Understanding CT image quality

IAEA RER/9/135 COURSE ON OPTIMIZATION IN COMPUTED TOMOGRAPHY Sofia, Bulgaria, 2017 Understanding CT image quality Dean Pekarovič UMC Ljubljana, Institute of Radiology Quality and Safety office

Role of radiographer Gatekeepers for delivered dose to patient. In optimisation process we can define different roles. Radiologist (expert) Define ImQ Visualise anatomy and pathology Should understand how exposure parameters affect dose Radiographer (expert) Implementation of tips and tricks Manipulation with X- ray modality Especially for individual patient (size, age, condition, clinical indication..) Medical Physicist (expert) Understand technology Acceptance test QC Dose and ImQ Evaluation - DRL Train all tips and tricks

CT comparison Radiologist compare ImQ MP can perform QC test Radiographer? accept protocol (vendor/hospital) and with need or time they are optimised

Radiographer explore CT What IF Radiographer evaluate the CT scanner possibility according to the background knowledge! Experienced radiographer What we need : Competent to evaluate ImQ regarding clnical needs and age(size) Dose understanding No hard formulas Test Object Appropriate questions for Radiologist and MP!!

CT exposure parameters Direct Indirect Localizer Type Sequential or Helical Scan length kv mas AEC Rotation time Table feed/speed and Pitch Collimation/detector configuration Number of phases Reconstruction Slice thickness Reconstruction algorithm(kernel) FBP, IR

Localizer Start and end of scan range. Used for appropriate tube current modulation (AEC in use). Incorrect position in isocenter leads to inadequate tube current modulation (higher or lower). Incorrect position lead to incorrect geometry and size of scanned object. Body region CTDI (mgy) E_DLP (msv mgy -1 cm -1 ) +2,5 cm (msv) Chest x-rays Head 60 0,0021 0,32 3 Thorax 12 0,014 0,42 4 Abdomen 15 0,015 0,56 6

Isocenter geometry 1. Positioned in isocenter measured diameter = 330 mm in center position 2. Positioned above isocenter (+6 cm) observed diameter increased to 342 mm in topogram geometry Image courtesy: Jim Kofler, Ph.D. Too close Isocenter Too far

Isocenter noise and CR Cylindrical 16 cm CTDI-head phantom scanned in center position and while lowered by 60 mm. Noise (1SD HU) increases vertically across the phantom in lower position as the beam shape is non-optimally targeted in the scan. isocenter isocenter noise increase slight contrast change 20 110 18 100 Noise (1SD HU) 16 14 12 10 Contrast (HU) 90 80 70 8 0 50 100 150 y-coordinate (mm) centered 60 mm below 60 0 50 100 150 y-coordinate (mm) centered 60 mm below Ref : Eurosafe, M. Kortesniemi, D. Pekarovic, D.Sheppard

Effect of Topogram on CTDIvol Planning part (before SPIRAL scan was performed and after each SCOUT) After scan was performed part (patient protocol info) Weight (kg) Height (cm) Age Gende r PA Scout CTDI vol LAT Scout CTDI vol mas CTDI vol ref. mas mas* Collim. Cycle time 1 78 167 1951 F 27,65 14,71 222 12,17 210 184 24x1,2 0,5 2 78 160 1942 F 16,91 12,01 181 10,30 210 155 24x1,2 0,5 3 79 174 1936 M 15,37 11,24 170 9,80 210 148 24x1,2 0,5 4 63 167 1947 M 13,83 10,85 164 9,31 210 140 24x1,2 0,5 5 100 190 1954 M 16,64 12,39 187 10,85 210 164 24x1,2 0,5 6 70 163 1939 F 14,49 10,91 165 9,31 210 140 24x1,2 0,5 7 78 169 1959 F 18,29 12,45 188 10,3 210 155 24x1,2 0,5 8 80 178 1930 M 11,84 9,20 139 8,26 210 159 24x1,2 0,5 9 65 167 1966 F 10,69 8,53 118 6,97 210 122 64x0,6 0,5 10 75 155 1968 F 11,06 11,00 152 8,83 210 122 64x0,6 0,5 CTDI vol, DLP and effective dose reduce cca 20 % Evaluate your CT!

AXIAL vs. HELICAL AXIAL Better ImQ No windmill artefacts Lower dose No 3D recon No thin slices HELICAL Lower ImQ Windmill artefacts PFO Higher dose 3D yes Thin slices available Detector configuration??? 16 x 0,75 or 16x 1,2 Lower is Better + Less Scatter and Better slice sensitivity profile - More rotations and slightly higher dose Is it true for both technique Axial and Helical?

Seq vs Helical Seq kv 120 mas 400 AEC off Collimation 1,2 Slice thic. 3 mm CTDI vol 57,09 (16cm) DLP 480 Seq kv 120 mas 400 AEC off Collimation 3 Slice thic. 4,8 mm CTDI vol 53,71 (16cm) DLP 290 helical kv 120 mas 380 AEC off Collimation 0,6 Slice thic. 3 mm CTDI vol 54,71 (16cm) DLP 968 helical kv 120 mas 285/410 AEC on Collimation 0,6 Slice thic. 3 mm CTDI vol 41,10 (16cm) DLP 670 3D reconstructions, Artefactcs, No gantry tilt, Lower noise (ImQ)

kv Lowering kv will results Improved contrast Lower dose (if mas constant be careful with AEC) More artefacts (beam hardening and photon starvation) corrected with IR More noise (can be reduced with Iterative Reconstructions)

kv which effect we want more Adapted from Yu et al. 2011 and Fletcher et al. 2009 Attenuation differences (contrast) decrease with increased photon energies when Compton effect dominates increasingly over photoelectric effect. Due to k-edge of iodine, use of lower kilovoltages may provide contrast and optimisation benefits, depending of the iodine enhancement and object size (which affects to total attenuation). Ref: Eurosafe,M.Kortesniemi,D.Pekarovic

kv and dose Dose ratio with different voltages 80 kvp 100 kvp 120 kvp 140 kvp 35% 64% 100% 140% Dose Voltage 2.5 Dose increases strongly with higher tube voltage, and radiation is more penetrating. Thus, contrast is decreased but if ma is not changed, noise is also decreased. Ref: Eurosafe,M.Kortesniemi,D.Pekarovic

kv and contrast Fixed mas and other exposure parameters Lower kv - Increase contrast, increase noise Tip: Iodine, MSK, Chest Paediatric lower attenuation kv Bone Nylon Air 120 1496 98 1004 100 1685 88 1005 80 2023 71 999 kv mas CTDIvol DLP 120 380 53,68 220 100 380 32,28 132 80 380 15,02 62

kv - noise Fixed mas and other exposure parameters Lower kv Increse nosie Tips : Collimation, slice thickness, AEC (increase ma), rotation time kv mas CTDIvol DLP 120 380 53,68 220 100 380 32,28 132 80 380 15,02 62

kv and HU kv 100 mas 88 AEC ON CTDI vol (32 cm) 5,21 mgy Collimation 64x0,6 Sl thic. 1 mm kv 120 mas 76 AEC ON CTDI vol (32 cm) 9,18 mgy Collimation 64x1,2 Sl thic. 2 mm

Hip 80 kv Routine 120 kv AEC ON kv 80 mas 205/210 64 x 0,6 CTDIvol 3,89 DLP 129 Lenght 32,3 cm 100 kv 3/3 5/5 3/3 X 2 3/3 80 kv 100 kv 5/5 3/3 AEC ON kv 100 mas 193/210 64 x 0,6 CTDIvol 8,19 DLP 265 Lenght 33,1 cm 33 y, 72kg

Knee kv 100 mas 94 1 mm 0,75 CTDIvol 5,93 DLP 90 RECON 2/2 kv 120 mas 102 1 mm 0,75 CTDIvol 9,25 DLP 181

Paediatric 3y 100 Kv AEC ON 90mAs/110 16x0,75 (3mm) CTDIvol 22,5 DLP 380 2,5 y 80 Kv AEC Off 60mAs 16x0,75 (2mm) CTDIvol 1,85 DLP 14 2 y 100 Kv AEC On 25mAs/25 16x0,75 (2mm) CTDIvol 1,32 DLP 24 Images Courtesy : UMC Ljubljana

DRL Comparision Leta (l) Kv mas ref mas CTDIvol (mgy) DLP (mgycm) 54 100 96 200 4,09 122 49 100 136 210 5,78 182 32 100 156 250 6,63 229 48 100 200 200 8,54 296 46 100 237 210 10,01 282 54 100 98 210 4,19 140 32 100 156 250 6,63 215 49 100 113 210 4,81 141 48 100 205 200 8,68 304 46 100 160 200 6,77 199 28 100 160 180 6,79 191 23 100 216 210 9,17 227 Povprečeje 6,85 210,6 DRL CT Pelvis in SI is 525 mgycm - 40% lower than DRL SI Lower ImQ only for 3D navigation

Automatic kv selection Care kv Siemens kv Assist GE Crucial Localizer tissue attenuation (patient in isocenter!). Do we need one or two localizers (in different angle AP,LAT at 90 Δ). Defined noise (mas, Noise Index) by user. Suggested regarding clinical needs (MSK, Iodine, LCR-soft tissue). System adopt mas and kv to gain same image noise (ImQ).

Summarize The lower the tube voltage the higher the iodine contrast in the image because the average energy of the spectrum gets closer to the k-edge of iodine. In small patients, has almost no effect. Larger objects absorb a significant amount of low energy photons noise can rise dramatically For small patients the CNR increases when going to lower kv, for bigger patients it has a maximum and drops again when lowering the tube voltage. Ref: https://health.siemens.com

Tube current -AEC Fixed or AEC. Fixed To avoid tube current manipulation. Paediatric very small patients below e.g. 10 kg. Size or indication specific (shoulder arthro..). AEC AEC standard vendor specific regarding x,y and z axis tube current modulation. Requires localizer (1 or 2). AEC change when slice thickness, pitch and kv (vendor specific). Do we understand how it works on our modality.

AEC well known facts 200 180 160 140 120 AEC On 120 kv 125mAs/210 ref. mas 64x0,6 mm (1,5 mm) Pitch 0,6 Rot. time 0,5 s Scan time 9,57 s CTDI vol 9,03 mgy DLP 198,6 mgy*cm 100 80 AEC ON AEC OFF 27 mas 165 mas 60 40 20 0 1 5 9 13172125293337 414549 535761 65 AEC Off 120 kv 160mAs 64x0,6 mm (1,5 mm) Pitch 0,6 Rot. time 0,5 s Scan time 9,57 s CTDI vol 11,62 mgy DLP 255,6 mgy*cm 160 mas 160 mas

AEC noise level Can be defined by user. Time to adopt or general use? Very strong Average Very weak Test Object 16 cm Cylindrical PMMA Type kv mas/ ref CTDIvol mgy (16 cm) DLP mgy*cm Detector 1.Very strong 100 104/160 8,47 123 64x0,6 2. Average 100 115/160 9,39 136 64x0,6 11% Advantages? Planned for different patient size 3.Very weak 100 135/160 11,01 149 64x0,6 30%

AEC - Noise or Ref mas Always same upper level of noise what about patient size kv 120 mas 190/ ref 180 Collimation 64x1,2 Rot Time 0,5 s Slice thic. 2mm CTDIvol 12,66 mgy (32 cm) DLP 557 Scan lenght 43,9 cm Δ 38 % kv 120 mas 264/ ref 250 Collimation 64x1,2 Rot Time 0,5 s Slice thic.2mm CTDIvol 17,53 mgy (32 cm) DLP 770 Scan lenght 43,9 cm

kv and AEC Art. Phase 100 Kv 114 mas/ ref 180 mas 64x0,6mm CTDIvol 4,84 mgy (32cm) Recon 1mm/0,7mm Recon 2mm/2mm MIP 7/5 - CM Venous phase 120 Kv 127 mas/ ref 140 mas 64x 1,2 mm CTDIvol 8,44 mgy (32cm) Recon 2mm/1mm Recon 2mm/2mm MIP 7/5 Vnous phase

CT Head trauma -4 m 1st AEC No 100 kv 380 mas CTDIvol 32,9 DLP 492 15 cm AEC No 100 kv 190 mas CTDIvol 15,42 DLP 264 17 cm Few h later

Detector configuration vs. dose and noise (AEC on) 120 kv AEC ON 125mAs/210 mas 64 X 0,6 mm Pitch 0,6 Rot time 0,5 s Scan time 9,57 s CTDIvol 9,03 mgy DLP 198,6 mgy*cm 1,5 mm 3 mm 5 mm 120 kv AEC ON 125mAs/210 mas 64 X 1,2 mm Pitch 0,6 Rot time 0,5 s Scan time 6,63 s CTDIvol 8,35 mgy DLP 190,4 mgy*cm 1,5 mm 3 mm 5 mm Tip size of the object, artefact reduction, noise -same

FOV according to the needs Small FOV allows increased spatial resolution until when? Pixel- inversely related to detail, smaller is better Matrix- directly related to detail, bigger is better directly FOV- inversely related to detail, smaller is better Pixel = FOV/ matrix FOV 500 mm Pixel 0,98 mm FOV 100 mm Pixel 0,20 mm Ref: E.Seeram, Computed Tomography: Physical Principles, Clinical Applications, and Quality

FOV noise FOV 300 mm Pixel size 0,6 mm FOV 260 mm Pixel size 0,5 mm FOV 164 Pixel size 0.32 mm Δ tissue

FOV- SR FOV 300 mm Pixel size 0,6 mm FOV 260 mm Pixel size 0,5 mm FOV 164 Pixel size 0.32 mm

Pitch Not so big factor in dose managements anymore. CT control the dose much better. u New scanners (check for your CT!) Siemens, Philips Δ Pitch leads to Δ ma (to get same nosie) Only speed with similar dose (paediatric, cardiac..) GE, Toshiba Some Δ ma when Δ pitch Higher pitch Lower dose (Scanning speed up) Lower pitch Higher dose (scanning speed down)

Pitch-dose is changed (+AEC) Pitch 1 vs pitch 2 (50 % lower dose). TEST CAD 1 Kv mas/ref. CTDIvol(mGy) DLP mgy x cm rot. time c Sl mm Dosr L/S Pitch 0,8 120 213/330 14,30 373 1,00 1,20 0% L/S Pitch 1,0 120 219/330 14,48 395 1,00 1,20 6% L/S Pitch 1,5 120 233/330 15,42 455 1,00 1,20 22% Example 16 slice CT Evaluate for your scanner! Pitch 0,8 Pitch 1,0 Pitch 1,5

Pitch dose is constant (+AEC) Δ Pitch Noise??

Pitch-dose is constant-3d Be careful some vendors are changing tube current to achieve constant ImQ. Then you will get only speed. Pitch 0,5 0,8 1,2 1,5 Pitch 0,5 0,8 1,2 1,5

Bow tie filters Imege Courtesy : http://xrayphysics.com 120 kv 200mAs AEC no Rot time 0,5 64x0,6 CTDIvol 11,47mGy (32cm) Type ( 0018,1160) : Wedge_2 120 kv 200mAs AEC no Rot time 0,5 64x0,6 CTDIvol 14,42mGy (32cm) Type ( 0018,1160) : Flat Is the scanner component that modifies the energy spectrum and spatial distribution of the primary beam. Affects CTDI vol. Inside organ program (difficult to see).

Rotation time CTDI vol may not change in the expected manner if the scanner automatically adjust other parameters when the exposure time per rotation is changed. The relationships between CTDIvol and exposure time per rotation is vendor dependant. 0,33 s 0,5 s 1 s Rotation Time Scan Time (s) kv mas/ ref CTDIvol mgy (32 cm) DLP mgy*cm Detector? 0,33 s 2,3 120 270/275 17,88 286 24x1,2 Δ aprox. 38% 0,5 s 3,48 120 388/275 25,72 412 24x1,2 Why? 1 s 6,96 120 377/275 24,98 400 24x1,2

Noise and slice thicknes 1 mm 2 mm 3 mm 5 mm Until when?

Reconstructed slice thickness 16 x 1,5 ; 2/1 recon 16 x 1,5 ; 3/3 recon 2/1 recon 3/3 recon 3/3 Image man. 2/1 Image man..

Reconstruction algorithms - Kernel Impact on : Noise/ Low Contrast Resolution Sapatial Resolution Important : Be aware of quality of diagnostic monitors - radiologist. Additional reconstruction is for free and takes 20 seconds

Play, explore 120 kv 166 mas 0,6 mm sl. Thic. /0,5 mm Increment Bone window Bone kernel 120 kv 166 mas 1mm sl. Thic. /1mm Increment Bone window Bone kernel 120 kv 166 mas 1mm sl. Thic. /1mm Increment Bone window Soft kernel 120 kv 166 mas 1mm sl. Thic. /1mm Increment Bone window Bone kernel 120 kv 166 mas 0,6 mm sl. Thic. /1mm Increment Bone window Soft kernel DLP 144 mgy*cm

Extented CT Values PigTitanum and Steel Soft window+ Extenced CT, Bone window 140 Kv 80 kv 80 mas 120 mas 0,6 mm 1,20 mm B 30f B70f Ref: A.Janežič, UMC LJ,SI

Dual Energy metal arifact MIP 3 mm/2 mm TRA 3MM/2MM VRT CTDIvol 7,99 DLP 160 mgycm SAG 3 mm/2 mm

Conclusion Explore your CT Use existing experience After acceptance test ask MP and Radiologist to get additional information Evaluate and optimise CT protocols Understanding the NOISE.

Thank you!