- Email: [email protected]
Quality assurance applied to mammographic equipments using phantoms and software for its evaluation
ARTICLE IN PRESS Nuclear Instruments and Methods in Physics Research A 619 (2010) 372–374
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research A journal homepage: www.elsevier.com/locate/nima
Quality assurance applied to mammographic equipments using phantoms and software for its evaluation Patricia Mayo a,, Francisco Rodenas b, Juan Manuel Campayo c, Gumersido Verdu´ d a
´gicos S.L., Grupo Dominguis, Apartado 46015,Valencia, Spain Titania Servicios Tecnolo ´tica Aplicada, Universidad Polite ´cnica de Valencia, Apartado 46022,Valencia, Spain Departamento de Matema c ˜ez, Apartado 46017,Valencia, Spain Hospital Clı´nico Universitario de Valencia, Avda. Blasco Iban d Departamento de Ingenierı´a Quı´mica y Nuclear, Universidad Polite´cnica de Valencia, Apartado 46022, Valencia, Spain b
a r t i c l e in f o
a b s t r a c t
Available online 1 February 2010
The image quality assessment in radiographic equipments is a very important item for a complete quality control of the radiographic image chain. The periodic evaluation of the radiographic image quality must guarantee the constancy of this quality to carry out a suitable diagnosis. Mammographic phantom images are usually used to study the quality of images obtained by determined mammographic equipment. The digital image treatment techniques allow to carry out an automatic analysis of the phantom image. In this work we apply some techniques of digital image processing to analyze in an automatic way the image quality of mammographic phantoms, namely CIRS SP01 and RACON for different varying conditions of the mammographic equipment. The CIRS SP01 phantom is usually used in analogic mammographic equipments and the RACON phantom has been specifically developed by authors to be applied to acceptance and constancy tests of the image quality in digital radiographic equipments following recommendations of international associations. The purpose of this work consists in analyzing the image quality for both phantoms by means of an automatic software utility. This analysis allows us to study the functioning of the image chain of the mammographic system in an objective way, so an abnormal functioning of the radiographic equipment might be detected. & 2010 Elsevier B.V. All rights reserved.
Keywords: Quality control Radiographic phantom Automatic analysis
1. Introduction Nowadays, digital radiographic equipments are replacing traditional film-screen equipments and it is necessary to evaluate the image parameters to guarantee the quality of the new technologies [1]. The use of digital systems allows the automatic analysis of the obtained radiographic images, increasing the objectivity in the evaluation of the image. The application of digital image tools to study the image quality of digital mammographic equipments is the main purpose of this work. The objective consists in to characterize the constancy of the mammographic imaging chain and guarantee acceptable image quality with low level doses [2–4]. The development of specific phantoms to study the image obtained by digital radiographic equipments is an important task to analyze physical aspects of the imaging chain related to the evaluation of the image, and the quality control of radiographic equipments in normal operating conditions. In this sense we have considered for the image quality analysis the phantom RACON developed by the authors. This phantom has been Corresponding author. Tel.: + 34 963540300; fax: + 34 963540340.
E-mail address: [email protected] (P. Mayo). 0168-9002/$ - see front matter & 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2010.01.015
designed with different test objects recommended by international associations, as low contrast objects varying in diameter and size for the threshold contrast resolution, high resolution test for the limiting spatial resolution, dynamic step wedge for the dynamic range of the system, homogeneity zone and alignment marks for position and size of radiation field. For the study, we have also considered the commercial phantom CIRS SP01. Both phantoms are enough sensitive to the operating conditions of the radiographic system and the automatic image evaluation enables to study the global state of the image system in an objective way. The digital image processing tools for the phantom image analysis are based in standard digital image processing techniques, although they have been specifically developed for the analysis of each test object into the phantom.
2. Methodology 2.1. CIRS SP01 and RACON phantoms The mammographic phantoms that we have used are CIRS SP01 and RACON.
ARTICLE IN PRESS P. Mayo et al. / Nuclear Instruments and Methods in Physics Research A 619 (2010) 372–374
The phantom commercialized by CIRS, model SP01, contains test objects to simulate typical pathologies of the breast as microcalcifications, masses or fibers with different thickness and diameters, and has other test objects for the calibration of the image like the reference 50–50% optical density, 100% gland, 100% fat and the horizontal and vertical resolution measured as line pairs per millimetre. The phantom sketch is reflected in Fig. 1, showing the number of test objects, shape and distribution. The RACON phantom to acceptance and constancy test was made in a square block of plexiglas of size 325 325 10 millimeters with several test objects embedded into it. Fig. 2 gives an image of this phantom. The test objects contained into the RACON phantom are: 1. Low contrast objects varying in diameter and size for the threshold contrast resolution. This test zone is a block of aluminium that has specific objects of contrast-detail combinations which are cylindrical holes of determined diameter related with resolution, in a range from 0.3 to 1.6 mm, and in depth related with contrast in a range from 0.14 to 1.28 mm. 2. High resolution test for the limiting spatial resolution which varies from 0.5 to 10lp/mm that is 451 rotated to avoid the heel effect in X-ray tube emission. 3. Dynamic copper step wedge with different step thickness from 0.3 to 2.3 mm for the evaluation of the dynamic range of the image system. 4. Homogeneity zone which the grey mean level is related with the type of exposition. Alignment marks for position, and size of radiation field is evaluated within the squared marked areas.
each test zone of the phantom [5]. All the algorithms have been implemented in Matlab 7.0@. The image processing algorithms for CIRS SP01 and RACON phantoms are not identical, however they have common features. Specific details about the image processing algorithms could be seen in a previous work [6]. The process of image analysis can be summarized as follows: 1. The digital phantom image is captured by the computer program. 2. The program firstly searches the representative geometrical marks in the image after these marks have been found, the original phantom image is separated in subimages corresponding to the different test objects. 3. Specific algorithms are applied to each subimage, these are based on digital image processing techniques such as denoising filters for noise removal, pattern recognition, edge detectors, thresholding and morphological operators to detect each test.
2.2.. Image processing techniques The main purpose of our work is to carry out an automatic analysis of the image quality in radiography systems by using phantom images and digital image processing techniques. To achieve this goal, we have developed specific software tools to be applied to the analysis of CIRS SP01 and RACON phantom images. The algorithms are based on standard digital image processing techniques, although, they have been designed specifically for
373
Fig. 2. Sketch of the radiographic phantom RACON.
Fig. 1. Sketch of the radiographic phantom CIRS SP01.
ARTICLE IN PRESS 374
P. Mayo et al. / Nuclear Instruments and Methods in Physics Research A 619 (2010) 372–374
Table 1 Microcalcifications zone results for CIRS phantom. Images
Diameter (mm)
Ca
Cr (%)
SNR (dB)
23kv136mAs 26kv65mAs 28kv42mAs 30kv30mAs 33kv20mAs
0.344 0.345 0.357 0.329 0.327
17.95 23.07 26.45 26.77 27.54
10.58 16.16 20.17 23.28 32.33
28.40 24.81 23.01 21.59 18.33
Table 2 Horizontal resolution results for RACON phantom.
Fig. 3. Image processing techniques applied to SP01 CIRS phantom in microcalcifications zone.
4. After that we can obtain specific parameters for each test object as contrast with its background, geometrical features and exact location. Some of these parameters can be used to evaluate the image quality defining quality image indexes. As an example, in Fig. 3 the scheme of the image processing analysis for one of the microcalcifications zone of the CIRS SP01 phantom is shown.
3. Results and discussion Two series (one for each phantom) of test mammographic images were acquired from a digital radiographic equipment in Dicom 3.0 format for several functioning conditions (kV and mAs) of the equipment. The working conditions have an important influence in the contrast and in the results for the image quality of the digital images for both phantoms, so they were chosen as the usual clinical conditions of the equipment. Each image was analyzed by means of our software described above and several features and image parameters were computed. For the CIRS SP01 and RACON phantom images, the features evaluated have been: 1. Absolute contrast (Ca) and relative contrast (Cr %) to its background, SNR as the signal to noise ratio present in the image. 2. Horizontal and vertical resolution as lp/mm. 3. IQF as the parameter related with detected contrast-detail combinations in low contrast zone [7]. 4. Horizontal and vertical uniformity zone. In Table 1 and Table 2 some of these results for the microcalcifications zone after the automatic image processing techniques are given.
Images
Resolution threshold
Grey mean level-background mean
IQF
23kv136mAs 26kv65mAs 28kv42mAs 30kv30mAs 33kv20mAs
7 8 8 8 8
82.71 76.30 72.17 74.37 72.57
51.38 55.06 65.62 90.96 95.95
4. Conclusions The analysis carried out in this work indicates that automatic image processing techniques can be used to determine the image quality of CIRS SP01 and RACON phantoms because they are sensitive to detect variations in the operating conditions of the equipment (kv, mA and ms) due to the dependency on detection of the test objects inside both phantoms. The application is useful to obtain a relationship with acceptable image quality with low level dose. The code enables the comparison of different images between themselves and with a reference image obtained by determined functioning conditions of the equipment. The developed software is able to detect efficiently and automatically the test objects. In this sense, the application can be useful in the quality control of the equipment detecting abnormal functioning of the equipment.
Acknowledgments The authors wish to thank Clinical Hospital of Valencia their collaboration in testing these phantoms in their radiographic facilities and the project IAP-560610-2008-24 from National Programme of Applied Research. References [1] Working Party of Radiologists, Quality Assurance Coordinating Group, Computer aided detection in mammography, NHSBSP Publication no. 48, 2001. [2] J.A. Thomas, K. Chakrabarty, R. Kaczmark, A. Romanyukha, Medical Physics 33 (3) (2005) 807. [3] D.P. Chakraborty, Medical Physics 24 (1997) 1269. [4] K.W. Brooks, J.H. Trueblood, K.J. Kearfott, D.T. Lawton, Medical Physics 24 (1997) 709. [5] P. Mayo, F. Rodenas, G. Verdu´, J.I. Villaescusa, J.M. Campayo, Computer Methods and Programs in Biomedicine 73 (2) (2004) 115. [6] P. Mayo, F. Rodenas, J.M. Campayo, A. Pascual, B. Marin, G. Verdu´, Nuclear Technology 168 (2009) 235. [7] A. Pascoal, C.P. Lawinski, I. Honey, Physics in Medicine and Biology 50 (2005) 5743.
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research A journal homepage: www.elsevier.com/locate/nima
Quality assurance applied to mammographic equipments using phantoms and software for its evaluation Patricia Mayo a,, Francisco Rodenas b, Juan Manuel Campayo c, Gumersido Verdu´ d a
´gicos S.L., Grupo Dominguis, Apartado 46015,Valencia, Spain Titania Servicios Tecnolo ´tica Aplicada, Universidad Polite ´cnica de Valencia, Apartado 46022,Valencia, Spain Departamento de Matema c ˜ez, Apartado 46017,Valencia, Spain Hospital Clı´nico Universitario de Valencia, Avda. Blasco Iban d Departamento de Ingenierı´a Quı´mica y Nuclear, Universidad Polite´cnica de Valencia, Apartado 46022, Valencia, Spain b
a r t i c l e in f o
a b s t r a c t
Available online 1 February 2010
The image quality assessment in radiographic equipments is a very important item for a complete quality control of the radiographic image chain. The periodic evaluation of the radiographic image quality must guarantee the constancy of this quality to carry out a suitable diagnosis. Mammographic phantom images are usually used to study the quality of images obtained by determined mammographic equipment. The digital image treatment techniques allow to carry out an automatic analysis of the phantom image. In this work we apply some techniques of digital image processing to analyze in an automatic way the image quality of mammographic phantoms, namely CIRS SP01 and RACON for different varying conditions of the mammographic equipment. The CIRS SP01 phantom is usually used in analogic mammographic equipments and the RACON phantom has been specifically developed by authors to be applied to acceptance and constancy tests of the image quality in digital radiographic equipments following recommendations of international associations. The purpose of this work consists in analyzing the image quality for both phantoms by means of an automatic software utility. This analysis allows us to study the functioning of the image chain of the mammographic system in an objective way, so an abnormal functioning of the radiographic equipment might be detected. & 2010 Elsevier B.V. All rights reserved.
Keywords: Quality control Radiographic phantom Automatic analysis
1. Introduction Nowadays, digital radiographic equipments are replacing traditional film-screen equipments and it is necessary to evaluate the image parameters to guarantee the quality of the new technologies [1]. The use of digital systems allows the automatic analysis of the obtained radiographic images, increasing the objectivity in the evaluation of the image. The application of digital image tools to study the image quality of digital mammographic equipments is the main purpose of this work. The objective consists in to characterize the constancy of the mammographic imaging chain and guarantee acceptable image quality with low level doses [2–4]. The development of specific phantoms to study the image obtained by digital radiographic equipments is an important task to analyze physical aspects of the imaging chain related to the evaluation of the image, and the quality control of radiographic equipments in normal operating conditions. In this sense we have considered for the image quality analysis the phantom RACON developed by the authors. This phantom has been Corresponding author. Tel.: + 34 963540300; fax: + 34 963540340.
E-mail address: [email protected] (P. Mayo). 0168-9002/$ - see front matter & 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2010.01.015
designed with different test objects recommended by international associations, as low contrast objects varying in diameter and size for the threshold contrast resolution, high resolution test for the limiting spatial resolution, dynamic step wedge for the dynamic range of the system, homogeneity zone and alignment marks for position and size of radiation field. For the study, we have also considered the commercial phantom CIRS SP01. Both phantoms are enough sensitive to the operating conditions of the radiographic system and the automatic image evaluation enables to study the global state of the image system in an objective way. The digital image processing tools for the phantom image analysis are based in standard digital image processing techniques, although they have been specifically developed for the analysis of each test object into the phantom.
2. Methodology 2.1. CIRS SP01 and RACON phantoms The mammographic phantoms that we have used are CIRS SP01 and RACON.
ARTICLE IN PRESS P. Mayo et al. / Nuclear Instruments and Methods in Physics Research A 619 (2010) 372–374
The phantom commercialized by CIRS, model SP01, contains test objects to simulate typical pathologies of the breast as microcalcifications, masses or fibers with different thickness and diameters, and has other test objects for the calibration of the image like the reference 50–50% optical density, 100% gland, 100% fat and the horizontal and vertical resolution measured as line pairs per millimetre. The phantom sketch is reflected in Fig. 1, showing the number of test objects, shape and distribution. The RACON phantom to acceptance and constancy test was made in a square block of plexiglas of size 325 325 10 millimeters with several test objects embedded into it. Fig. 2 gives an image of this phantom. The test objects contained into the RACON phantom are: 1. Low contrast objects varying in diameter and size for the threshold contrast resolution. This test zone is a block of aluminium that has specific objects of contrast-detail combinations which are cylindrical holes of determined diameter related with resolution, in a range from 0.3 to 1.6 mm, and in depth related with contrast in a range from 0.14 to 1.28 mm. 2. High resolution test for the limiting spatial resolution which varies from 0.5 to 10lp/mm that is 451 rotated to avoid the heel effect in X-ray tube emission. 3. Dynamic copper step wedge with different step thickness from 0.3 to 2.3 mm for the evaluation of the dynamic range of the image system. 4. Homogeneity zone which the grey mean level is related with the type of exposition. Alignment marks for position, and size of radiation field is evaluated within the squared marked areas.
each test zone of the phantom [5]. All the algorithms have been implemented in Matlab 7.0@. The image processing algorithms for CIRS SP01 and RACON phantoms are not identical, however they have common features. Specific details about the image processing algorithms could be seen in a previous work [6]. The process of image analysis can be summarized as follows: 1. The digital phantom image is captured by the computer program. 2. The program firstly searches the representative geometrical marks in the image after these marks have been found, the original phantom image is separated in subimages corresponding to the different test objects. 3. Specific algorithms are applied to each subimage, these are based on digital image processing techniques such as denoising filters for noise removal, pattern recognition, edge detectors, thresholding and morphological operators to detect each test.
2.2.. Image processing techniques The main purpose of our work is to carry out an automatic analysis of the image quality in radiography systems by using phantom images and digital image processing techniques. To achieve this goal, we have developed specific software tools to be applied to the analysis of CIRS SP01 and RACON phantom images. The algorithms are based on standard digital image processing techniques, although, they have been designed specifically for
373
Fig. 2. Sketch of the radiographic phantom RACON.
Fig. 1. Sketch of the radiographic phantom CIRS SP01.
ARTICLE IN PRESS 374
P. Mayo et al. / Nuclear Instruments and Methods in Physics Research A 619 (2010) 372–374
Table 1 Microcalcifications zone results for CIRS phantom. Images
Diameter (mm)
Ca
Cr (%)
SNR (dB)
23kv136mAs 26kv65mAs 28kv42mAs 30kv30mAs 33kv20mAs
0.344 0.345 0.357 0.329 0.327
17.95 23.07 26.45 26.77 27.54
10.58 16.16 20.17 23.28 32.33
28.40 24.81 23.01 21.59 18.33
Table 2 Horizontal resolution results for RACON phantom.
Fig. 3. Image processing techniques applied to SP01 CIRS phantom in microcalcifications zone.
4. After that we can obtain specific parameters for each test object as contrast with its background, geometrical features and exact location. Some of these parameters can be used to evaluate the image quality defining quality image indexes. As an example, in Fig. 3 the scheme of the image processing analysis for one of the microcalcifications zone of the CIRS SP01 phantom is shown.
3. Results and discussion Two series (one for each phantom) of test mammographic images were acquired from a digital radiographic equipment in Dicom 3.0 format for several functioning conditions (kV and mAs) of the equipment. The working conditions have an important influence in the contrast and in the results for the image quality of the digital images for both phantoms, so they were chosen as the usual clinical conditions of the equipment. Each image was analyzed by means of our software described above and several features and image parameters were computed. For the CIRS SP01 and RACON phantom images, the features evaluated have been: 1. Absolute contrast (Ca) and relative contrast (Cr %) to its background, SNR as the signal to noise ratio present in the image. 2. Horizontal and vertical resolution as lp/mm. 3. IQF as the parameter related with detected contrast-detail combinations in low contrast zone [7]. 4. Horizontal and vertical uniformity zone. In Table 1 and Table 2 some of these results for the microcalcifications zone after the automatic image processing techniques are given.
Images
Resolution threshold
Grey mean level-background mean
IQF
23kv136mAs 26kv65mAs 28kv42mAs 30kv30mAs 33kv20mAs
7 8 8 8 8
82.71 76.30 72.17 74.37 72.57
51.38 55.06 65.62 90.96 95.95
4. Conclusions The analysis carried out in this work indicates that automatic image processing techniques can be used to determine the image quality of CIRS SP01 and RACON phantoms because they are sensitive to detect variations in the operating conditions of the equipment (kv, mA and ms) due to the dependency on detection of the test objects inside both phantoms. The application is useful to obtain a relationship with acceptable image quality with low level dose. The code enables the comparison of different images between themselves and with a reference image obtained by determined functioning conditions of the equipment. The developed software is able to detect efficiently and automatically the test objects. In this sense, the application can be useful in the quality control of the equipment detecting abnormal functioning of the equipment.
Acknowledgments The authors wish to thank Clinical Hospital of Valencia their collaboration in testing these phantoms in their radiographic facilities and the project IAP-560610-2008-24 from National Programme of Applied Research. References [1] Working Party of Radiologists, Quality Assurance Coordinating Group, Computer aided detection in mammography, NHSBSP Publication no. 48, 2001. [2] J.A. Thomas, K. Chakrabarty, R. Kaczmark, A. Romanyukha, Medical Physics 33 (3) (2005) 807. [3] D.P. Chakraborty, Medical Physics 24 (1997) 1269. [4] K.W. Brooks, J.H. Trueblood, K.J. Kearfott, D.T. Lawton, Medical Physics 24 (1997) 709. [5] P. Mayo, F. Rodenas, G. Verdu´, J.I. Villaescusa, J.M. Campayo, Computer Methods and Programs in Biomedicine 73 (2) (2004) 115. [6] P. Mayo, F. Rodenas, J.M. Campayo, A. Pascual, B. Marin, G. Verdu´, Nuclear Technology 168 (2009) 235. [7] A. Pascoal, C.P. Lawinski, I. Honey, Physics in Medicine and Biology 50 (2005) 5743.