You are on page 1of 4

Dental Intraoral System Supporting Tooth

Segmentation
Truong Quang Vinh, Bui Minh Thanh, Nguyen Ngoc Tai
Vietnam National University - Ho Chi Minh City (VNU-HCM)
Ho Chi Minh City University of Technology (HCMUT)
Faculty of Electrical and Electronics Engineering, Ho Chi Minh, Vietnam
1
tqvinh@hcmut.edu.vn ; 2bmthanh@hcmut.edu.vn; 3ngoctai0308@gmail.com

Abstract—Embedded systems are being further developed and touch screen GUI, dental image processing, patient record, and
applied in medical devices. In this paper, we present an Vietnamese language for domestic clinic.
embedded design for dental intraoral system which supports In this paper, we present an embedded system design for
dental image capturing and image tooth segmentation. This dental intraoral system. This system utilizes ARM
device assists dentists in diagnosis by using dental images, which
microprocessor running on embedded Linux operating system.
is captured from dental camera. Moreover, we propose advanced
features for the dental intraoral system including touch screen Our system is supported by QT and the libraries of Open-CV,
with Vietnamese graphic user interface (GUI), dental image V4L, Libcurl. Our system has following functions:
processing, patient records, and dentist’s diagnosis note. • Capturing dental image
Especially, the segmentation of teeth is important for examining • Touch screen GUI, supporting Vietnamese language.
and extracting teeth features from dental images. A teeth • Dental image processing including brightness, contrast,
segmentation method based on active contour without edge gamma, and tooth segmentation.
algorithm has been proposed in this paper. Consequently, our • Patient records and dentist’s diagnosis note
system is portable, economic and ready to be applied at dental The remaining parts of this paper are organized as follows.
clinics. The system can help dentists examine at patient’s home Section II presents system description of the proposed dental
and voyages, not only in clinics.
imaging system. The proposed algorithm for tooth
Keywords— Dental intraoral system, dental image, tooth segmentation is then described in section III. Section IV
segmentation, ARM, Micro2440, embedded system, Linux, presents experimental results. Finally, the conclusions are
OpenCV, Libcurl described in the last section.

I. INTRODUCTION II. SYSTEM DESCRIPTION


Nowadays, embedded systems with ARM microprocessor The system is partitioned into 2 parts: hardware and
have become widely used in medical systems due to the software. Hardware is built based on Friendly ARM
advantages of small size, low power, and high performance. Mirco2440 board. Dental images are captured from a dental
Most of the medical applications are based on image camera through USB and showed on 7-inch touch screen LCD.
acquisition systems [4], [5]. These kinds of systems are Software is built in Embedded Linux environment. An
currently applied much in medical devices such as X-ray embedded program is developed by using QT creator with
scanner, computed tomography (CT) scanner, magnetic OpenCV[3], and Libcurl[8] library. The program supports
resonance imaging (MRI), and dental intraoral system. functions including management patients, supports dentists for
The image acquisition systems typically support three diagnosis by dental images, dental image processing, and
functions: capturing images from camera, processing image patient’s data storage
data, and displaying results. For examples, Yong-lin LIU A. Hardware Description
proposed an image acquisition system using ARM and Linux
[1]. In this design, the embedded image processing system
integrated the functions of image capturing, displaying, and
processing. In other design, Yuanping Ni proposed an image
acquirement with Bluetooth technique [2]. This system
acquires images from USB camera and transmits them to
medical data centre.
Dental intraoral system is one of image acquisition
applications. KAB Dental [6], Miharu [7] have presented
dental devices which can capture tooth images, and display
them on LCD. However, these current products do not support
Figure 1. Block diagram of hardware design

978-1-4673-2088-7/13/$31.00 ©2013 IEEE 326


processing, surveillance camera, identification, and robot
The hardware of our system based on Micro2440 is shown programming. Libcurl, developed by Daniel Stenberg, is a
in Figure 1. Micro2440 is ARM9 development board and has library which supports for connecting and communicating
many applications in embedded systems. Micro2440 uses different types of servers with many different types of
S3C2440 microprocessor which is a 32-bit microcontroller protocols. Libcurl supports the http, https, ftp, gopher, telnet,
fully integrated ARM920T kernel as a core controller. dict, file, and ldap protocols. Libcurl also supports HTTPS
S3C2440 has interface with SD Card, USB Camera (via USB certificates, HTTP POST, HTTP PUT, FTP uploading, HTTP
port), PC (via RS232 port) and Ethernet. Besides, touch screen form based upload, proxies, cookies, and user-password
authentication.
LCD, camera, and memory are equipped.
The final part of our system is an embedded program
Camera unit capture images and send them to S3C2440 via
presented in Figure 3. The embedded program is divided into
USB port. The 1.3M Dental Intra-Oral USB Camera which is two main parts: Graphic User Interface and Dental Image
small, high sensitivity and high definition has been used in our Capturing and Processing.
system. Dental images are sent to CPU and are showed on
LCD unit. The user can easily manipulate and control the
system by using available 7-inch touch screen LCD.
Furthermore, the images and data can be stored in SD card.
Micro2440 board supports 256M NAND Flash and 64M
SDRAM, including two 32M HY57V561620 chips with the
frequency 100MHz, for storing Linux kernel and embedded
program.
B. Software Description
Figure 2 shows the software structure of our system. The
software system consists of 4 parts, namely Bootloader, Figure 3. Structure of Graphical User Interface
Embedded Linux, Library and Application.

Start

Capture device information

N
Look for hardware device
Figure 2. Software structure
Y
Firstly, bootloader is started when the power is on. The
Open device
bootloader Vivi developed by MIZI Company is used as a
bootloader in the system. Bootloader helps hardware to
transport the kernel of the operating system into RAM Take IplImage
memory. Moreover, the arm-linux-gcc-4.4.3 package is used
to compile necessary files for Embedded Linux.
Secondly, after the bootloader has been run, the Linux Change to Pixmap Image
kernel and Root File System is loaded into system memory.
Linux kernel is a core of operating system (OS) which contains
Capture data Image
controlling functions of hardware. The Kernel Linux-2.6.32.2
which is suitable for Mirco2440 board is utilized in our system.
Besides bootloader and kernel, the operating system needs a
Process Image and display
root directory tree, which stores executable programs, libraries,
and system configurations. Root file system of Linux OS
contains main folders such as bin, sbin, boot, etc, dev, home, N
lib, usr, opt, root… Close device
The third part of our system is a set of libraries consisting
Y
of OpenCV and Libcurl. OpenCV (Open Computer Visison) is
an open source library developed by Intel in 1993. OpenCV End
plays an important role in research and development of
Figure 4. The flow chart of image acquisition subroutine
computer vision area. OpenCV has over 500 functions,
includes 4 parts: CxCore, CV, HighGUI, CvCam. It serves for
many fields such as security image processing, medical image

327
Graphical user interface helps dentists to manage patients which is the result of these computational stages is presented in
easier. Our Graphical User Interface includes 5 functionalities: Figure 6d. Finally, the teeth area is extracted based on the
patient information, diagnosis note box, data and time mask. After teeth segmentation process, the region outside the
management, Virtual keyboard, and data storage. teeth area is removed as shown in Figure 6e.
The dental image capturing and processing consists of
brightness, contrast, and gamma adjustment, and tooth IV. IMPLEMENTATION AND RESULT
segmentation. The dental images are acquired from a 1.3M The dental intraoral system implemented on Mirco2440
dental camera through USB port, and then stored in pixmap board using Embedded Linux 2.6.32.2 has been presented in
array supported by OpenCV library. Consequently, dental this paper. The embedded software is developed using Qt4
images can be processed by brightness, contrast, and gamma creator, OpenCV and Libcurl libraries. The experiment has
adjustment, or tooth segmentation before being displayed. been carried out on PC and Micro2440 board. The interface is
Figure 4 shows the flow chart of the image acquisition presented in Figure7.
subroutine. The brightness, contrast, and gamma adjustment is
implemented by OpenCV functions.
III. TOOTH SEGMENTATION
We propose an algorithm for tooth segmentation in order to
helps dentist easily to examine the teeth. The algorithm of the
teeth segmentation is shown in Figure 5.

Original image a) Original image b) De-noised gray image

Denoise

Active Contour without Edge

Discard noisy and small blobs

c) ACWE image d) Mask image


Fill the main blob

Dilation and Median Filter

Extract the tooth area using the filled


blob as a mask

Output image
e) Tooth segmentation image

Figure 5. The algorithm of teeth segmentation process Figure 6. The illustration images of teeth segmentation process

The original RGB image is converted to gray image to


make it effective to be processed in the subsequent stages.
Gaussian noise generally appears in the medical image, the
Weiner filter has been used to remove Gaussian noise without
affecting to the image. Subsequently, active contour without
edge (ACWE) [9] which is the method used to describe the
object’s shape in an image is employed to segment the teeth’s
contour from the de-noised image. The teeth’s contour is found
as shown in Figure 6c. Then global region based segmentation
image is calculated to locate the area of the teeth. In order to
find the full mask of the teeth, some additional computational
steps including discard noisy and small blobs, fill blobs need to
be applied to the global region image. Dilation operation and
median filter are also applied to the region image to expand the Figure 7. The GUI in Vietnamese of our application.
mask and smooth the contour of the mask. The mask image

328
The tooth image captured from camera and the tooth storage patient records and dental images. Consequently, the
segmentation image are shown on Figure 7 and Figure 8, proposed system is very useful when being applied to dental
respectively. In order to store data into the server, an clinics.
application form has been designed to ask the login
information from the user as shown in Figure 9. Figure 10 TABLE 1: A COMPARISON AMONG OUR SYSTEM
presents our system built on Friendly ARM Micro2440. AND OTHER SYSTEMS
Specification [6] [7] Our system
ARM9
CPU Computer base Computer base
S3C2440
Image size 640*480 600*450 320*240
Graphic User
None None Supported
Interface
Language Engligh English Vietnamese
Brightness,
Contrast,
Image Digital image Gamma
None
Processing processing adjustment,
Tooth
segmentation
SD Card
Storage SD Card None Flash
Network
Figure 8. The GUI with the result of teeth segmentation V. CONCLUTION
In this paper, a dental intraoral system based on
FriendlyARM Mirco2440 board and Embedded Linux was
described. The system supports various functions including
dental image capturing, patients’ information management,
image storing, and especially tooth segmentation. The dentists
and patients can observe dental images in real time on LCD
touch screen. The system is portable, economic and ready to
be applied at dental clinics. Moreover, this system can help
dentists examine at patient’s home and voyages, not in clinics.

REFERENCES
Figure 9: Configuration for uploading data on server
[1] Yong-lin LIU, Ying LIANG and Yao-yu CHENG “Design and
Implementation of Image Acquisition System Based on ARM and
Linux,” 2011 International Conference on Electronics and
Optoelectronics (ICEOE 2011), pp.49–52, 2011.
[2] Yuanping Ni, “Image acquirement with bluetooth technique under
embedded Linux environment”, 3rd International Conference on
Biomedical Engineering and Informatics (BMEI), pp. 2650-2653, 2010.
[3] OpenCV (Open Computer Vision), www.opencv.org
[4] Zheng Guo-rong, Xiong Chang-zhen, Zhang Yan, “A Method of
Embedded Video Surveillance based on OpenCV” Institute of Intelligent
Transport System of NCUT, Beijing, China, 2011
[5] Yu Fanhua, Wu Dengfeng, Zhao Hongwei, Zhen Zhiqiang “Video image
Acquisition System based on embedded Linux” 2010 International
Conference on Digital Manufacturing & Automation, 2010, pp.907-909
[6] Mini SD Card Camera System, http://kabdental.com/dental-
equipment/high-tech-equipment/tpc-advance-cam.htm
Figure 10: Demonstration of dental intraoral system on Micro2440 [7] Miharu - Dental Intraoral Plaque Detection Camera Oral hygiene
cleaning, http://www.japantrendshop.com/miharu-dental-intraoral-
A comparison among our system and other systems is plaque-detection-camera-p-71.html.
shown on Table 1. According to the table, our system has [8] Libcurl, www.curl.haxx.se/libcurl/
some advantages compared with other system. The proposed [9] T. F. Chan and L.A. Vese,” Active Contours Without Edges,” IEEE
Transactions on Image Processing, vol. 10, no. 2, Feb 2001
system is independent of computer, so it’s portable. Besides,
[10] P. Choorat, W. Chiracharit, K. Chamnongthai, “A Single Tooth
our design supports graphic user interface which helps dentists Segmentation Using PCA-Based Stacked Gabor Orientation”,
to manage patients easier. Additionally, the GUI supports International Workshop on Advanced Image Technology 2011.
Vietnamese which is very useful for domestic dentists.
Moreover, SD card, Flash, and network are supported for

329

You might also like