7/21/2014 - amos onlineamos3.aapm.org/abstracts/pdf/90-25729-346462-108671.pdf · 7/21/2014 1 1...

7/21/2014

1

Philips Research - Hamburg Confidential 1

“Partners in Solution” session

Philips CT Dose Optimization tools and Advanced

Reconstruction

Tsvi Katchalski Philips

July 24, 2014


Content (I) Dose Reduction and Dose Optimization Technologies

• Dose Right and Tube Current Modulation (Dose Right, Z-DOM, 3D-DOM) • Special Acquisition Modes (Cardiac Dose Right, Prospective gating) • Beam Filtration and Beam Shapers • Eclipse Collimator and Clear Ray Collimator • NanoPanel Elite Detector • NEMA XR-25 – Dose check – Dose structured reporting

(II) Advanced Reconstruction Technology

• iDose4

• IMR (Iterative Model)


PART (I)

(I) Dose Reduction and Dose Optimization Technologies

• Dose Right and Tube Current Modulation (Dose Right, Z-DOM, 3D-DOM • Special Modulation Modes (Cardiac Dose Right, Prospective gating) • Beam Filtration and Beam Shapers • Eclipse Collimator and Clear Ray Collimator • NanoPanel Elite Detector • NEMA XR-25 – Dose check – Dose structured reporting (II) Advanced Reconstruction Technology

• iDose4 • IMR (Iterative Model)

7/21/2014

2


Dose Right and Tube Current Modulation Dose Right Index Consistent IQ for Different Patient Size Dose Right Index + Current Modulation Consistent IQ for Different Patients and at Different Z-Locations

DoseRight recommendation: 389mAs DoseRight recommendation: 240mAs


Dose Right Index – Definition (by Reference)

Dose Right – How Does it Work

Averages patient size is calculated from the surview

Patient size is compared to the EC

Reference size

Avg. mAs for the patient is calculated

according to the DRI

DRI=24 200 mAs @120 kV @ average patient size (29 cm)

0

100

200

300

400

500

600

700

800

900

1000

0 10 20 30 40 50

Cu

rren

t x

Tim

e [m

As]

Average Size [cm]

Exponential Adaptation to Patient size


Increase and Decrease in DRI (Dose Right Index) - Adults

DRI 25cm 29cm 35cm

…

18 71 102 202

19 80 114 226

20 90 128 254

21 101 143 284

22 113 160 317

23 127 179 355

24 142 200 400

25 159 224 443

26 178 250 497

…

1 DRI:

+12% mAs

6 cm: +100% (doubled) mAs

6 DRI:

+100% mAs

6

7/21/2014

3


Z- Dose Modulation (Z-DOM) Consistent IQ Along Patient Liver Detection Head Detection

0

100

200

300

400

500

0 5 10 15 20 25

mA

s

Z-position


3D- Dose Modulation (3D-DOM) Reduction of Streak Artifacts Improve Noise Uniformity


3D- Dose Modulation vs. Z-DOM - Reduction of Streak Artifacts - Improve Noise Uniformity

7/21/2014

4


Automatic Pitch/Rotation Time Algorithm Basic equations:

𝑠𝑐𝑎𝑛 𝑡𝑖𝑚𝑒 =𝑠𝑐𝑎𝑛 𝑙𝑒𝑛𝑔𝑡ℎ

𝑏𝑒𝑑 𝑠𝑝𝑒𝑒𝑑

𝑏𝑒𝑑 𝑠𝑝𝑒𝑒𝑑 =𝑝𝑖𝑡𝑐ℎ × 𝑐𝑜𝑙𝑙𝑖𝑚𝑎𝑡𝑖𝑜𝑛

𝑟𝑜𝑡𝑎𝑡𝑖𝑜𝑛 𝑡𝑖𝑚𝑒

𝑚𝐴𝑠 =𝑚𝐴 × 𝑟𝑜𝑡𝑎𝑡𝑖𝑜𝑛 𝑡𝑖𝑚𝑒

𝑝𝑖𝑡𝑐ℎ

Bed speed

Pitch / rotation

time

mAs

Tube current


Special Modulation Modes - Cardiac Heart Cycle – The Heart is in Constant Motion

General motion profile: • Relaxation: Heart is filled with blood.

• Contraction: Heart ‘pushes’ blood to the body/Lungs

“Quiet phases”: Phases with relatively less motion • End systolic (“40%”): Relaxation Expansion

• End diastolic (“75%”): Relaxation Contraction

Systole Diastole

“40%” “75%” “0%” “100%”


Cardiac Cycles Reconstructing at Different Phases – ECG Gated

20 35 75

Cardiac Phase

7/21/2014

5


Special Modulation Modes Cardiac Retrospective

• ECG triggered dose modulation in Helical/Spiral scans

57 34

Z

20 %

Cardiac DoseRight - Prospective

100 %


Special Modulation Modes Cardiac Step and Shoot – Prospective Gating

• Low dose (3–5 mSv) diagnostic scan

– Dose reduction up to 80% compared to standard

retrospectively-gated spiral scans

– Image quality comparable to that of standard retrospectively-

gated spiral scans

• 3D axial cone beam reconstruction algorithm

• Real-time arrhythmia handling algorithms


Special Modulation Modes Dose Reduction from ECG Dose Modulation – Dependency on HR

Compared to ECG-gated spiral scans without dose modulation

0

10

20

30

40

50

60

70

80

90

60 bpm 70 bpm 80 bpm 90 bpm 100 bpm S&SC

Do

se R

ed

uct

ion

(%

)

Klass O, Walker MJ, et al, Eur Rad 2008

http://www.wardray-premise.com/xray/misc/signage.html










7/21/2014

6


Beam Filtration and Beam Shapers InteliBeam Filters SmartShape Wedge (Bowtie) Filters

- Filtration is optimized to increase Low Contrast and reduce Skin Dose and is done per protocol

- Three filtration modes: “Half”/”Full”/”Off-center enabled”


Eclipse Collimator and Clear Ray Collimator Helical Dynamic Collimation to Reduce Over Scanning

Walker MJ, Olszewski ME, Desai MY, et al, Intl Journal of Cardiovasc Imaging, 2009


Eclipse Collimator and Clear Ray Collimator 2D Anti Scatter Grid

7/21/2014

7


Clear Ray Scatter and Beam Hardening Correction

- Monte Carlo based scatter estimation

- Analytical Beam Hardening Correction

- Scatter correction is proportional to patient size


NanoPanel Elite New Tile Detector Scintillator

– IGOS (Gd2O2S) High Quality Ceramic Scintillator

– Fast Response

– High Light Output

– Low Cross-Talk

ASIC

– Very Low Noise

– Improved linearity over all the dynamic range

– Front End Electronics with 256 Channels

PDA (Photo Diode Array)

– Back Illuminated Using Thin and High purity Silicon

– High Quantum efficiency.

– High Shunt Resistance

– Low Cross-talk


PART (II) (I) Dose Reduction and Dose Optimization Technologies

• Dose Right and Tube Current Modulation (Dose Right, Z-DOM, 3D-DOM • Special Acquisition Modes (Cardiac Dose Right, Prospective gating) • Beam Filtration and Beam Shapers • Eclipse Collimator and Clear Ray Collimator • NanoPanel Elite Detector

(II) Advanced Reconstruction Technology • iDose4 • IMR (Iterative Model)

7/21/2014

8


iDose advanced reconstruction How it works

iterative reconstruction technique: How does it work?

iDose uses the raw data output from the NanoPanel detector

to determine and remove noise while keeping structure intact


iDose – Mode of Operation

Noise Reduction – Reduce noise according to iDose Level

selection

– For each protocol the achievable iDose

Level were determined

– No degradation in Resolution and Low Contrast

(Tolerance: 10% in MTF10% and 1mm in Low Contrast)

High mA IQ Improvements – Keep the Dose Level the same and improve spatial resolution and Low

Contrast

Hardware Enabler

– RapidView IR, Next Generation Intel

Multi-Core Processors

– Up to 20 Images/sec

Compensating increase in noise

(noise reduction %)

11 %

16 %

23 %

29 %

37 %

45 %

55 %

Philips Research - Hamburg Confidential 24 Philips Healthcare / Clinical Science & Application CT EMEA, June 6, 2008

2

4

50% Dose (125mAs) FBP 50% Dose (125mAs) iDose

Courtesy TU Munich, Germany

Follow up with

50% Dose Reduction

Comparison Full Dose / 50% Dose Reduction + iDose Full Dose (250mAs) FBP

7/21/2014

9


2

5

120 kV, 100 mAs (6.6 mGy)/ 80 kV, 100 mAs (1.9 mGy) 70% Dose Reduction

70% Dose Reduction (1.9mGy) - iDose

Courtesy TU Munich, Germany

Full Dose (6.6mGy) - FBP

Comparison Full Dose / 70% Dose Reduction + iDose


2

6

120kV, 70mAs, 6.4mGy, Prospective Gated Aortic CTA (Step & Shoot Complete)

iDose FBP

Courtesy Cleveland Clinic, USA

Improving Image Quality: Artifact Reduction


2

7

Courtesy OHSU, USA

120kV, 280 mAs, 22.9mGy, Prospective Gated Cardiac CTA (Step & Shoot Cardiac)

iDose FBP

Improving image quality in inherently noisy acquisitions

7/21/2014

10


Knowledge-Based Iterative Reconstruction Technique

Name


Knowledge-Based Iterative Reconstruction

What makes this possible Fast and compact next-gen computing devices

using GPU (under 3min for most protocols)

Integrated Workflow

User has flexibility as to the final result.

Body Soft Tissue Routine Sharp Plus


Performance Evaluation

Low Contrast Detectability A measure of a person’s ability to perform a particular task: The determination of a Low Contrast object

– Is influenced by all Physics IQ metrics (Noise, Resolution, Uniformity and CT number)

– A statistical approach is needed

– Human observer study, 4 Alternative Forced Choice (4AFC) Study was used

Test 42 out of 100 Test 57 out of 100 Test 73 out of 100 Test 77 out of 100 Test 94 out of 100

Test 57 out of 100Test 42 out of 100 Test 94 out of 100Test 77 out of 100Test 73 out of 100

7/21/2014

11

31

0

1

2

3

4

StandardReconstruction

IMR

1.4

3.9

Low

-con

trast

Dete

cta

bili

ty I

ndex 2.5 - 3.6x

Low-contrast Detectability Index*

Standard

Reconstruction

* Low-contrast detectability was assessed using 4-AFC test on a reference abdomen protocol, tested on the MITA CT IQ phantom. Data on file.

Improve low-contrast detectability Revealing subtle differences

32

0

20

40

60

80

100

0 2.5 5 7.5 10 12.5

Gain

(%

)

Frequency (lp/cm)

Standard Resolution MTF Measurements

FBP (Std Filter)

FBP (Sharp Filter)

IMR (High-res setting)

Revealing fine detail

High resolution Imaging

33

GGO visualized on Chest CT with IMR

80 kVp, 10 mAs, 0.2 mGy, 0.11 mSv

Chest X-Ray

0.05 mSv

Standard

Reconstruction

BODY

Courtesy: UCL, Belgium

7/21/2014

12

34

Standard

Reconstruction

120 kVp, 300 mAs, 14.3 mGy, 1.8 mSv, < 2 min recon time

71 year old man with hemorrhagic lesions not seen on FBP

NEURO

1.0 mm slice thickness

Courtesy: UCL, Belgium

3.0 mm slice thickness 1.0 mm slice thickness

Standard

Reconstruction

FBP iDose 4 IMR L3

Non-calcified plaque visualization

Mixed plaque in LM / LAD

in severe obese female : 130kg, 170cm, BMI 45

Step&Shoot scan ! Max tube settings 120kV,

300mAs

7/21/2014

13

FBP iDose

IMR L3

Excessive noise hinders

segmentation

Pre-TAVI assessment Entire ECG Gated Aorta

with ONLY 40cc’s of IV contrast 20cc @2cc/s + 20cc @ 1cc/s

Semmelweis Univ. Budapest

Video: Maimondes Medical Center https://www.youtube.com/watch?v=X1Uk9Jvbut0

TAVI scan with only 40cc of contrast

7/21/2014

14

TAVI ISP6, annular plane

TAVI ISP6, all planes

TAVI ISP6, angle prediction

7/21/2014

15


Acknowledgements - Scott Smith - Richard Thompson - Yoad Yagil - Igor Uman - Elazar Gerland - Shmuel Glasberg - Ami Altman - Amar Dhanantwari - Dhruv Mehta - Tom Morton - Mani Vembar - Galit Kafri - Naama Cohen-Dahan - Andrei Feldam - Mani Vembar


7/21/2014 - amos onlineamos3.aapm.org/abstracts/pdf/90-25729-346462-108671.pdf · 7/21/2014 1 1...

Documents