Abdominal DECT: How to Integrate Into Your Practice

Transcription

1 Abdominal DECT: How to Integrate Into Your Practice Eric Tamm, M.D. Department of Diagnostic Radiology Division of Diagnostic Imaging MD Anderson Cancer Center Houston, TX

2 Disclosure I have no relationships with commercial interests to disclose. 2

3 Overview Background of multispectral (MSpCT) imaging Ideas for where MSpCT can be integrated into an abdominal imaging practice (applications) How to create/convert to a dual energy protocol (RSDECT, DSDECT) Workflow, scanner to PACS Workflow, an approach to reading studies 3

4 Background of MSCT Different Materials Attenuate Differently from one another at Different X-ray Energies 4

5 Types of Multispectral CT Cammin Proc SPIE 7961, 2011 DSDECT RSDECT SS Dual Layer MSCT Flux 80 kvp 140 kvp Flux 120 or 140 kvp kvp DICOM 140 kvp DICOM Raw Data Raw Data MDI, MONO, Blended PC MDI, MONO, QC image 120 kvp DICOM MDI, MONO 5

6 Two Material Decomposition= Material Density Images (GE, Phillios) Known curves (NIST) Unknown Material Iodine Curve Water Curve 70 kev Water(- iodine) ROI=1022 mg/cm3 Iodine(- water) ROI=27 mg/cm3 Each material has unique attenuation curve Each curve can be represented by combination of two curves of known materials Can decompose voxel to the amount of water and iodine needed to mimic the actual curve of that voxel 7

7 (DSDECT): 3 Material Decomposition, VNC and Iodine Image Virtual NonContrast Iodine Map 9

8 Iodine: Iodine Map and Iodine Overlay 10

9 Monochromatic Images Conventional images are formed from polychromatic sources Monochromatic images show image data as if an x-ray source of only one energy (monochromatic), i.e. 70 kev had been used Terms to know: kvp= the maximum energy in a polychromatic spectra kev= the unit of energy used to denote one energy level. Used to describe a monochromatic beam. Flux Polychromatic Dual Energy Spectra 70 kev 80 kvp Energy 140 kvp 11

10 40-140keV Monochromatic Energy Images 40keV 60keV 80keV 100keV 120keV 140keV 12

11 Blends of Polychromatic Beam (DSDECT): Low, Optimized, High Blends (80-100&140KVP) 3mm Blended Emphasizing Low Energy 3mm Blended Emphasizing Optimum Contrast 3mm Blended Emphasizing High Energy 13

12 Where Can DECT Be Integrated Into an Abdominal Imaging Practice (Applications) 14

13 Broad Classes of Applications Iodine: Increasing its conspicuity. Material composition What is it made of? Removing a material from images Removing/decreasing artifacts from metal 15

14 Hepatocellular Carcinoma 70keV ~ 120kvP 40 kev Iodine (-water) Material density *Shuman, W. P. (2014). AJR. 16

15 Small Bowel Carcinoid Tumor Late Arterial Phase 70keV 70 kev 50keV Iodine(water) 17

16 Pancreatic Cancer: Increasing Conspicuity 120 kvp image, at baseline 70 kev axial, 1 month later Iodine (-water) Material Density 1 month later 18

17 Visualizing Internal Enhancement of Cysts 70keV 50keV 20

18 Color Map to Identify Enhancement Iodine (-water) image 70 kev image Iodine overlay color map

19 Increasing Conspicuity of Vessels 50 kev axial MIP Late arterial (PP) phase* *Rauch et al. RSNA

20 Material Composition: What is it made of? 25

21 Tools Virtual unenhanced (VUE): Are several evolving types (2 MD, 3MD, MSI) May reduce dose by avoiding pre contrast Unique material decomposition Comparison against NIST values (Virtual atomic number, histogram, spectral HU curves) 26

22 Virtual Unenhanced Images: Types mono kev Energy level furthest from k-edge of iodine (33keV), minimizes iodine s impact Water(iodine)= water MINUS iodine, MDI Problem: Units are unfamiliar mg/cc, not HU Material Suppressed Imaging for Iodine ( MSI ) Substitute blood for equal volume of iodine= HU! 3 material decomposition = HU! VUE values may*, or may not, depend on contrast phase from which VUE images are derived Relationship between various VUE and true non-contrast is not simple** *Sahni, V. A. (2013). Clin Radiol, ) ** Morgan, D. E. (2013) True Unenh MSI 27

23 Lipid Rich Adenoma: Late Arterial Phase RSDECT 70kev, postcontrast, 90HU True 140 kev MSI water(iodine) fat(iodine) noncontrast 8 HU 11.8HU 16.4HU 942mg/cc 988mg/cc Lipid rich vs. poor adenoma * 9.5 sens 64% spec 94% LRA * Morgan, D. E. (2013). JCAT **Glazer AJR 2014,***Mileto Radiology 2014 * 994 sens 59% **<992 LRA * 987 sens 50% LRA *** 997 Adenoma vs Nonaden, sens 96% spec 100%

24 Ex: Renal Calculi 70keV True noncontrast 140 kev MSI water(iodine) HU mg/cc HU 30

25 Stone Analysis: Pitfall Atomic Z : Theoretical atomic number for a material behaving same as that in a given voxel at 80/140kVp, compare with knowns Pitfall: Volume averaging with iodine. Best approach: DECT on unenhanced scan (Wang, J. (2012). Eur Radiol) 31

26 Removal of a Material From Images 33

27 Angiography MIP 70 kev MIP iodine (HAP) 34

28 Pancreatic Cancer: Resected, Surgical Clips in Operative Site 120 kvp, polychromatic 86 kev, monochromatic 86 kev, monochromatic MARS (metal artifact reduction sequence)

29 Workflow: Creating Protocols Converting Existing Protocols to Dual Energy RSDECT vs DSDECT 41

30 Context: Scope of MD Anderson Operations 21 CT Scanners In 4 Separate Facilities 72 CT Technologists 53 Nurses 120 Radiologist (48 Body) 48 residents 9 Fellows pts. Per wkday 200 pts. Sat & Sun. 80% = body: AP, CAP 42

31 Creating a RSDECT Protocol RSDECT can t use tube current modulation, but does have ASIR noise reduction (can choose how strong) Manufacturer has created presets: Are combinations of tube rotation, fixed tube current, filter, and pitch for body sizes that produce a fixed CTDI These are listed in a table Based on a needed CTDI, can choose appropriate preset For abdomen, usually large body 43

32 Converting to a RSDECT Protocol Identify phase to convert to DECT e.g. late arterial phase Choose a parameter for patient size DFOV, diameter, BMI Get list of patients For each patient study, record Size dimension CTDI (tube output) Sort patients by dimension Identify convenient breakpoints for size Note the highest CTDI for each group and choose matching GSI preset Modify timing based on pitch/ rotation time, scan time for GSI At end, have a protocol for each size range. CTDI N1 CTDI N2 CTDI N3 DFOV GSI x GSI y GSI z NonDECT 44

33 Alternative: AutoGSI When technologist puts in noise index that is wanted Scanner evaluates scout Suggests GSI setting to use that will produce closest average CTDI for scan to achieve that NI Noise Index + GSI Preset 45

34 RSDECT: MDACC Scanner to PACS Workflow Dual energy data pushed to AW Server and PACS keV with full GSI data GE Advantage Windows Thin Client-Server Used for problem solving Constraints: Limited installations, user needs to be comfortable with software Scanner can automatically create & push to PACS: Water/Iodine/calcium MDI MPRs, kev monochromatic QC : Quality control= 140kVp image* AW Server *manufacturer says not for diagnosis RSDECT isite PACS 46

35 Example: Pancreas Protocol: Data to PACS 3.8mm True UE 2.5mm QC (140kVp) 0.6mm70keV DE Data 0.6mm PV phase (120kVp) 2.5mm sag PV phas mm 70kev 2.5mm 50keV 2.5mmH2O(-iod)VUE 2.5mm iod(-h2o) 2.5mm cor PV phase

36 Converting to a Dual Source DECT Protocol Tube current modulation is available as are noise reduction techniques Can choose 80 or 100kVp for lower energy, (usually 100kVp), and 140kVp for larger Limiting factor: dual energy data only available over coverage of both tubes (33 cm) Pathology needs to be within circle 48

37 Converting to a DSDECT Protocol Identify phase to convert to DECT Identify reference ma for phase For multiphasic, will be DE on those close together, e.g. Art/PV Set Tube A 80 or 100 kvp, Tube B as 140kVp Create image sets for recons (0.6-2mm, q kernel) from Tube A &Tube B Choose nature of blended polychromatic (e.g. optimum), thickness (3mm), kernels (e.g. i30 or i31, and strength (f1-3) Decide upper limit of patient size on which to do dual energy (33cm circle) Monochromatic, virtual noncontrast, iodine quantitation images all are created on workstation Radiologist uses Syngo Via thin client software Alternative: create 3mm slices of 100kVp Tube A which is of entire FOV 49

38 DSDECT: Image Sets at Scanner Pushed to PACS 0.75mm LA TubeA 100kVp 0.75mm LA TubeB 100kVp 0.75mm PV TubeA 100kVp 0.75mm PV TubeB 100kVp 3mm True UE 3mm Opt LA PC Blend 3mm OPT PV Blend 3mm Cor PV Blend 3mm Sag PV 50 Blend

39 Blends of Polychromatic Beam: Low, Optimized, High 3mm low KVP Blended PC 3mm opt Blended PC High low KVP Blended PC 52

40 100kVp Tube A versus Optimized Mono + 3mm Tube A 100 KVP 3mm Tube A 100 KVP 53

41 Virtual Unenhanced and Fused Iodine Map 54

42 Future Directions: Automated Processing Current: isite PACS Syngo Via Future: Virt Uneh, iodine, mono Syngo Via 55

43 Pitfalls/Solutions 56

44 Breast Shield: Streak Artifact Solution: No breast shields for DECT studies Decrease dose for chest portion of studies 57

45 Arm Down: Streak Artifact Use non-dect if arms, tubes, lines are anticipated as a problem DECT PP Phase Non-DECT PV Phase 58

46 Blurring (Large Patients, DFOV 48) Other solutions: Edge enh. Filters or Use Non-DECT Techniques DECT 70 kev monoch Solution: 140 kvp QC Image

47 Alternative Solve: 80 kev Monochromatic, With or Without Edge Filter 2.5mm 70 kev 3mm 80 kev 3mm 80 kev Edge Filter 140kVp QC

48 Workflow: Approach to Reading a Study

49 Image Review: Ex: Pancreatic Cancer PV Phase 70 kev PP & 120 kvp PV in same stack Iodine MDI Max conspicuity **Thin client workstation software (rarely) for problem solving on PACS stations. 50 kev Problem solver 62

50 Reviewing DSDECT Images Syngo Via & Thin Client PV Phase Opt Blend 3mm PC Mono plus 40 kev Fused iodine 63

51 Conclusion Evolving approaches to multispectral imaging Several applications are developing in the abdomen Different workflows depending on the type of scanner Must keep in mind what can be done on the scanner, ease of use of workstation software, and work volume Can constrain use of the workstation to problem solving Can create several different types of images These can be used efficiently, as in MRI 64

52 Thank you!