Radiotherapy Support Tools, the Brazilian Project: SIPRAD

Diego Fiori de Carvalho

, José Antonio Camacho Guerrero

, Luis Javier Maldonado Zapata

Andrey Omar Mozo Uscamayta

, Heleno Murilo Campeão Vale

, Leandro Federiche Borges

Alexandre Collelo Bruno

and Harley Francisco de Oliveira

I-medsys, Innovative Medical Systems, Ribeirão Preto, São Paulo, SP, Brazil

Clinical Hospital at Ribeirao Preto, University de Sao Paulo, USP HCRP, SP, Brazil

Keywords: Radiotherapy, RTPS, SIPRAD, Portal 2D, the Fusion of Images.

Abstract: The radiotherapy planning process (teletherapy) is initially performed by the acquisition of Computed

Tomography images of the areas of interest to guide a series of health professionals in the work of vector

design of regions of interest for protection (risk organs) and radiation (tumors). All these steps are

performed using computational tools that extrapolate measurements and scales in the treatment plan. The

efficiency of the treatment depends on the recreation of the patient's positioning on the linear accelerator

stretcher with the previously acquired tomography images. For this, in this article, we present three modules

of the SIPRAD (Information Systems for Radiation Therapy Planning) project. With the name of

Radiotherapy Portal it is able to perform a fusion of planar images of the target region, made on the day of

treatment, with the digital recreation (DDR - Digital Reconstructed Radiographs) of this radiograph

generated from the Tomography of treatment planning, aiming to improve the reproducibility of the

positioning that the radiation dose delivered during all the radiotherapy treatment. The second module

named by LYRIA PACS RT provides a client/server architecture for storing, distributing and displaying

images from any systems using the DICOM RT Struct, Image, Plan and Dose modes. The third module

called Contouring is responsible for the training of new radiotherapists.

1 INTRODUCTION

There are several difficulties in radiotherapy

planning in latin american because few hospitals

have solutions that have a complete planning system

for the radiotherapy treatment.

In Brazil cancer treatment reality patients who

treat in the public system, only 16% start the

procedure within 30 days; the average waiting time

is 113.4 days. According to the Brazilian National

Nuclear Energy Commission (CNEN), there are 371

linear accelerators (LINAC) in operation in Brazil,

of which 260 are in the public health network. This

means that there is at least a 40% shortage in the

supply of machines. 55% of LINACs are located in

only 4 states (27 states in all), 70% is the average

yield of a LINAC due to a lack of professionals and

maintenance and operation processes (SBRT, 2018).

Besides the lack of LINACs, another problem is

related to a shortage on expert systems linked to the

Brazilian reality of treatment.

The patients' positioning and location of the

target regions (tumors) at the time of treatment,

which is fundamental in the efficacy of the

treatment, is carried out in an artisanal manner. In

some hospitals (at least 80%), it is attempted to

guarantee the reproducibility of the positioning

through the use of molds, and ink markings, as a

kind of tattoo for temporary marking of the correct

region of radiation application. Recently the Federal

Council of Medicine has chosen to recommend the

use of radiotherapy planning software generating

demand for this solution. It was necessary to build a

solution to increase the accuracy and improvement

of the radiotherapy treatment process in public and

private hospitals in Brazil, creating a robust product

and cost alternative with the Latin American reality

for such systems. The SIPRAD project (Information

Systems for Radiotherapy Planning) aims to build a

series of software for this urgent Brazilian need to

evolve the demarcation process and improve

treatment.

Fiori de Carvalho, D., Guerrero, J., Zapata, L., Uscamayta, A., Vale, H., Borges, L., Bruno, A. and Francisco de Oliveira, H.

Radiotherapy Support Tools, the Brazilian Project: SIPRAD.

DOI: 10.5220/0007482901370143

In Proceedings of the 12th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2019), pages 137-143

ISBN: 978-989-758-353-7

137

In addition to creating a unique architecture on

the regional scene, SIPRAD is interested in building

interoperable systems with all linear accelerators

(LINAC) in the market as well as its proprietary

planning systems. Thus, communication between all

the radiotherapy treatment workflow entities are

integrated and operating together.

The SIPRAD project is currently in final

development, with tests in Brazilian hospitals. Some

modules are under analysis by local and

international health surveillance certifiers.

Among the SIPRAD solutions that were

developed: a contour design application, a PACS

RT, a radiological scheduling system integrated to

the RRIS radiotherapy and the Radiotherapy Portal.

For a 2D Portal system, the digital radiographic

image (portal image) is acquired from each patient

before the moment of radiotherapy treatment. The

portal image is then fused with a reconstructed

digital radiograph from the anterior tomography

allowing for corrections of its positioning (Maria Y

Law, 2009).

2 SIPRAD

The SIPRAD (Information Systems for

Radiotherapy Planning) is a tool for managing and

controlling the flow of radiotherapy treatment.

Provide an intuitive and efficient way to define all

steps of components suitable for a public or private

clinic or hospital (Carvalho, 2018).

SIPRAD has access via patient data entry and

Computed Tomography scan in the axial plane. The

system stores the system inputs and presents

specialized interfaces for each type of end user

according to their needs. From a computational

architecture point of view, SIPRAD can be

presented separately as a Client/Server (front

end/back end) system. The green parts in Figure 1

represent the back-end (server-side) solutions, while

the pink solutions are front-end (client-side)

solutions. For each service offered to medical or

patient clients, there is a specialized server

containing services hosted in clinics, hospitals or

data centers (cloud). On the back end side, we have

the RIS RT and Lyria PACS RT (Radiology

Information System for Radiotherapy) that are

directly connected to the workflow module. The 2D

Portal and IGRT are fundamental for the

improvement of the accuracy/quality in the

radiotherapy sessions. The Plotting and Simulation

tools are linked directly to the Contours module. The

Planning and Management Software is already

linked to the planning and management modules

respectively. Also, finally, the Onco APP is a way of

accessing patients to the critical items of their

treatment. The Onco APP is an extension of the

workflow module aimed at a restricted view of

information within the interest of the user to the type

of patient.

The system stores the inputs and presents

specialized interfaces for each type of end user

according to their needs. To facilitate its

interoperability and communication actions the

DICOM (DICOM, 1999) standard is used. More

specifically the DICOM RT (Maria Y. Law, 2009)

standards that are used: Dose, Plan, Image, and

Struct.

Figure 1: Architecture of SIPRAD.

For better understanding and separation of the

DICOM RT storage actions, the SIPRAD was

divided into four modules. The four modules of

SIPRAD are Workflow, Contours, Planning, and

Management. Workflow is the largest module; it is

present in practically the entire SIPRAD solution.

The Workflow module has the role of extracting the

patient data from the Hospital/Clinic worklist by

using the Lyria PACS (Carvalho, 2015) worklist

solution and creating interfaces of all the system

registries, including those of patients. It also

monitors the patient during the treatment process at

the hospital and brings their agenda of radiotherapy

appointments and sessions, also allows patient event

logs and reports. Workflow actions are stored in

SQL like database.

The Contour module has DICOM with Query

/Retrieve communication with Lyria PACS

(Carvalho, 2015) and performs Computed

Tomography scans (CTs) of the patient making them

available, through a specialized interface, the

BIOIMAGING 2019 - 6th International Conference on Bioimaging

138

contouring of the areas of interest (GTV –Gross

tumor Volume, CTV- Clinical Tumor Volume, and

PTV - Planning Tumor Volume) and OARs (Organs

at Risk). After the delimitation of these important

structures for their treatment, the data is transferred

to the Planning module.

The Planning module is responsible for the

creation of Isodoses, BEV (Beam Eye View), DVH

graphs (Dose Volume Histogram), and areas to be

irradiated in the radiotherapy treatment. Still, in the

Planning module, the user has a friendly interface

available that interacts with 3D objects to facilitate

the process. It also has the possibility of printing

data for the assignment to the corresponding person

in charge according to the process of the institution.

In Figure 2, you can view the layout of the Contours

module.

Figure 2: Contours Module Delimitation System.

In Figure 2, we can visualize the interface of the

delimitation system; basically, we have a Computed

Tomography scan that doctors and physicists

perform the generation of contours of areas of

interest utilizing specialized delimitation tools which

generate vectors. The Management module is

responsible for the creation of molds and filters that

are inserted in the particle or linear accelerator

(LINAC). This module is also responsible for

communicating with the linear accelerator to

perform the exam. It is also possible to send data to

the radiologist technician who performs the

treatment session, such as the table position in X, Y,

Z, and gantry angle of irradiation of the apparatus.

3 2D PORTAL

3.1 Marking Process

Before and on a weekly basis during the treatment,

2D orthogonal (anterior-posterior and lateral-lateral)

radiographic images of the target region are

acquired. The images are acquired with Cross-Hair,

an accessory that generates a scale with the origin at

the center of the target and provides real

magnification, recorded with patient information and

saved in DICOM format. The same coordinate axis

scale is generated digitally in the digital

reconstruction of the orthogonal 2D radiographic

images (DRR). These scales are of fundamental

importance for the process of comparison of the

radiographic images, since the coincidence of the

origin of the scales, the relative distances of the bone

structures with the scales are the same parameters

used to confirm the exact reproducibility of the

location with the planning, this is maintained during

treatment. As the radiotherapy today is mostly

isocentric (the target lies in the center of the axis of

rotation, and the radiation source is around the

patient) the comparison of orthogonal 2D

radiographic images with the reconstruction of the

same from the planning tomography (DRR),

increase the accuracy of planning reproducibility

based on bone marking. The overlapping of these

images (radiographs and DRRs) through a digital

fusion facilitates the comparison of them by the

responsible doctor. This analysis is based on the

coincidence of the coordinate axis scales, with the

DRR being the reference image for the comparison.

The time for demarcation and analysis with this

tool facilitated the process and increased accuracy. It

is estimated that 28% of the time was gained with a

60% increase in accuracy for testing in 100 adult

patients.

3.2 Computer Tool

Initially, for better control of the axial, sagittal and

coronal images of the project, it was necessary to

implement the MPR algorithm (Multiplanar

Reconstruction), enabling the 2D Portal to generate

new planes interpolated from the default axial plane

from the acquired tomography examination. This

computational process of generation of new

interpolated planes has algorithmic complexity O

(n³). Thus, it was necessary, the implementation of

computational parallelism routines with adapted

programs to be processed to user threads in Java.

Once the MPR was stabilized, the stage of fusion

of radiological images implemented according to the

needs of the medical team was carried out, in order

to facilitate the patient's positioning and image

acquisition process at the time of planning, thus a

specialized interface was developed to perform

fusions images (CRs, DRs, and DXs) with

Computed Tomography (CTs) images. In Figure 3,

we can visualize the interface created for this action;

we can visualize in “A” the target with the sagittal

Radiotherapy Support Tools, the Brazilian Project: SIPRAD

139

positioning of a head/neck x-ray with the patient

already in the radiotherapy position.

Figure 3: Fusion of X-ray Images and DRR (Computed

Tomography).

In "A" of Figure 3, the dark rectangular region

corresponds to the holes of the lead block in the

positioning present in the gantry of the linear

accelerator. In this phase, it is important to calibrate

the quantity and pixel positioning of the DRR

(Digitally Reconstructed Radiographs) image scale.

In "B" of Figure 3, we can see a rotation of the block

concerning the sagittal plane. The user can rotate,

translate, zoom in and out to make it easier to

position.

To facilitate the visualization of two different

images in real time, a matrix visualization feature

(2x2) was created in which the main or secondary

diagonal is chosen to visualize the image that is in

the bottom layer; we call this action of "Fusion

Division". These options facilitate the change of

transparency between the overlaps made in the

fusion processes. From the mouse actions, half of

the DRR (the lightest region) and the remainder of

the standard radiography (darker region) are

visualized as shown in Figure 4.

After conferencing this placement, the system

reports the differences of distances in millimeters

between the (virtual) study model and the (actual)

patient at the moment before the radiation. In this

way, the technician can adjust the patient in the

correct position to be irradiated in LINAC.

Figure 4: Transparency matrix view for merged images.

4 LYRIA PACS RT

Strong customer demand on the national scene, and

not contemplated by any market solution currently,

refers to obtaining a PACS RT. This type of system

is independent of the architecture acquired by the

clinic or hospital that has radiotherapy services, as it

does the storage, viewing, and distribution of the

examinations of any system brands as long as they

work with the standard DICOM RT. More

specifically, the Lyria PACS RT (Picture Archiving

Communication System for Radiotherapy Planning)

is an extension of the Lyria PACS system (Carvalho,

2015) and works on the backend side (server)

responsible for the following services:

1. Storage of DICOM RT (Thiruchelvam, 2005)

images in magnetic or optical media;

2. Connections for health services and informatics

departments (HL7, XML, SGBDs via web services,

and others);

3. Recovery of images in the short or long term;

4. Viewing images in remote diagnostic stations;

View of DICOM RT (DICOM RT, 1997) structures

in a simplified and responsive universal

environment, in this case, web.

5. Image-friendly interfaces (Web and desktop

clients);

6. Fast and secure communication via computer

networks;

7. Patient Worklist Service and communication with

other equipment;

8. Interoperability for other systems through the use

of Web Services;

The C-ECHO SCP / SCU (ping) and C-STORE

SCP / SCU (storage) protocols are obtained directly

from Lyria PACS (Carvalho, 2015). Also, it was

necessary to extend the DCM4CHEE (Max, 2007)

library in the Java programming language by

BIOIMAGING 2019 - 6th International Conference on Bioimaging

140

determining a storage hierarchy of the DICOM RT

structures of the project on the servers.

The implementation of item 2 was performed via

integration with a SQL like database. The creation of

the resource of item 3 was done through a service

called Query/Retrieve, where we can create DICOM

entities called AETitles that have an IP (Internet

Protocol) and a port registered in the database. An

example of this action can be seen in Figure 5. In

this process, the equipment can exchange exams and

content information with each other.

Figure 5: SIPRAD project worklist integrating RIS RT and

PACS RT.

The implementation of items 4 and 5 occurred

with the integration of the SIPRAD contour interface

to search for exams with the DICOM RT protocol

(Maria Y Law, 2009). Item 5 refers to the friendly

PACS server interfaces, an example of the PACS RT

web interface can be seen in Figure 3. This web

interface is only for reading and manipulating PTZ

(Pan, Translate and Zoom). In Figure 3, it is possible

to visualize the main phases of the radiotherapy

treatment performed by the VARIAN Eclipse or

Aria system. This exam presented in Figure 6 was

obtained from the integration with a VARIAN server

of the HCRP (Hospital das Clinicas de Ribeirão

Preto).

Figure 6: Lyria PACS RT web interface.

In Figure 6, we can visualize on the left side the

DVH (Dose/Volume Histogram) graph in the center,

a tomography with the isodoses curves and the right

the structures plotted on a tomography in the axial

plane. To perform the visualization of structures, it

was necessary to parser many proprietary tags, in

addition to the tags common to the DICOM RT

standard. Two examples of proprietary tags from the

Finnish company VARIAN can be seen in Table 1.

Table 1: Proprietary tags.

Tag

Data

(3285,0

010)

[0034] [Varian Medical Systems

VISION 3285] Private Creator

Data Element

(3285,1

000)

[0066][\FE\FF\00\E0:\00\00\00\852\

10\00"\00\00\00 Varian Medical

Systems VISION 3285\852\

01\10\08\00STANDARD]

In TCP / IP communication between servers and

computers is done using the RESTful API and

HTTPS protocols in user authentication (item 6).

Figure 7 shows a complete worklist (item 7)

implemented for the HCRP hospital structure, where

it is possible to see a list of anonymous patients.

This worklist is generated in the morning of the day

containing the schedules of patients obtained from

the Workflow module. After the list is generated, it

is sent to the radiological acquisition equipment and

linear accelerators that are registered as DICOM

entities. Item 8 refers to interoperable integration

with other systems made according to the DICOM

standard version 3.0.

This system-to-system communication

functionality is paramount for PACS RT testing in

clinics and hospitals communicating with RIS and

HIS (Hospital Information System) systems.

Figure 7: Worklist of Lyria PACS RT.

5 CONTOURING TRAINING

TOOL

The contouring training tool is a version of RTPS

capable of measuring the degree of similarity

between a contour performed by a student and

another contour created by an experienced doctor

(teacher). Figure 8 shows the process of a typical

Radiotherapy Support Tools, the Brazilian Project: SIPRAD

141

test using the Training Contour Tool. In the first step

is to teacher choose exams to determine the contours

of OARs and TVs. After that, the teacher draws the

contours of the answer sheet. In the second step,

student's log in the system to create free contours in

the axial CT. In the third step, these students create a

fill contour for all OARs. In the fourth step, the

students create the drawings and fill of the TVs.

In the fifth step, the student analyzes his result in the

tool and can improve contours for only 2 times

(evaluation rule). Finally, in the last step, the tool

will generate students’ contours similarities and

calculate the student´s grade.

Figure 8: Process Tests Tool Contours.

In the process the Jaccard index is used, also

known as the Jaccard's coefficient of similaritiy, is a

statistical component to compare similarity and

diversity between two sets. The Jaccard index is

defined by the size of the intersection of two sets,

divided by the size of the union of the same sets

(Jaccard, 1901).

In this study it was used a structure (ROI)

represented by a set of spatial points (x, y, z). These

points are drawn in each axial section of the CT.

Each structure is represented by a set of many spatial

points, where each subset belongs to each cut of the

digital tomography. Each subset is represented by a

set of spatial points that form a convex figure

representing only the outline of the ROI structure.

The Figure 9 presents the Jaccard index in left and

the image results in right. The green region refers to

the area that the student did correctly, the yellow

region refers to the area that the student did not

draw. And the red region refers to the area that the

student has drawn wrong.

Figure 9: Training Contour Tools Interface.

6 CONCLUSIONS

The 2D Portal SIPRAD demonstrated precision in

the determination of the displacements from the

fusion when tested at Clinical Hospital at Ribeirao

Preto (HCRP-USP), and it found greater agreement

among the users on the positioning of the patient

when compared to the fusion with visual analysis.

Further testing with experts is required to verify the

accuracy of the process by comparing the results

with the existing models.

The Lyria PACS RT and Training contour tool is

currently being tested in two hospitals in the state of

São Paulo (INRAD in the capital and HCRP in the

interior) and an Australian cancer hospital (Illawarra

Cancer Care Center). Both hospitals have equipment

and systems from the VARIAN and ELEKTA

companies.

The SIPRAD project is currently in final

development, with tests in Brazilian hospitals. Some

modules are under analysis by local and

international health surveillance certifiers.

The cost of SIPRAD will be approximately USD

$10.000 for the basic solutions of the radiotherapy

planning process according to SUS (Brazilian Public

Health System).

In these tests, we will be able to analyze the

operational data loads in a real environment to be

presented in future work.

ACKNOWLEDGEMENTS

Clinical Hospital at Ribeirao Preto (HCRP),

University of Sao Paulo and FAPESP, Process

2016/19854-8.

BIOIMAGING 2019 - 6th International Conference on Bioimaging

142

REFERENCES

Carvalho D. F, Camacho J. A., Azevedo-Marques, P. M.,

Macedo, A. A. Lyria PACS A case study saves ten

million dollars in a Brazilian Hospital, 2015 IEEE

28th International Symposium on Computer-Based

Medical Systems, 326- 329, June 2015.

Carvalho, D. F, Uscamayta A. O, Vale H. M. C.,

Camacho-Guerrero J. A.,Oliveira H. F. SIPRAD: A

Radioherapy Planning System (RTPS), 2018 IEEE

31st International Symposium on Computer-Based

Medical Systems (CBMS),

10.1109/CBMS.2018.00081, June 2018.

DICOM Digital Imaging and Communications in

Medicine (DICOM) Supplement 29: Radiotherapy

Treatment Records and Radiotherapy Media

Extensions, 1999.

DICOM RT Supplement. Final Text, 4 June 1997 i.

TABLE OF CONTENTS. Part 3 Addendum

Radiotherapy Information Object Definitions.

Jaccard, P. Étude comparative de la distribution florale

dans une portion des Alpes et des Jura, Bulletin de la

Société Vaudoise des Sciences Naturelles, 1901, 37:

547–579.

Maria Y. Y. Law, Brent Liu. Informatics in Radiology

DICOM-RT and Its Utilization in Radiation Therapy.

RadioGraphics 2009, 29:655–667.

10.1148/rg.293075172.

Max J. Warnock, Christopher Toland, Damien Evans, Bill

Wallace, and Paul Nagy Benefits of Using the

DCM4CHE DICOM Archive, J Digit Imaging. 2007

Nov; 20 (Suppl 1): 125–129. 10.1007/s10278-007-

9064-1.

SBRT, Brazilian Society Radiotherapy, Report in

interview: Arthur Accioly Rosa, July 2018.

Thiruchelvam, N., Mostafid, H. and Ubhayakar, G.

(2005), Planning percutaneous nephrolithotomy using

multidetector computed tomography urography,

multiplanar reconstruction and three-dimensional

reformatting. BJU International, 95: 1280–1284.

doi:10.1111/j.1464-410X.2005.05519.x.

Radiotherapy Support Tools, the Brazilian Project: SIPRAD

143