Storage optimization of multi-slice computed tomographic data on PACS systems

Transcription

1 Storage optimization of multi-slice computed tomographic data on PACS systems Poster No.: C-3033 Congress: ECR 2010 Type: Scientific Exhibit Topic: Radiographers Authors: B. Begenau, C. Beigelman, O. Lucidarme, D. Toledano, P. A. Grenier; Paris/FR Keywords: PACS, picture archiving and communication system, Volumetric data saving and optimization of storage, Image quality of MDCT scan DOI: /ecr2010/C-3033 Any information contained in this pdf file is automatically generated from digital material submitted to EPOS by third parties in the form of scientific presentations. References to any names, marks, products, or services of third parties or hypertext links to thirdparty sites or information are provided solely as a convenience to you and do not in any way constitute or imply ECR's endorsement, sponsorship or recommendation of the third party, information, product or service. ECR is not responsible for the content of these pages and does not make any representations regarding the content or accuracy of material in this file. As per copyright regulations, any unauthorised use of the material or parts thereof as well as commercial reproduction or multiple distribution by any traditional or electronically based reproduction/publication method ist strictly prohibited. You agree to defend, indemnify, and hold ECR harmless from and against any and all claims, damages, costs, and expenses, including attorneys' fees, arising from or related to your use of these pages. Please note: Links to movies, ppt slideshows and any other multimedia files are not available in the pdf version of presentations. Page 1 of 63

2 Purpose 1. New generations of multidetector computed tomography scans (MDCT) generate high data flows. To ensure an efficient online capacity and a storage optimization of Picture Archiving and Communication System (PACS), often limited by saturation or extension restrictions, it seems necessary to reduce data volume. 2. The purposes of this study are to demonstrate that: The quality of images with higher slice thickness compared to the native CT images may be maintained or judged as sufficient and that their future use for follow-up and post-processing analyses will not be impeded. The total data volume stored in PACS could decrease by 55% with data volume savings per exam type by 22% to 75% according to the respective number of reconstructions and the corresponding reconstruction parameters. Consequently, online capacity could be more efficient and optimized if images with higher slice thickness are used for storage. Context Because online storage capacity on PACS reaches nearly the saturation point within 6 months by maintaining a fixed total online storage capacity of 36 TB in our hospital, a global saving of 30% for all medical imaging departments has been required. In fact, the direct consequence of saturation on PACS is a decrease of online duration time of images and a possible less convenient follow up comparison of exams if these are not available online anymore. To find a solution for these problems a study was proceeded on a beta site MDCT scan (256 slices) which produces high monthly data flow. Some elements concerning our hospital: Page 2 of 63

3 beds. - Two departments of radiology including neurology. Equipments connected to PACS: 3 MDCT scans 4 MRI 4 Angiography devices 11 Standard RX devices 5 Sonography devices Some elements concerning PACS: Total online storage capacity: 36 TB Total data volume sent to PACS per month (all imaging departments): 1.2 TB Current online storage duration: 2.5 years Actual need of online storage duration: 5 years Storage legally required online or offline: 20 years PACS compression coefficient: X2 Objectives Reduce the total image volume from 1.2 TB to 800 GB per month by maintaining a sufficient image quality to allow meaningful subsequent comparison with follow-up images and increase the online storage duration from 2.5 to 5 years. Page 3 of 63

4 Fig.: Objectives References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Page 4 of 63

5 Images for this section: Fig. 1: Objectives Page 5 of 63

6 Methods and Materials Identification of means to reduce data volume Means to reduce data volume in PACS had to be identified to reach data savings for all medical imaging departments in our hospital. Proposal Explored : Direct Effect Consequence Reduction of total number of images stored Gain of space on PACS Storage of images with higher slice thickness Fig.: Storage of images with higher slice thickness References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Proposal Optional : Direct Effect Consequence Eliminating useless data transfer Gain of space on PACS No automatic data transfer Page 6 of 63

7 Fig.: Abolishment of automatic data transfer of native images, implementation of manual transfer References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Proposal Direct Effect Gain of online space, privileging exams requiring Identification of frequent follow up unnecessary images online for immediate deletion Optional : Consequence Gain of space on PACS Fig.: Identification of unnecessary images online for immediate deletion Page 7 of 63

8 References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Proposal Direct Effect Gain of online space, privileging exams requiring Distinction between exams frequent follow up with regards to time interval for follow up studie Optional : Consequence Gain of space on PACS Fig.: Distinction between examinations concerning their online duration References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Proposal Optional : Transfer examination data of deceased patients from the online to the offline system Direct Effect Consequence Gain of online space, privileging exams requiring follow up Gain of space on PACS Page 8 of 63

9 Fig.: Transfer examinations of deceased patients from online to offline system References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Categories and types of exams included in study For proposal "Storage of images with higher slice thickness" MDCT, 256 slices images from following table of Categories and types of exams were studied. Page 9 of 63

10 Fig.: Table of categories and types for CT scan explored for study References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Qualitative scale used for assessment For each type of CT scan, quality of axial sections, coronal and sagittal reformats, including native and slabs with minimum intensity projection (mip), maximum intensity projection (MIP) and average modes and 3D-VR images of two to three different sets of reconstruction parameters were assessed. A qualitative three point's scale - good, acceptable, not acceptable - was used: First set: native reconstruction obtained in clinical routine Second and the third sets: obtained by increasing the slice thickness from 11% to 50% and 51% to 100%, and changing the increment by 50% and 70% respectively Page 10 of 63

11 For each examination type, the set chosen for storage in the PACS was the one that reached the best compromise between a good image quality and a lower data volume. Representative tables of reconstruction values The tables of reconstruction values in the sidebar show the average number of reconstructions per exam, the native reconstruction parameters and the approved higher slice thickness and increment according to each scan type which was analysed for this study. Representative examples: Table of reconstruction values cardiac CT scan Figure 1 on page 28 Table of reconstruction values thoracic aorta CT scan Figure 2 on page 28 Table of reconstruction values total aorta CT scan Figure 3 on page 29 Table of reconstruction values abdominal aorta and lower extremities CT scan Figure 4 on page 29 Table of reconstruction values chest CT scan Figure 5 on page 30 Table of reconstruction values chest-abdomino-pelvic CT Figure 6 on page 30 Table of reconstruction values hepatic CT scan multiphases Figure 7 on page 31 Table of reconstruction values uro CT scan Figure 8 on page 31 Table of reconstruction virtual colonoscopy CT scan Figure 9 on page 32 Note that all image reconstructions are performed with matrix 512 reconstructions. 2 except lung 2 Lung reconstructions are performed with matrix 768 in order to maintain the quality of the high resolution kernel that gives the highest spatial resolution and therefore the visualization of tiny details. Page 11 of 63

12 Representative images cases The following images cases for the studied scan types show only the most common image reformat used for analysis and the reformat where difference of image quality is most likely to be seen. Images cases cardiac CT scan View of the left main coronary artery, a segment of the LAD and the ostium of LCX in MIP 1.5 mm thickness. Despite a slight augmentation of blurring in the thicker reconstruction, analysis is not compromised. Appreciation of image 1 mm was assessed as good. Fig.: Coronary CT scan, Reformat: coronal oblique, Effective slice thickness: 0.8 and 1mm, Maxi intensity projection: 1.5mm References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Page 12 of 63

13 Right coronary artery reconstructed in curvilinear mode. Appreciation of image 1 mm was assessed as good. Fig.: Coronary CT scan, Curvilignar Reformat, Effective slice thickness: 0.8 and 1 mm References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Same patient's RCA reconstructed in elongated mode. No difference in terms of image quality visible for human eye. Appreciation of image 1 mm was assessed as good. Page 13 of 63

14 Fig.: Coronary CT scan, Reformat: elongated mode, Effective slice thickness: 0.8 and 1 mm References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Images cases thoracic aorta CT scan Sagittal oblique reconstruction of the thoracic aorta: the analysis of the dissection is not compromised whatever the slice thickness is. Appreciation of image 1.4 mm was assessed as good. Page 14 of 63

15 Fig.: Thoracic aorta CT scan, Reformat: sagittal oblique, Effective slice thickness: 0.9, 1.25 and 1.4 mm References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Images cases total aorta CT scan Coronal slab in Maximum Intensity Projection of the whole volume. A good assessment of the aneurysm and calcium deposits is obtained with this post-processing. Due to the better signal to noise ratio with 1.4 mm slice thickness; the identification of the vessels is slightly enhanced in 1.4mm using automatic mode. Appreciation of image 1.4 mm was assessed as good. Page 15 of 63

16 Fig.: Total aorta CT scan, Reformat: coronal volume rendering, Effective slice thickness: 0.9 and 1.4 mm References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Images cases abdominal aorta and lower extremities CT scan Analysis of aorto iliac prosthesis is not compromised with various slice thicknesses. A difference in terms of image quality is hard to recognise. Appreciation of image 1.4 mm was assessed as good. Page 16 of 63

17 Fig.: Abdominal Aorta CT scan, Reformat: coronal view, Effective slice thickness: 1, 1.25 and 1.4 mm, Maxi intensity projection: 11 mm References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Fibular artery and posterior tibial artery : No deterioration visible to human eye.the appreciation of image 1.4 mm was assessed as good. Fig.: Arteries of lower extremities CT scan, Reformat: coronal Effective slice thickness: 1, 1.25 and 1.4 mm Maxi intensity projection: 12 mm Page 17 of 63

18 References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Images cases chest CT scan Multinodular heterogeneous goiter. The noisy aspect with slice thickness of 1.5 mm is associated with better visualisation of details. A better signal to noise ration was assessed with a slice thickness of 3 mm, even though the voxels are not isotropic anymore. No loss in terms of diagnosis could be ascertained on this axial view. The image deterioration was acceptable, and appreciation of image 3 mm was assessed as good. Fig.: Chest CT scan, Reformat: axial view, Effective slice thickness: 1.5 and 3 mm References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Sagittal reconstruction of the same patient. Stairsteps shaped artefacts related to reconstruction interval are observed on the 3 mm image. Note that noise visualized as dots on the left image appears more elongated in the right image due to anisotropic voxels. Image quality is acceptable in terms of further use as comparaison. The quality of image 3 mm was considered as acceptable. Page 18 of 63

19 Fig.: Chest CT scan, Reformat: sagittal, Effective slice thickness: 1.5 and 3 mm References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Curved bronchus related to rounded atelectasis, located in front of a calcified pleural thickening. Even though bronchial walls are best delineated with the 0.8mm thick slice, the 1.4 mm thick slice image remains perfectly interpretable. The image 1.4 mm was assessed as good. Fig.: Chest CT scan parenchyma reconstruction, Reformat: coronal oblique, Effective slice thickness: 0.8, 1 and 1.4 mm Page 19 of 63

20 References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Images cases chest-abdomino-pelvic CT scan Coronal 6mm thick slab with MIP mode obtained from reconstruction of 1.5 and 3mm effective slice thicknesses. Inferior vena cava and mesenteric vessels are visible in both series. Note that image details are better seen with 1.5mm effective slice thickness. Image of 3mm thick was considered as acceptable. Fig.: Chest-Abdomino-pelvic CT scan, Reformat: coronal, Effective slice thickness: 1.5 and 3 mm, Maxi intensity projection: 6 mm References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Images cases hepatic CT scan multiphases Page 20 of 63

21 Axial view of hepatic lesions whose evaluation is not altered despite a slight augmentation of blurring in 3 mm image slice thickness. Image 3mm was assessed as acceptable. Fig.: Hepatic CT scan multiphases, Reformat:axial view, Effective slice thickness: 2 and 3mm References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Images cases uro CT scan Good spatial resolution with1.5 mm thick image is required for evaluation of details like subtle stenosis. The 3 mm thick image is less noisy but less precise. Nevertheless in terms of future use for comparaison the image deterioration is acceptable. Note that even the axial Maxi Intensity Projection of 4mm does not compensate the different images qualities. Image of 2.5mm slice thickness was assessed as acceptable. Page 21 of 63

22 Fig.: CT uronography, Reformat: axial view, Effective slice thickness: 1.5 and 2.5 mm Maxi Intensity Projection: 4 mm References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Coronal 11 mm reformat with Maximum Intensity Projection allowing perfect visualization of superior and inferior renal arteries. Better spatial resolution in 1.5 mm image, sharp details. Though image quality of 2.5mm is slightly deteriorated by appearance of stairsteps shaped artefacts, image quality is acceptable, diagnosis will not be compromised. Image of 2.5mm was assessed as acceptable. Page 22 of 63

23 Fig.: CT uronography, Reformat: coronal, Effective slice thickness: 1.5 and 2.5 mm Maxi Intensity Projection: 11 mm References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Images cases virtual colonscopy CT scan Two differents polyps. Deformation of round into oval structures with slice thickness 2 mm and 2.5mm due to anisotropic voxels can be observed. The 2mm image was assessed acceptable. Page 23 of 63

24 Fig.: Virtual Colonoscopy CT scan, Reformat: virtuel endoscopic images unfold mode, Effective slice thickness: 1.4, 2 and 2.5 mm References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Page 24 of 63

25 Identification of reference values to gauge data volume To estimate the average number of images per examination type and that way obtain realistic data volume average acquisition lengths had to be determined. Assumed hypothesis According to the official record 2005 with regards to the average body size of the French population: Reference value male: cm Reference value female: cm Method 266 examinations were evaluated to calculate average acquisition length according to all different scan types. Manual measurements with the help of anatomic leads were executed to estimate theoretic acquisition length on: Female co-worker's size: 163 cm, Male co-worker's size: 175 cm. Objective The first objective was to identify average male and female reference values in terms of acquisition length for each scan type in order to identify corresponding number of images and resulting average data volumes. Page 25 of 63

26 The second objective was to validate the hypothesis that the average patient size correlates with the average size of the French population. Results 68% of average acquisition length is very close to the comparative manual measurements performed on male and female co-worker, with increasing number of patients the obtained values get closer to the comparative manual measurements. Conclusion Hypothesis validated: average acquisition length is correlated to average body size of the French population (deviation less than 4%) Table of reference values of average acquisition lenght Green values point out the three differents scan types realized most frequently. Page 26 of 63

27 Fig.: Table of average scan lenght values References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Page 27 of 63

28 Images for this section: Fig. 1: Table of reconstruction values cardia CT scan. If retrospective multi cycle reconstructions are performed, only the phases with good visibility of coronary arteries should be stored in PACS besides ventricular analysis. In prospective mode, only the reconstruction with edge correction should be stored; the non edge correction reconstruction should be used as control for analysis only. Page 28 of 63

29 Fig. 2: Table of reconstruction values thoracic aorta CT scan. In prospective mode only the reconstruction with edge correction should be stored; the non edge correction reconstruction should be used as control for analysis only. Fig. 3: Table of reconstruction values total aorta CT scan Page 29 of 63

30 Fig. 4: Table of reconstruction values abdominal aorta and lower extremities CT scan Fig. 5: Table of reconstruction values chest CT scan Page 30 of 63

31 Fig. 6: Table of reconstruction values chest-abdomino-pelvic CT scan Fig. 7: Table of reconstruction values hepatic CT scan multiphases Page 31 of 63

32 Fig. 8: Table of reconstruction values uro CT scan Fig. 9: Table of reconstruction virtual colonoscopy CT scan. Only the acquisitions with sufficient insufflation should be stored in the PACS Page 32 of 63

33 Fig. 10: Coronary CT scan, Reformat: coronal oblique, Effective slice thickness: 0.8 and 1mm, Maxi intensity projection: 1.5mm Page 33 of 63

34 Fig. 11: Coronary CT scan, Curvilignar Reformat, Effective slice thickness: 0.8 and 1 mm Page 34 of 63

35 Fig. 12: Coronary CT scan, Reformat: elongated mode, Effective slice thickness: 0.8 and 1 mm Page 35 of 63

36 Fig. 13: Thoracic aorta CT scan, Reformat: sagittal oblique, Effective slice thickness: 0.9, 1.25 and 1.4 mm Page 36 of 63

37 Fig. 14: Total aorta CT scan, Reformat: coronal volume rendering, Effective slice thickness: 0.9 and 1.4 mm Page 37 of 63

38 Fig. 15: Abdominal Aorta CT scan, Reformat: coronal view, Effective slice thickness: 1, 1.25 and 1.4 mm, Maxi intensity projection: 11 mm Fig. 16: Arteries of lower extremities CT scan, Reformat: coronal Effective slice thickness: 1, 1.25 and 1.4 mm Maxi intensity projection: 12 mm Page 38 of 63

39 Fig. 17: Chest CT scan, Reformat: axial view, Effective slice thickness: 1.5 and 3 mm Fig. 18: Chest CT scan, Reformat: sagittal, Effective slice thickness: 1.5 and 3 mm Page 39 of 63

40 Fig. 19: Chest CT scan parenchyma reconstruction, Reformat: coronal oblique, Effective slice thickness: 0.8, 1 and 1.4 mm Fig. 20: Chest-Abdomino-pelvic CT scan, Reformat: coronal, Effective slice thickness: 1.5 and 3 mm, Maxi intensity projection: 6 mm Page 40 of 63

41 Fig. 21: Hepatic CT scan multiphases, Reformat:axial view, Effective slice thickness: 2 and 3mm Page 41 of 63

42 Fig. 22: CT uronography, Reformat: axial view, Effective slice thickness: 1.5 and 2.5 mm Maxi Intensity Projection: 4 mm Page 42 of 63

43 Fig. 23: CT uronography, Reformat: coronal, Effective slice thickness: 1.5 and 2.5 mm Maxi Intensity Projection: 11 mm Page 43 of 63

44 Fig. 24: Virtual Colonoscopy CT scan, Reformat: virtuel endoscopic images unfold mode, Effective slice thickness: 1.4, 2 and 2.5 mm Page 44 of 63

45 Results The average data volume was calculated based on the total number of examinations per month (n=707) to gauge the impact on the PACS system. Number of exams per scan type and per month: Fig.: Number of exams per scan type and per month References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE All reference values were added to an excel board which was used to treat all elements of the study. According to average patient and scan type rates per month, average aquisition lenght and resulting number of images per examination type, information about the image data Page 45 of 63

46 volume depending on its matrix; average monthly data volume with native slice thickness and higher slice thickness could be determined. Total volumetry of MDCT 256 slices per month and annual estimation with native and higher slickness: This graphic shows the estimated total volumetry generated by MDCT 256 slices; storing natives slice thickness and higher slice thickness. By applying the new reconstruction parameters for each scan type, data volume decreased significantly. Volumetric savings of 55 % can be observed.the annual saving could reach 1.74 TB, which is 45% more than the total data volumetry of all medical imaging departments sent to PACS per month. Fig.: Total volumetry of MDCT 256 slices per month and annual estimation with native and higher slickness References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Page 46 of 63

47 Estimated volumetry in GB per scan type and per month: The two following graphics show the distribution of the estimated volumetry generated by MDCT 256 slices by storing natives slice thickness and higher slice thickness according to each scan type per month. The top 3 most important exams in terms of volumetry are the chest-abdomino-pelvic CT scan, followed by chest CT scan and total aorta CT scans. Note that data volume depends on image matrix. The data volume per image for lung 2 reconstructions is therefore doubled due to the 768 matrix. Fig.: Volumetry per scan type per month with native and higher slice thickness References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Page 47 of 63

48 Fig.: Volumetry per scan type and per month with native and higher slice thickness References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Savings in GB per scan type and per month: This graphic shows the distribution of the estimated total savings of 148 GB according to each scan type generated by the MDCT 256 slices if higher slice thickness will be stored in PACS. Page 48 of 63

49 Fig.: Estimated Savings in GB per scan type in June 2009 References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE Volumetry in Gigabyte (GB) and % of repartition for all medical imaging departments: The two graphics represent the repartition of volumetric data of the MDCT 256 slices by storing native images and by storing higher slice thickness compared to all other medical imaging departments maintaining their current data volume. The yellow part symbolizes the estimated 12 % of savings attained by the MDCT 256 slices regarding the current monthly 1.2TB total volume for all departments. Page 49 of 63

50 Fig.: Total monthly volumetry of all departements References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE What does the total monthly data volume of 1.2 TB sent to PACS at our hospital represent in termes of video, audio or text? compressed episodes of a TV serie, hours or 139 days non stop broadcast pieces of compressed music, hours or 347 days non stop listening pages of Text using Microsoft Word. Some technical terms for your information: A byte is a unit of digital information in computing: Page 50 of 63

51 1 Kilobyte (kb) = byte 1 Megabyte (MB) = kb 1 Gigabyte (GB) = MB 1 Terabyte (TB) = GB Matrix: Matrix 512 : 0.5 MB per image Matrix 768 : 1 MB per image 2 2 Images for this section: Page 51 of 63

52 Fig. 1: Number of exams per scan type and per month Page 52 of 63

53 Fig. 2: Volumetry per scan type per month with native and higher slice thickness Page 53 of 63

54 Fig. 3: Volumetry per scan type and per month with native and higher slice thickness Page 54 of 63

55 Fig. 4: Estimated Savings in GB per scan type in June 2009 Page 55 of 63

56 Fig. 5: Total monthly volumetry of all departements Page 56 of 63

57 Fig. 6: Total volumetry of MDCT 256 slices per month and annual estimation with native and higher slickness Page 57 of 63

58 Conclusion Storing images with higher slice thickness and larger reconstruction interval generated by a new generation MDCT 256 slices may be a valuable method to enhance online capacity of PACS. Total volumetric data saving up to 55 % monthly may be achieved without major penalty for subsequent comparison of CT exams. Additional savings might be possible by applying the other proposed means to reduce data volume but they are difficult to estimate in advance. Can the objective of 30% global volumetric data savings for all medical imaging departments be attained? The requested saving was 400 GB of volumetry per month for all medical imaging departments. (30% of monthly 1.2TB) Considering that MDCT scan represents nearly one quarter of the total volumetry of the hospital, its proportionally required saving would be 92 GB (23% of the total volume) to obtain the objective of 30% global savings per month if all departments participated at the same ratio. If PACS storage of reconstruction with higher slice thickness would be applied, the volumetric saving of the MDCT 256 slices could reach 148 GB, representing 41% of the total data volume of all departements and demonstrating that the objective for the 256 slices scan of 92 GB could be surpassed by 35%. The other departements are therefore allowed to provide less savings than proportionally required to attain the 30% global volumetric data savings. Page 58 of 63

59 Fig.: Repartition of monthly savings of all departments in Gigabyte (GB) and % of partition References: B. Begenau; Radiologie, Hospital Pitiè Salpetrière, Paris, FRANCE How to succeed in such a project? Find and put a person in charge to coordinate all essential required actions to ensure project success, Communicate extensively so that the project becomes an issue for everyone, Have oversight of all actions; calculate obtained savings throughout the project. This way you can, if necessary, re-focus and make corrections at an early stage. Page 59 of 63

60 Images for this section: Fig. 1: Repartition of monthly savings of all departments in Gigabyte (GB) and % of partition Page 60 of 63

61 References 1.National institut of statistics and economy (french website): Institut national de la statistique et des études économiques 2.Philippe Coulon, Clinical Scientist Philips Healthcare, France 3.Pascal Giry, Person in charge Department of Informatics Pitié Salpêtrière Hospital 4.Sebastian Georget, Person in charge Biomedical Departement Pitié Salpêtrière Hospital 5.Philippe Grenouilleau Person in charge of operations HE / BU Imaging Informatics - RIS/PACS Agfa HealthCare, France 6.Wikipedia: Page 61 of 63

62 Personal Information Bettina Begenau Ct Radiographer Senior, MDCT 256 slices Page 62 of 63

63 Pitiè-Salpetrière Hospital, Department of Radiology Prof. Philippe Grenier 47-83, boulevard de l'hôpital PARIS Cedex 13 France Page 63 of 63