5814.
[11] M. Petryk, M. Bachynskyi, V. Brevus, I. Mudryk, and D. Mykhalyk, “Analysis technology
of neurological movements considering cognitive feedback influences of cerebral cortex
signals” in Proceedings of the 2nd International Workshop on Information Technologies:
Theoretical and Applied Problems (ITTAP 2022), Eds., in CEUR Workshop Proceedings,
vol. 3309. Ternopil, Ukraine: CEUR, Nov. 2022, pp. 45–54. Accessed: Sep. 10, 2023.
[Online]. Available: https://ceur-ws.org/Vol-3309/#paper4
[12] Petryk M.R. Methods and models for identification of parameters of complex
multicomponent nanoporous and nanoscale structures and processes: monograph / M.R.
Petryk, I.V. Boyko, M.I. Lebovka, V.M. Brevus– Ternopil: Ivan Pulyuy TNTU Publishing
House, 2023 – 156 p.
[13] Buijink, A. W., van der Stouwe, A. M., Broersma, M., Sharifi, S., Tijssen, M. A., &
Speelman, J. D. (2015). Motor network disruption in essential tremor: a functional and
effective connectivity study. Brain, 138(10), 2934-2947.
[14] Kim, Christine Y., et al. "Repeated spiral drawings in essential tremor: a possible limb-
based measure of motor learning." The Cerebellum 18 (2019): 178-187.
[15] Mcgurrin, Patrick, et al. "Quantifying tremor in essential tremor using inertial sensors—
Validation of an algorithm." IEEE journal of translational engineering in health and
medicine 9 (2020): 1-10.
[16] JPen, 2024. URL: https://github.com/nicarran/jpen
[17] San Luciano, Marta, et al. "Digitized spiral drawing: A possible biomarker for early
Parkinson’s disease." PloS one 11.10 (2016): e0162799.
[18] Wintab basics, 2024. URL: https://developer-
docs.wacom.com/docs/icbt/windows/wintab/wintab-basics/
[19] Gil-Martín, Manuel, Juan Manuel Montero, and Rubén San-Segundo. "Parkinson’s disease
detection from drawing movements using convolutional neural networks." Electronics 8.8
(2019): 907.
[20] Korunović, Ana, and Siniša Vlajić. "An Example of Integration of Java GUI Desktop
Technologies Using the Abstract Factory Pattern for Education Purposes." ETF Journal of
Electrical Engineering 29.1 (2023): 3-11.
[21] Wintab diagnostic, 2024. URL: https://developer-support.wacom.com/hc/en-
us/articles/9354461019927-Wintab-diagnostic