- Email: [email protected]
Performance evaluation of different digital breast tomosynthesis systems: Dose and image quality assessment
e82
Abstracts/Physica Medica 32 (2016) e71–e96
Radiological Sciences and Public Health, University of Brescia, Brescia, Italy; c School of Medical Physics, University of Milan, Milano, Italy; d Medical Technology Laboratory, Rizzoli Orthopaedic Institute, Bologna, Italy; e Neuroimaging Lab, Medical and Surgical Specialties, Radiological Sciences and Public Health, University of Brescia, Brescia, Italy Introduction: Bone volume fraction (BV/TV), trabecular thickness (Tb.Th) and trabecular separation (Tb.Sp) are the structural parameters used to represent trabecular bone architecture. In general, micro computed tomography (micro-CT) is considered the standard for evaluating human trabeculae but it is limited to specimen due to FOV restrictions. In the present study the spatial resolution of a CBCT scanner was characterized for in-vivo bone structural measurements and the derived trabecular parameters were compared to micro-CT results. Materials and Methods: For QR Newtom 5G CBCT installed at Spedali Civili (Brescia, Italy) Noise Power Spectrum (NPS) and Modulation Transfer Function (MTF) were measured using Catphan600 phantom for different voxel sizes (75, 100, 125, 150 μm). Four human specimens with different trabeculae concentrations were scanned by both CBCT and a Skyscan microCT model 1072 (Bruker MicroCT, Konthich, Belgium) installed at Rizzoli Orthopedic Institute (Bologna, Italy). For CBCT Gray Level Distributions (GLDs) of trabecular volumes were compared among the different voxel sizes. Percentage differences and intercalibration curves were calculated between CBCT (all voxel sizes) and micro-CT trabecular parameters. Results: For CBCT, MTF curve moved toward right while NPS integral increased when decreasing voxel size. GLDs representing bone marrow and cancellous bone were resolved for all voxel sizes except for 75 μm. For all voxel sizes CBCT systematically overestimated BV/TV and Tb.Th. Intercalibration curves were monotonic but not linear. Conclusion: MTF improvement does not justify the use of the 75 μm voxel size because of noise increase. GLDs comparison points out that the 150 μm voxel size has the highest discrimination capability between bone marrow and cancellous bone. Intercalibration curves against micro-CT are not linear due to a CBCT saturation effect for higher trabecular concentrations, but can be used to correct CBCT trabecular parameters. http://dx.doi.org/10.1016/j.ejmp.2016.01.280
B.276 PERFORMANCE EVALUATION OF DIFFERENT DIGITAL BREAST TOMOSYNTHESIS SYSTEMS: DOSE AND IMAGE QUALITY ASSESSMENT A. Maldera *,a, P. De Marco b, P.E. Colombo a, D.A. Origgi b, A. Torresin a. a Ospedale Niguarda Ca’ Granda, Milano, Italy; b Istituto Europeo di Oncologia, Milano, Italy Introduction: Digital Breast Tomosynthesis (DBT) is a recently introduced technique for detection of breast cancer. The aim of this work is to give a physical characterization of three commercial systems in order to evaluate their performances in clinical practice. Methods and Materials: The systems investigated are Senograph Essential® (GE), Mammomat Inspiration® (Siemens) and Selenia Dimensions® (Hologic). They use different angular ranges (15°–50°), number of projections (9–25), type of acquisition (step and shoot – continuous) and reconstruction algorithms (FBP – iterative). Average Glandular Dose (AGD) and image quality parameters were assessed. AGD was estimated with a calibrated ionization chamber according to the model proposed by Dance. Quality parameters of 3D images include spatial resolution evaluated with the ‘‘in plane’’ modulation transfer function (MTF) (Zhao 2008), ‘‘in depth’’ resolution (Marshall 2012) and signal difference to noise ratio (SDNR). The artifact was investigated measuring the Artefact Spread Function (ASF, Wu 2004) of spherical details of various materials and diameters. Results: Measured AGD was always below acceptable limits of full field digital mammography (EUREF 2006). We found the best “in plane” resolution on GE system; this could be related to the used step&shoot acquisition modality. The “in depth” resolution, measured with the FWHM of a PSF in z-direction, improved with increasing scan angle. The Hologic system showed the best SDNR, probably due to its pixel binning, while for the Siemens system low contrast detectability was poor.
As expected, ASF depends not only on detail dimension but is especially sensitive to scan angle range: the persistency of the artifacts away from the in-focus plane is larger for smaller range systems. Conclusions: Physical characterization of DBT systems is important to evaluate dose and to determine image quality parameters, which can influence clinical detectability of pathological tissues. http://dx.doi.org/10.1016/j.ejmp.2016.01.281
B.277 ESTIMATE ORGAN DOSE FROM CT EXAMINATION: A SOFTWARE COMPARISON A. Maldera *,a, M. Sutto a, A. Gaeta a, M. Maddalo b, P.E. Colombo a, A. Torresin a. a Ospedale Niguarda Ca’ Granda, Milano, Italy; b Spedali Civili Brescia, Brescia, Italy Introduction: Risk estimate in CT examination is directly connected to organ dose calculation. The aim of this work is to evaluate the effect of simulation parameters on organ dose estimation in CT examination with different commercial software. In this study CT-Expo® (Sascrad, EX), ImpaCT Dose® (CT Imaging GmbH, ID) and VirtualDose® (Virtual Phantoms Inc., VD) were compared. Materials and Methods: The most evident difference among SW is the adopted phantom: EX and ID employ stylized phantoms, whereas VD includes a family of hybrid phantoms. Differences between adult phantoms, adapted to ICRP “standard male and female” have been evaluated with CT scan simulations that encompass the whole body. VD and ID allow to simulate on phantoms with different body mass indices: mean scan parameters of obese patients have been simulated both on “normal weight” phantom and on “obese phantoms” to analyze organ dose differences. EX and ID can simulate tube current modulation (TCM): the effect on organ dose has been evaluated comparing a chest-abdomen-pelvis scan simulated with constant current and one with TCM. Results: Dose variability of 30% has been found for totally included organs of “standard male and female” phantoms due to different dimension and position of organs. Variability increases for partially included organs up to 70%. Scan parameters of an obese patient simulated on an “obese phantom” results in, at least 10% less organ dose than if simulated on a “normal weight phantom”, with percentage increasing with obesity degree. Finally, organ doses estimated with and without TCM can vary up to 40%. Conclusion: A precise patient dose estimate is difficult to obtain, also because phantoms are eventually just a representation of the patient’s anatomy and the actual scanner can differ from the one used by software. Nevertheless, a comparison between different SW can help understand the existing error bars across dose estimates for further improvement. http://dx.doi.org/10.1016/j.ejmp.2016.01.282
B.278 HEALTH DOSE ASSESSMENT OR HEALTH TECHNOLOGY ASSESSMENT? L. Manco *, A. Turra. Medical Physics Unit, Arcispedale S.Anna Hospital, Ferrara, Italy Introduction: The diffusion of radiation dose management software (DMS) nationwide should allow dose-population’s monitoring? The use of DoseWatch (GE) software during the two past years allows a continuous monitoring of CT dose at Sant’Anna Hospital in Ferrara. The purpose of this work is to evaluate the effective reduction of CT dose after installation of iterative reconstruction method (ASiR, GE). Materials and Methods: More than 30,000 exams were collected on DoseWatch from two CT scanners GE Lightspeed VTC 32 slice. One is used for body exams and the other for neuro ones. The analysis was carried out comparing DLP values before and after installation of ASiR in different study protocols. Results: The use of iterative reconstruction method on CT scanners allows a reduction of average dose of 25% for protocol thorax-abdomen and 30% for protocol head. The compare between before and after ASiR, on all exams, shows a dose reduction of 52% on CT Boby and 30% on CT Neuro.
Abstracts/Physica Medica 32 (2016) e71–e96
Radiological Sciences and Public Health, University of Brescia, Brescia, Italy; c School of Medical Physics, University of Milan, Milano, Italy; d Medical Technology Laboratory, Rizzoli Orthopaedic Institute, Bologna, Italy; e Neuroimaging Lab, Medical and Surgical Specialties, Radiological Sciences and Public Health, University of Brescia, Brescia, Italy Introduction: Bone volume fraction (BV/TV), trabecular thickness (Tb.Th) and trabecular separation (Tb.Sp) are the structural parameters used to represent trabecular bone architecture. In general, micro computed tomography (micro-CT) is considered the standard for evaluating human trabeculae but it is limited to specimen due to FOV restrictions. In the present study the spatial resolution of a CBCT scanner was characterized for in-vivo bone structural measurements and the derived trabecular parameters were compared to micro-CT results. Materials and Methods: For QR Newtom 5G CBCT installed at Spedali Civili (Brescia, Italy) Noise Power Spectrum (NPS) and Modulation Transfer Function (MTF) were measured using Catphan600 phantom for different voxel sizes (75, 100, 125, 150 μm). Four human specimens with different trabeculae concentrations were scanned by both CBCT and a Skyscan microCT model 1072 (Bruker MicroCT, Konthich, Belgium) installed at Rizzoli Orthopedic Institute (Bologna, Italy). For CBCT Gray Level Distributions (GLDs) of trabecular volumes were compared among the different voxel sizes. Percentage differences and intercalibration curves were calculated between CBCT (all voxel sizes) and micro-CT trabecular parameters. Results: For CBCT, MTF curve moved toward right while NPS integral increased when decreasing voxel size. GLDs representing bone marrow and cancellous bone were resolved for all voxel sizes except for 75 μm. For all voxel sizes CBCT systematically overestimated BV/TV and Tb.Th. Intercalibration curves were monotonic but not linear. Conclusion: MTF improvement does not justify the use of the 75 μm voxel size because of noise increase. GLDs comparison points out that the 150 μm voxel size has the highest discrimination capability between bone marrow and cancellous bone. Intercalibration curves against micro-CT are not linear due to a CBCT saturation effect for higher trabecular concentrations, but can be used to correct CBCT trabecular parameters. http://dx.doi.org/10.1016/j.ejmp.2016.01.280
B.276 PERFORMANCE EVALUATION OF DIFFERENT DIGITAL BREAST TOMOSYNTHESIS SYSTEMS: DOSE AND IMAGE QUALITY ASSESSMENT A. Maldera *,a, P. De Marco b, P.E. Colombo a, D.A. Origgi b, A. Torresin a. a Ospedale Niguarda Ca’ Granda, Milano, Italy; b Istituto Europeo di Oncologia, Milano, Italy Introduction: Digital Breast Tomosynthesis (DBT) is a recently introduced technique for detection of breast cancer. The aim of this work is to give a physical characterization of three commercial systems in order to evaluate their performances in clinical practice. Methods and Materials: The systems investigated are Senograph Essential® (GE), Mammomat Inspiration® (Siemens) and Selenia Dimensions® (Hologic). They use different angular ranges (15°–50°), number of projections (9–25), type of acquisition (step and shoot – continuous) and reconstruction algorithms (FBP – iterative). Average Glandular Dose (AGD) and image quality parameters were assessed. AGD was estimated with a calibrated ionization chamber according to the model proposed by Dance. Quality parameters of 3D images include spatial resolution evaluated with the ‘‘in plane’’ modulation transfer function (MTF) (Zhao 2008), ‘‘in depth’’ resolution (Marshall 2012) and signal difference to noise ratio (SDNR). The artifact was investigated measuring the Artefact Spread Function (ASF, Wu 2004) of spherical details of various materials and diameters. Results: Measured AGD was always below acceptable limits of full field digital mammography (EUREF 2006). We found the best “in plane” resolution on GE system; this could be related to the used step&shoot acquisition modality. The “in depth” resolution, measured with the FWHM of a PSF in z-direction, improved with increasing scan angle. The Hologic system showed the best SDNR, probably due to its pixel binning, while for the Siemens system low contrast detectability was poor.
As expected, ASF depends not only on detail dimension but is especially sensitive to scan angle range: the persistency of the artifacts away from the in-focus plane is larger for smaller range systems. Conclusions: Physical characterization of DBT systems is important to evaluate dose and to determine image quality parameters, which can influence clinical detectability of pathological tissues. http://dx.doi.org/10.1016/j.ejmp.2016.01.281
B.277 ESTIMATE ORGAN DOSE FROM CT EXAMINATION: A SOFTWARE COMPARISON A. Maldera *,a, M. Sutto a, A. Gaeta a, M. Maddalo b, P.E. Colombo a, A. Torresin a. a Ospedale Niguarda Ca’ Granda, Milano, Italy; b Spedali Civili Brescia, Brescia, Italy Introduction: Risk estimate in CT examination is directly connected to organ dose calculation. The aim of this work is to evaluate the effect of simulation parameters on organ dose estimation in CT examination with different commercial software. In this study CT-Expo® (Sascrad, EX), ImpaCT Dose® (CT Imaging GmbH, ID) and VirtualDose® (Virtual Phantoms Inc., VD) were compared. Materials and Methods: The most evident difference among SW is the adopted phantom: EX and ID employ stylized phantoms, whereas VD includes a family of hybrid phantoms. Differences between adult phantoms, adapted to ICRP “standard male and female” have been evaluated with CT scan simulations that encompass the whole body. VD and ID allow to simulate on phantoms with different body mass indices: mean scan parameters of obese patients have been simulated both on “normal weight” phantom and on “obese phantoms” to analyze organ dose differences. EX and ID can simulate tube current modulation (TCM): the effect on organ dose has been evaluated comparing a chest-abdomen-pelvis scan simulated with constant current and one with TCM. Results: Dose variability of 30% has been found for totally included organs of “standard male and female” phantoms due to different dimension and position of organs. Variability increases for partially included organs up to 70%. Scan parameters of an obese patient simulated on an “obese phantom” results in, at least 10% less organ dose than if simulated on a “normal weight phantom”, with percentage increasing with obesity degree. Finally, organ doses estimated with and without TCM can vary up to 40%. Conclusion: A precise patient dose estimate is difficult to obtain, also because phantoms are eventually just a representation of the patient’s anatomy and the actual scanner can differ from the one used by software. Nevertheless, a comparison between different SW can help understand the existing error bars across dose estimates for further improvement. http://dx.doi.org/10.1016/j.ejmp.2016.01.282
B.278 HEALTH DOSE ASSESSMENT OR HEALTH TECHNOLOGY ASSESSMENT? L. Manco *, A. Turra. Medical Physics Unit, Arcispedale S.Anna Hospital, Ferrara, Italy Introduction: The diffusion of radiation dose management software (DMS) nationwide should allow dose-population’s monitoring? The use of DoseWatch (GE) software during the two past years allows a continuous monitoring of CT dose at Sant’Anna Hospital in Ferrara. The purpose of this work is to evaluate the effective reduction of CT dose after installation of iterative reconstruction method (ASiR, GE). Materials and Methods: More than 30,000 exams were collected on DoseWatch from two CT scanners GE Lightspeed VTC 32 slice. One is used for body exams and the other for neuro ones. The analysis was carried out comparing DLP values before and after installation of ASiR in different study protocols. Results: The use of iterative reconstruction method on CT scanners allows a reduction of average dose of 25% for protocol thorax-abdomen and 30% for protocol head. The compare between before and after ASiR, on all exams, shows a dose reduction of 52% on CT Boby and 30% on CT Neuro.