medical image processing and radiology … for thaiscience/article/62/10028895.pdfp. suapang, s....

P. Suapang, S. Yimman and K. Dejhan 61

Medical Image Processing and RadiologyInformation System

Piyamas Suapang 1 , Surapun Yimman 2 , and Kobchai Dejhan 3 ,

ABSTRACT

The development of comprehensive picture archiveand communication systems (PACS) has mainlybeen limited to proprietary developments by ven-dors, though a number of freely available softwareprojects have addressed specific image managementtasks. The Medical Image Processing and Radiol-ogy Information System project aims to provide anopen source, common foundation upon which notonly can a basic PACS be readily implemented, butto also support the evolution of new PACS function-ality through the development of novel imaging ap-plications and services. The project consists of fourmain software modules: 1) image order entry, whichenables the ordering and tracking of structured imagerequisitions in single frame and multiframes i.e. DI-COM, BITMAP, JPEG and JPEG2000; 2) an agent-based image server framework that coordinates dis-tributed image services including routing, image pro-cessing, and querying beyond the present digital im-age and communications in medicine (DICOM) ca-pabilities; 3) an image viewer, supporting standarddisplay and image manipulation tools i.e. DICOM,INTERFILE, BITMAP, JPEG and JPEG2000 pre-sentation states, and structured reporting i.e. Effec-tive Renal Plasma Flow (ERPF), Glomerulaar Fil-tration Rate (GFR) and Renogram; and 4) reportingand result dissemination, supplying web-based wid-gets for creating integrated reports. All componentsare implemented using Borland Delphi 6.0, PHP andMySQL to encourage cross-platform deployment. Todemonstrate the usage of this project, a preliminaryapplication supporting primary care/specialist com-munication was developed and is described herein.Ultimately, the goal of the project is to promote thewide-scale development and usage of PACS and imag-ing applications within academic and research com-munities.

Keywords: DICOM, JPEG, JPEG2000, PACS,RIS.

Manuscript received on August 10, 2012 ; revised on October1, 2012.1 The author is with the Biomedical Engineering Program,

Faculty of Sicence, Rangsit University, Pathumthani 12000,Thailand, E-mail: [email protected] The author is with the Faculty of Applied Science, King

Mongkuts University of Technology North Bangkok 10800,Thailand, E-mail: [email protected] The author is with the Faculty of Engineering, King

Mongkuts Institute of Technology Ladkrabang, Bangkok10520, Thailand, E-mail: [email protected]

1. INTRODUCTION

The increasing presence of medical imaging withinclinical care is evident: as an objective source of doc-umentation and as a means to improve communi-cation, imaging serves as a tenet of evidence-basedmedical practice [1]. Moreover, images contain im-portant biomarkers, providing in vivo snapshots ofanatomical and physiological processes. Thus, imag-ing also plays an emergent role in research, expand-ing the understanding of normal and disease states.With the growing estimates from tera- to peta-bytesof imaging data acquired annually, which will needto be compressed before sending or collecting databecause of the limited of the band width and alsolimited capacity of the data saving space [2]. There-fore, compression technology in PACS is more impor-tant part than any part in store or transmission andthe effective management of imaging data is now aparamount necessity. Unlike standard medical data,however, images pose additional, distinctive manage-ment challenges. These problems are only amplifiedin consideration of the rapid developments in medi-cal imagingranging from core technical investigationsinto new imaging modalities, to novel applications ofexisting imagingall of which are capable of marshalingan evolution in healthcare; but such changes must beintegrated into the clinical environment and researchin order to be practical. Facilitating this objectiverequires an infrastructure that can support both thecommon requirements of today imaging applications,while further serving as a springboard for future ex-plorations.

The Medical Image Processing and Radiology In-formation System project leverages earlier work bythe authors in teleradiology and medical imaging in-formation systems to provide an open source archi-tecture for a picture archiving and communicationsystem (PACS), creating a foundation for integratedimaging applications. Three considerations perme-ate the design of the project: 1) cross-platform de-velopment and deployment, leveraging both BorlandDelphi 6.0 and web-based technologies; 2) standardscompliance, focusing on archiving, compression andpresentation about digital image and communicationsin medicine (DICOM) protocols; and 3) extensibil-ity, allowing additional types of functionality (e.g.,distributed image processing, non-DICOM querying)and standards (e.g., web services) to be incorporatedin a unified manner by abstracting image-handlingprocesses.

The project framework loosely considers the mul-

62 INTERNATIONNAL JOURNAL OF APPLIED BIOMEDICAL ENGINEERING VOL.5, NO.1 2012

tiple aspects of image management in terms of work-flow, starting from the point of ordering an image, toits interpretation and use in documentation, throughto the final distribution of results. Collectively, thesoftware under this project aim to provide a common,comprehensive suite of tools that can be adapted bydevelopers to readily implement PACS-based func-tionality.

The remainder of this paper is organized as followsSection 2, the methodology and algorithm are out-lined briefly provides background on related work inthe area of PACS and imaging informatics, focusingon image acquisition, image compression, image pro-cessing and image management issues. Whats moreprovides an overview of the project architecture, fol-lowed by the design issues and implementation detailsof each major module making up the system. The ex-perimental results and discussion using this project tosupport rapid communication between primary carephysicians and specialists is described in Section 3.Finally, we conclude on the project issues, and futuredirections for this project in Section 4.

2. METHODOLOGY

2.1 Image Capture [3][4]

This research has done with collecting the imagingdata by selecting the data from video signals whichhave sent to the monitor console by using video cap-ture card (model: Life View Chip BT 8.7) processingthrough AVICAP Windows Class. These function asthe interface between Application and Device Driverto control the image receiver to collect the image sig-nal in single or multiple windows bitmap file (.BMP)type.

2.1...1 DICOM

DICOM is a standard, which has been developedby ACR/NEMA. The ACR is the American Collegeof Radiology and NEMA is the National ElectricalManufacturers Association. This standard allows dif-ferent manufacturers equipment to communicate. Forexample, all CT images are stored in the same for-mat irrespective of manufacturer and the header in-formation pertaining to the CT images going to alsobe in the same format irrespective of whose devicegenerated the image. The early standard is focus onviewing data between each other dissimilar companiesbased on point-to- point connection. Therefore, DI-COM standard with rapid development of computernetwork and PACS is used present through the con-

Fig.1:: Collecting the imaging data by using videocapture card.

Fig.2:: DICOM as standard of image format.

Fig.3:: Information Object Definition (IOD) Model.

cept controlling information in network standards asshown in figure 2. Furthermore, DICOM standardhas strong adaptability, it is necessary to adopt DI-COM standard in several professional fields. Stan-dard number has form of ”PS 3.X-YYYY”. Here, Xis a part number and YYYY is publication year. Forexample, DICOM Part 2 is ”Conformance” and doc-ument number ”PS 3.2 - 2003”. DICOM standardrules are divided into 16 parts [5].

2.1...2 DICOM Information Object Definition

DICOM specifies that image information repre-sents an Information Object which is defined in Infor-mation Object Definition (IOD) as shown in figure 3[5], and commanding word is relating to Service Classwhich is defined in DICOM Message Service Element(DIMSE).

IOD specify information for medical image whereis corresponding to patient’s name, examination type,date and it looks like a format dealing with standard-ized medical information. With these items, if thereare real values on each item, it is called InformationObject instance.

2.1...3 Data Element

DICOM communicates each attribute as shown inFigure 4 One data element communicates one at-tribute, and several data element should be necessaryfor IOD instance whole. That is, several data ele-ment must be combined as shown in Figure 4 to make


Fig.4:: Components and forming of a DICOM file.

Fig.5:: An example of a specific tag.

any person IOD instance. One data element’s detailsstructure of them is a form in Figure 4 lower columns.Figure 5 shows the example of any tag which has Tag(0010, 0010) that is consisted of two integers repre-senting Patient’s Name. VR(Value Representation)specifies the characteristics of information that hasform of PN. If value field is a name (date), it can berepresented as PN (DA). Value field used often likepatient’s name may understand even if do not clar-ify PN. It is known that this is implicit VR. Becausepatient’s name corresponds to implicit VR, it is sameeffective on the results for transmission. Otherwise,we are called explicit VR in the case of whether weshould inform the information characteristics.

Sometimes VR may be omitted. The value lengthis the number that displays how much length of datain value field and value field corresponds to data ac-tually. Here, PN means the patient name and DAmeans date.

Figure 6 shows the steps of DICOM-format imagearchive as flows 5. Firstly, data analysis and clas-sification analyze whether raw (input) data is com-pressed or other archive. Later, the data separatetext data such patients curing history and binary codedata such image data from medical instrumentation.Secondly, text data addition which is data impor-tant to the original image don’t have curing historyor data. Thirdly, decoding text data from the firststep to the stimulation (model) of DICOM format bysending the tag number into encoding module from(respective number) first to end which image data on

Fig.6:: The steps of data transformation to DICOMformat.

the last place. Finally, decoding into DICOM formattaking the tag number from the third step to com-pare with standard dictionary program, and then getthe value (number) for encoding to absolutely dataelement.

2.2 Image Compression

A medical image, that unlike an image that is usedusually, requires very high quality. For example, achest image that is acquired in Compression Ratio(CR) amounts one image size to 7 - 8 MB. Like this,when an image acquiring from various equipments isdeciphered by interpretation doctor or stores for con-servation to long term storage device etc., it shouldbe compressed in extent that do not influence on nextinterpretation. Therefore, compression technology inPACS is more important part than any part in storeor transmission.

In DICOM standard, the compression technol-ogy specify in lossy or lossless methods such asJPEG, run-length encoding, or JPEG-LS. Currently,JPEG2000 in [5] is added in new standard of DI-COM image compression. Thus, compress the origi-nal imaging data based on the JPEG and JPEG2000image compressing standard.

2.2...1 JPEG

The JPEG standard for image compression is com-prised of a toolkit that has three distinct components:baseline lossy, extended lossy, and lossless. Base-line lossy JPEG, which the most widely implementedof the three, utilizes the discrete cosine transform(DCT) to decompose an image into sets of spatialfrequency coefficients.

Figure 7 show the main procedure for all DCT-base encoding and decoding processes. The DCT isdone on an 8× 8 pixel block-adaptive basis. Baselinelossy JPEG supports 8 bits-perpixel source imagery,offers a simple quantization scheme that enables usersof the algorithm to trade off the degree of file size re-duction, i.e., compression ratio, with image quality,and utilizes sequential Huffman entropy coding. Ex-tended lossy JPEG is also based on the 8 × 8 pixelblock adaptive DCT.

2.2...2 JPEG2000 [2][6][7][8][9]

JPEG2000 characteristic can embody lossy andlossless compression at the same time in one encodedbit stream, and is shown quality of more excellentlyeminent image than existent JPEG in high compress-ibility. In addition, JPEG2000 in a sense of ROI (re-gion of interest) coding is possible, and can be ap-plied to technique of watermarking, labeling etc. forsecurity of image. Also, it has various bit depth to1 bit through 16 bits in compression and supportscompression of motion image.

Figure 8, shows a bIock diagram for JPEG2000 al-gorithm. The strength of JPEG2000 is that it is capa-ble of tiling unlike JPEG (Figure 9). Since JPEG pro-cesses 8x8 subimage tiling and DCT (discrete cosine


Fig.7:: DCT-based encoder and decoder simplifiedblock diagram.

Fig.8:: A block diagram for JPEG 2000 encoder anddecoder.

transform), it is tended to block shape artifact in highcompression ratio, On the other hand, JPEG2000 isadvantage that can enhance quality of image or re-duce use of memory controlling by tiling of imagerandom. Also, DCT in JPEG changes an image tocharacteristic of frequency, but JPEG2000 is scalableto an image by scale or resolution because DWT (dis-crete wavelet transform).

Recently, moving picture compression of medicalimage was discussed at MPEG in 2002 but that issuewas not decided yet as DICOM base standard in 2003apart Moving picture present in PACS means thatfirst several still image were captured, multiframe DI-COM images were generated followed by animated.

Fig.9:: Tiling, DC level shifting and DWT of eachimage tile component.

3. ROI BOUNDARY DETECTION BASEDON GEOMETRIC ACTIVE CONTOURMODEL [10]

A region of interest (ROI) is a portion of an imagethat you want to filter or perform some other oper-ation on. You define an ROI by creating a binarymask, which is a binary image that is the same sizeas the image you want to process with pixels thatdefine the ROI set to 1 and all other pixels set to0. Which can define more than one ROI in an im-age. The regions can be geographic in nature, suchas polygons that encompass contiguous pixels, or theycan be defined by a range of intensities. In the lat-ter case, the pixels are not necessarily contiguous.Thence, ROI boundary detection based on geometricactive contour model for sets ROI semi-automaticallyon region of expected disease which is used in this re-search. Active contour model consist internal forceand external force from edge detection based on vari-ance image and Canny edge detection. Finally, com-parison time and efficiency of ROI processing betweenvariance image and Canny edge detection.

4. FUNCTIONAL REQUIREMENTS

Free DICOM viewers such as ezDICOM [1112]which runs on Windows computers. Available as astandalone Windows program or as an ActiveX com-ponent (allowing plug-and-play use with Delphi, Vi-sualBasic, C#, Visual C, Internet Explorer and otherActiveX aware programs). It is able to display mosttypes of DICOM image (many other viewers are lim-ited to showing uncompressed grayscale DICOM im-ages) and can automatically detect and open Analyze,DICOM, Genesis, Interfile, Magnetom, Somatom andNEMA images. ezDICOM is part of the Source forgecommunity. This programs have weaknesses in pro-viding facilities for image processing and image anal-ysis. The one of main purpose of this research is tofacilitate physicians to display, processes and analyzesthe images easily.

Essential functional requirements in parts of DI-COM viewer, image processing and image analysis arelisted in the below. These requirements are obtainedfrom physician and technical staffs in the radiologydepartment of hospital in Thailand.

⊙ Display various standard formats of digital im-age data both single frame and multiframes: suchas DICOM, INTERFILE, BITMAP, JPEG andJPEG2000.⊙ Zoom in/out: zooming full image or partial imagein and out for better image analysis.⊙ Contrast/Color map: adjusting the contrast of adim image by reassigning pixel brightness levels. Lin-early adjusting the contrast range (by assigning thedarkest pixel value to black, the brightest value towhite, and each of the others to linearly interpolatedshades of gray) makes the better use of the displayand enhances the visibility of features in the image.Color mapping is applied to image according to the


difference of the image pixel value.⊙ Quantitative Analysis: the quantitative analysisis to obtain average, maximum, minimum, dispersionand standard deviation of the pixel values in a ROIthat can be set up in various forms such as rectangle,ellipse, line and polygon. More than one ROI canbe set in a single image. Thence, Semi-auto ROIsetting that sets ROI semi-automatically on region ofexpected disease.⊙ Computer-aided diagnosis program: such as calcu-lation of Glomerular Filtration Rate (GFR), EffectiveRenal Plasma Flow (ERPF), and plot of the TimeActivity Curve or Renogram.

5. OVERVIEW OF MEDICAL IMAGE PRO-CESSING AND RADIOLOGY INFOR-MATION SYSTEM

In Addition, a Radiology Information System(RIS) can be defined as a system that provides medi-cal image services over long distance. The RIS serviceis one of the newest medical services that providestransmission of medical image and information for di-agnosis, consultation, medical treatment and healtheducation by using a communication network suchas local area network, wireless network and inter-net. Thus, demands on an effective RIS have beenever increasing. With the advent of computer andcommunication technologies, construction of an ef-fective RIS system becomes possible. Medical imagedatabases play an integral role in a digitized clini-cal environment starting from storing of electronicpatient records, scanned medical images generatedby Modalities like X-Ray and MR to reporting andbilling. The operations in the clinical environmentsuch as image storage and retrieval, viewing, and postprocessing can be served by a reusable database com-ponent using which different modalities can developtheir own clinical applications [12].

Component of the Medical Image Processing andRadiology Information System project can be used bythe various modalities for reusable storage as shownbelow in figure 10.

The medical database storage component can beconsidered to be made up of two parts: 1) Data Ac-cess Layer 2) The underlying storage media. Themedical database storage component exposes its func-tionality through the data access layer as shown inthe figure 11. The data access layer of the medi-cal database repository implements a published set ofdata access interfaces to store, retrieve, modify andquery the data present in the underlying repository.These interfaces are concerned with access and trans-fer of medical data. The data access layer abstractsthe underlying storage media, which in our case hap-pens to be a commercially available RDBMS from theclient.

The architecture of the reusable medical databasecomponent gives a provision where in the lowest layer,which accesses the storage media, can be a plug-incomponent in itself. This gives a provision for vari-

Fig.10:: Reusable storage component.

Fig.11:: Data access layer.

ous modalities to plug in their specific storage access-ing mechanism based on the commercial off the shelfdatabase they opt as shown in figure 12. In addi-tion to the performance tuning done at the RDBMSlevel, it is also important that the data access layeris designed for good performance. One of the keyhigh performance use cases of any medical databaseis the fast rendering of a medical image. Medicalimages contain pixel data and it is very importantto ensure that this pixel data is stored for extremelyeasy and fast access. In our database component, thepixel data is stored in a separate file on the file systemwhilst the Meta information is stored in the database.This gives a dual benefit:1) Fast query of the Meta information as the data isstored in the database.2) Fast access to pixel data given that the pixel datais stored on the file system and file IO performancecan be maximized when the CPU is busy processingthe meta information.

Medical images comprise of pixel data. This pixeldata can be stored by modalities on different hard-ware configurations. The architecture of medicaldatabase storage component gives a provision where


Fig.12:: Modality specific Plug-in.

in the modality could choose to plug-in its own pixelhandling plug-in. As with off the shelf databases, thecomponent is provided with a default implementa-tion of the pixel plug-in which can readily deployedon a Windows based operating system in any modal-ity. The component also has the provision of workingwith customized schemas in a modality. As notedearlier, certain modality might already have an ex-isting schema on which certain specific applications(like for e.g. reporting) might be using. In such sit-uations, in order to make our component work withmodality private schema, Database Views are used.The modality owners can then provide the mappingfrom views to private tables. Having a provision forthe modality to develop their own accessor plug-incomponent that understands the underlying schemafor storing data provides this flexibility.

Medical images stored in the DICOM formatwould contain redundant information. For example,let us say 10 CT Images are acquired out of a CTscan. Each of these 10 images in the DICOM formatwill contain the same information with respect to thepatient details such as patient name, patient Id, birthdate, and study date, study time, series modality in-formation. It is good to avoid this redundancy andfor this the data must be normalized and stored inthe repository. On normalization, the patient infor-mation, which is identical in all the DICOM files, isstored as a single entity in the Patient table in thedatabase. Similarly, the study and series informa-tion, which are identical in all the DICOM files arestored as just two entities in the respective Study andSeries tables.

6. RESULTS AND DISCUSSION

The Medical Image Processing and Radiology In-formation System project by using Borland Delphi6.0, PHP and MySQL for image archive, image com-pression, image processing, image analysis and imagetransmission. To control the image capturing throughthe image capturing device, the program can connect

Fig.13:: The facilities in a part of image archiving.

to the hardware and bring video signals from the ra-diological modality. Then collect the imaging datafrom the video signal. The collection was sent to thesingle frame, 15 image/sec rated multiframe and fixedmultiframe which could fix time range and frame rateradiological modality inWindows Bitmap file format(.BMP) without changing of image shape. The facil-ities in a part of image archiving as shown in Figure13. For converting data to the DICOM data con-struction, the program can convert the data to theDICOM data construction after both various imagecollecting process and also automatic image collectingby converting without changing the image point dataor not changing any the image data in the process.

The image compression, the program can compressfrom collect the imaging data in single frame file for-mat and multiframe file format before compress toJPEG and JPEG2000 by having the same compress-ing ratio as shown in figure 14 and 15. The signif-icant advantage of JPEG2000 over normal JPEG isthat the error from JPEG2000 compression is smallerthan the error from JPEG. Moreover, comparison ofan image quality, a general evaluation tool, RMSE(Root Mean Square Error), has been adopted. TheRMSE can be written by following.

RMSE =

[1

MN

M−1∑x=0

N−1∑y=0

{f(x, y)− g(x, y)}2] 1

2

(1)

Table 1 and figure 16 shows the comparison re-sults based on compression ratios with JPEG andJPEG2000. Which show that compressing image inJPEG2000 has less error than normal JPEG com-pression. Both compression processes will increasethe number of errors when the compressing ratio hasincreased.


Table 1:: The RMSE Comparison JPEG andJPEG2000.

compression ration15:1 35:1 65:1

compression standardImage .jpg .jp2 .jpg .jp2 .jpg .jp2

Single Frame 2.29 1.60 4.77 3.08 15.14 4.82Multiframe 3.81 2.71 5.29 3.75 21.14 6.19

Fig.14:: JPEG and JPEG2000 single frame image(compression ratio 23:1).

As shown in Figure 14, 15, 16 and Table 1, theoverall performance based on JPEG2000 shows bet-ter result than JPEG. The system have developed aprototype in parts of DICOM viewer, image process-ing and image analysis. Figure 17 shows the GUI ofDICOM viewer a part of this system. As depicted infigure 17, frequently used function are provided in theform of ICONs. Also, at image loading time, the pro-gram analyzes the DICOM 3.0 header and displaysthe header information with the corresponding im-age. In addition, the viewer has an image processingand image analysis tool bar that can be docked [12].

Functions provided with prototype can be summa-rized as follows: (1) DICOM viewer which can dis-play various standard formats of digital image databoth single frame and multiframes such as DICOM,INTERFILE, BITMAP, JPEG and JPEG2000. Asdepicted in figure 17. (2) Image processing whichhas also the capabilities to zoom in/out and con-trast/color map. (3) Image analysis which deter-mine the pixel values in a ROI that can be set upin various forms such as rectangle, ellipse, line andpolygon. Thence, Semi-auto ROI setting that setsROI semi-automatically on region of expected dis-ease. Additionally, more than one ROI can be setin a single image. As depicted in figure 18. (4) Nu-clear medicine diagnosis which are more functions on

Fig.15:: JPEG and JPEG2000 multiframe image(compression ratio 68:1).

Fig.16:: RMSE comparison curve.

Fig.17:: DICOM viewer.

nuclear medicine for renal study such as calculationof Glomerular Filtration Rate (GFR), Effective RenalPlasma Flow (ERPF), and plot of the Time ActivityCurve or Renogram. As depicted in figure 19, 20 and21.

These requirements are obtained from physiciansand technical staffs in the radiology department ofhospital in Thailand. As depicted in figure 16.

Moreover, time of ROI processing and efficiencybetween variance image and Canny edge detectionare compared. The results shown that edge detec-tion based on variance image have less time of ROIprocessing than edge detection based on Canny edgedetection but have nearly efficiency of ROI process-ing both variance image and Canny edge detection.This research was tested with the DICOM 3.0 stan-dard image compared the APEX -XPERT Program.The result showed that no significant difference canbe found (= 0.05).

Figure 16 (a) depicts the architecture of thisproject. The DICOM server is connected to local/re-mote image acquisition system. Upon acquiring animage, the server stores the image in the DICOMv 3.0 format by using the DICOM encoding mod-ule. The stored image can be displayed by using theDICOM decoding module. The server also includesmodules for image compression and DICOM viewer.


Fig.18:: Time Activity Curve or Renogram.

Fig.19:: The facilities for defination of personalvalue.

Fig.20:: Calculation of Effective Renal Plasma Flow(ERPF).

Fig.21:: Calculation of Glomerular Filtration Rate(GFR).

Fig.22:: Basic structure and sequence of interactionsbetween clients and server in the system.


Clients (a personal computer, a notebook or aPDA equipped with a Web browser) can downloadand install the DICOM viewer from the DICOMserver via WWW. From then on, clients can beselected and opened single frame and multiframes,each of which can exhibit information in .dcm fileformat and digital image processing based on localcontrast enhancement, adaptive interpolators tech-niques, colour transformation and cine loop. Also,for a situation in that a client wants to discuss withother client, the server provide facilities for makingWebboard.

Finally, The results of functional testing of the con-structed the project shows that a client can connect tothe DICOM server throughWWW, authentication byusing a login processing is performed. The client canexamine the list of image stored in DICOM server,select the desired image and download the selectedimage. When a client issues a command, then thecommand is transferred to DICOM server throughthe Web server.

After, DICOM server finished the requested job,the result image is sent to the client through theWeb server. In the event of image display and im-age processing is performed by DICOM viewer whichthe client can download the DICOM viewer in theform of plug-in from the server.

7. CONCLUSION

The Medical Image Processing and Radiology In-formation System project is designed and developeda prototype running on personal computer under theWindows OS. The prototype provides various func-tion for image archive, image compression, image pro-cessing, image analysis and image transmission. Fur-thermore, the system has been developed and pro-vided medical image services over long distance whichshowed the usefulness of our approach. Past worksin imaging and open source projects have remainedfragmented, only offering niche solutions. Thus, de-velopers are often left with the task of reinventingor integrating dissimilar software components; ulti-mately, it is hoped that this project, as an umbrellaframework for all these efforts, can serve as startingpoint to foster new developments in PACS and theuse of imaging in support of evidence-based medicalpractice, research, and education.

References

[1] R. Wootton, N. G. Patil, R. E. Scott, K. Ho,“Telehealth in the Developing World.,” Royal So-ciety of Medicine Press Ltd., 2009.

[2] P. Buonora, F. Liberat, “A Format for DigitalPreservation of Images: A Study on JPEG 2000File Robustness.,” D-Lib Magazine, Vol. 14, No.7-8, JulyAuguet, 2008.

[3] O. Le Meur, A. Ninassi, P. Le Callet, D. Barba,“Do video coding impairments disturb the visualattention deployment.,” Elsevier, Signal Process-

ing: Image Communication, Vol. 25, Issue 8, pp.597-609, 2010.

[4] S. Ardekani, U. Sinha:, “Geometric distortion cor-rection of highresolution 3T diffusion tensor brainimages,” Magn. Reson. Med., Vol. 54, No. 5, pp.1163-1171, 2005.

[5] M. D. Adams, “A Flexible Content-AdaptiveMesh-Generation Strategy for Image Representa-tion,” IEEE Transactions on Image Processing,Vol. 20, No. 9, pp. 2414-2427, September, 2005.

[6] OFFIS Computer Science Institute, “DCMTKDI-COM Toolkit main website, [Online],” Available:http://dicom.offis.de/dcmtk, March, 2012.

[7] Van Der Knijff, Johan, “TIFF to JPEG2000: re-port on migration and characterisation experi-ments.,” National Library of the Netherlands re-port, 2010.

[8] Van Der Knijff, Johan, “JPEG 2000 for Long-term Preservation: JP2 as a Preservation For-mat.,” D-Lib Magazine, Vol. 17, No. 5-6,MayJune, 2011.

[9] M. D. Adams, D. Xu, “Optimal Design of High-Performance Separable Wavelet Filter Banks forImage Coding.,” Signal Processing, Vol. 90, No.1, pp. 180-196, January, 2010.

[10] Auli-Llinas, F., Serra-Sagrista, J., Bartrina-Rapesta, “Enhanced JPEG2000 quality scal-ability through block-wise layer truncation.,”EURASIP Journal on Advances in Signal Pro-cessing, Vol.2010, 2010a.

[11] K. Johnson, “A descriptive processmodel for open source software devel-opment.,” M.Sc. thesis, Dep. Comput.Sci., Univ. Calgary, [Online]., Available:http://sern.ucalgary.ca/students/theses-/KimJohnson/toc.htm, June, 2012.

[12] I. Drnasin, G. Gogic, S. Tonkovic, “Interac-tive Teleradiology.,” World Congress on MedicalPhysics and Biomedical Engineering., September712, 2009.

Piyamas Suapang received the B.Sc.Physics from Nareasuan University inPhitsanulok, Thailand in 1999 andthe M.Sc. Medical Insrument fromKing Mongkuts University of Technol-ogy North Bangkok in Bangkok, Thai-land in 2004. Since 2009, she has stud-ied in the Doctors degree program atthe Electrical Engineering Department,Faculty of Engineering, King MongkutsInstitute of Technology Ladkrabang inBangkok, Thailand. In the past ten

years, she has focused my research on medical image process-ing. Moreover, her current research includes computer vision,pattern recognition and computer-assisted diagnosis (CAD) inmedicine.


Surapun Yimman received the M.Eng.in Electrical Engineering from KingMongkuts Institute of Technology NorthBangkok, Bangkok, THAILAND, in1999. and D.Eng in Electrical Engi-neering at King Mongkuts Institute ofTechnology Ladkrabang in 2008. Hisresearch interests are digital signal pro-cessing and applications.

Kobchai Dejhan was born in 1955in Petchburi, Thailand. He receivedB.Eng. and M.Eng. degree in ElectricalEngineering from Faculty of Engineer-ing, King Mongkuts Institute of Tech-nology Ladkrabang (KMITL), Bangkok,Thailand in 1978 and 1980, respec-tively and the Docteur from Ecole Na-tionale Superieure des Telecommuni-cations Paris (ENST/Telecom Paris),France in 1989. Since 1980, he has beena faculty member of KMITL. He is cur-

rently an associate professor at Telecommunication Engineer-ing Department and also works as a dean of Engineering Fac-ulty. He also works for Research Center for Communicationand Information Technology (ReCCIT) at KMITL as a chiefof Communications Circuit Designs laboratory. His researchinterests are in area of Communications Circuit Designs, Sig-nal Processing, VLSI and CMOS Integrated Circuit Design.He is a member of IEICE, Japan and senior member of IEEE,USA.

medical image processing and radiology … for thaiscience/article/62/10028895.pdfp. suapang, s....

Documents