Using DRLs for CT optimization: Sources and management of dose - - PDF document

▶
using drls for ct optimization sources and management of
SMART_READER_LITE
LIVE PREVIEW

Using DRLs for CT optimization: Sources and management of dose - - PDF document

Sep 05, 2022 313 likes •742 views

11/21/19 Using DRLs for CT optimization: Sources and management of dose variability Ehsan Samei (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission only. 1 11/21/19 Overarching premise Medicine: Discerning and

slide-1
SLIDE 1

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

1

Ehsan Samei

Using DRLs for CT optimization: Sources and management of dose variability

slide-2
SLIDE 2

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

2

Overarching premise

Medicine: Discerning and intervening in the health state of the patient with sufficient accuracy, precision, and safety for definitive clinical outcome Healthcare is about the patient, not the particularities

  • f the techniques – techniques and quantities are

valued to the extent they are relevant to the patient

slide-3
SLIDE 3

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

3

Reality check 1:

Clinical practice

Heterogeneous and Complex:

  • Varying technologies
  • Varying technical parameters
  • Varying patients
  • Varying human operators
  • Competing interests

Variability in the quality of care

Variability across practice

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Total DLP of study (Gy*cm) Exam Date

Abdomen-Pelvis CT

slide-4
SLIDE 4

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

4

GE HD750

5 10 15 20 25 30 35 20 25 30 35 40 45

noise (HU) patient diameter (cm)

5 10 15 20 25 30 35 40 45 50 20 25 30 35 40 45

patient diameter (cm)

Noise versus patient size: Phantom

Siemens Flash

Ria et al, AAPM 2018

GE HD750

5 10 15 20 25 30 35 20 25 30 35 40 45

noise (HU) patient diameter (cm)

5 10 15 20 25 30 35 40 45 50 20 25 30 35 40 45

patient diameter (cm)

Noise versus patient size: Patients

Siemens Flash

slide-5
SLIDE 5

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

5

Reality check 2:

There is a cost!

  • Most people will experience at least one diagnostic

error in their lifetime

– 10% of patient deaths – 6-17% hospital adverse events – Leading type of paid medical malpractice – Claims twice as likely to result in death – Highest proportion of total payments.

Improving Diagnosis in Healthcare, NAM 2015

Drive towards high-quality, consistent, patient-centric, evidence-based, precise, and safe healthcare

slide-6
SLIDE 6

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

6

Drive towards high-quality, consistent, patient-centric, evidence-based, precise, and safe healthcare

Patient-oriented

  • ptimization of

quality and safety

Quality and safety optimization

A process to enable and ensure

high-quality (accurate), consistent (precise), patient-centric, and safe,

use of imaging in medicine

slide-7
SLIDE 7

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

7

Optimization framework

Safety indices (organ dose,

  • eff. dose, …)

Quality indices (d’, Az, …) Scan factors (mAs, kVp, pitch, recon, kernel, …) B e n e f i t r e g i m e R i s k r e g i m e

Different patients and indications

Optimization regime

Optimization taxonomy

Benefit Risk Physical Quality indices (eg, MTF) Output indices (eg, CTDI) Clinical Performance indices (eg, d’, AUC) Safety indices (eg, ED, RI)

Independent variables (imaging parameters) Dependent variables

slide-8
SLIDE 8

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

8

Quality vs dose: the right balance?

Radiation Dose

?

Image Quality

Metrology matters!

Diagnosis Preference Radiation Dose Image Quality

? ?

Metrology matters!

Diagnosis Preference Radiation Dose Image Quality

? ?

slide-9
SLIDE 9

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

9

What are the right metrics?

1. Relevant: As much as possible, patient-/indication-centric (not modality or machine) 2. Robust: To ensure reliability and applicability (quantitative not subjective) 3. Smart: Maintained balance between robustness and relevance 4. Relatability: Surrogates relatable to clinical task/safety 5. Practical: Economic to measure

Radiation Dose Image Quality

Variability from systematic (age, size) and

  • perational effects
slide-10
SLIDE 10

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

10

DRL as variability management

Radiation Dose Image Quality Range w/ DRL

DRL

Image Quality Range w/o DRL

Reference range as variability management

Radiation Dose Image Quality Range w/ RR

RR

Image Quality Range w/o DRL

slide-11
SLIDE 11

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

11

Precision by inference

Technology assessment

Precision by prescription

Prospective use definition

Precision by outcome

Retrospective quality audit

Precision by inference

Technology assessment

Apply constraints to machine performance in concordance with self, peer, or mandates

slide-12
SLIDE 12

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

12

Characterizing relevant intrinsic performance of CT system, TG233

3 5 5 m m 2 2 m m 2 9 m m 1 7 7 m m 1 1 2 m m

60 mm 85 mm 85 mm 60 mm 90 mm

  • Mercury Phantom 4.0
  • Size matching population cohorts

Automated Characterization

  • HU, Contrast, Noise, CNR, MTF, NPS
  • TCM-dependency
  • Detectability index
  • Detectability indices for reference tasks

– 1, 5, 10 mm, 10 and 100 HU, designer, rect, Gaussian

dNPWE

'

( )

2

= MTF 2(u,v ) WTask

2 (u,v )

∫∫

E 2(u,v )dudv " # $ %

2

MTF 2(u,v ) WTask

2 (u,v )

∫∫

NPS(u,v )E 4(u,v ) + MTF 2(u,v ) WTask

2 (u,v )Ni dudv

slide-13
SLIDE 13

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

13

Automated Characterization

6 7 8 9 10 11 12 05/15/2013 02/23/2014 05/15/2014 07/24/2014 10/18/2014 01/08/2015 03/26/2015 06/10/2015 08/27/2015 11/24/2015 02/11/2016 04/14/2016 07/19/2016 09/20/2016

Detectability Date

QC Detectability - ACR Phantom

SIEME NS - Force - DMPCT 3 SIEME NS - Flash - ERCT1 SIEME NS - Flash - DMPCT3 SIEME NS - Flash - CCCT5 SIEME NS - Flash - CCCT4 SIEME NS - Flash - C3 SIEME NS - Flash - C1 GE - LSXtra - J1 GE - LSVCT - ERCT2 GE - LSVCT - CCCT3 GE - LSVCT - CaryCT GE - LS16 - SP GE - DIQ - PETCT1 GE - DCT75 0 HD - DMPCT 2 GE - DCT75 0 HD - DMPCT 1 GE - DCT75 0 HD - CCCT2 GE - DCT75 0 HD - CCCT1 GE - DCT75 0 HD - B5 GE - D690 - PETCT2 GE - BrightSpeed - DMP67 0 GE - BrightSpeed - D670

Detectability Indices Across Systems

Intersystem Variability 8.0% Intrasystem Variability SIEMENS - Force - DMPCT3 0.1% SIEMENS - Flash - ERCT1 1.0% SIEMENS - Flash - DMPCT3 3.5% SIEMENS - Flash - CCCT5 1.9% SIEMENS - Flash - CCCT4 4.2% SIEMENS - Flash - C3 2.1% SIEMENS - Flash - C1 2.4% GE - LSXtra - J1 2.4% GE - LSVCT - ERCT2 3.1% GE - LSVCT - CCCT3 2.0% GE - LSVCT - CaryCT 1.7% GE - LS16 - SP 2.1% GE - DIQ - PETCT1 3.2% GE - DCT750 HD - DMPCT2 2.8% GE - DCT750 HD - DMPCT1 1.4% GE - DCT750 HD - CCCT2 1.8% GE - DCT750 HD - CCCT1 2.2% GE - DCT750 HD - B5 3.0% GE - D690 - PETCT2 3.0% GE - BrightSpeed - DMP670 2.4% GE - BrightSpeed - D670 3.8%

Intra-system variability: 1-4% Inter-system variability: 8%

slide-14
SLIDE 14

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

14

Precision by prescription

Prospective use definition

Apply constraints to protocol definitions in concordance with self, peer, or mandates

Detectability Indices Across the US

ACR-RSNA-Duke Collaborative project

Zhang et al, RSNA, 2018

slide-15
SLIDE 15

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

15

Matching protocols across imaging systems

?

GE Siemens Siemens Siemens

Sharpness

slide-16
SLIDE 16

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

16

GE to Siemens “best match”

STD ASiR 40% i40 SAFIRE 3

GE to Siemens “best match”

BONE PLUS ASiR 0% b70 SAFIRE 0

slide-17
SLIDE 17

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

17

Average resolution (f50) Average noise texture (favg)

Log of dose needed for a target noise of 10

Matching Image Quality Across CTs

Winslow et al, Med Phys 2018

Protocol MatcherTM

slide-18
SLIDE 18

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

18

slide-19
SLIDE 19

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

19

Quality vs dose: the right balance?

Radiation Dose Image Quality

Iso-dose?

Radiation Dose

Inconsistency in image quality

Image Quality

slide-20
SLIDE 20

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

20

Iso-quality?

Radiation Dose Image Quality

Wide variation in dose

Iso-balance

Radiation Dose Image Quality

Smart balance between quality and dose

slide-21
SLIDE 21

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

21

Optimization of dose for size

for-your-child.blogspot.com http://www.pedrad.org

Iso-balance in practice

ALARA?

Age

Samei et al, JMI, 2017 Samei et al, RPD, 2018

slide-22
SLIDE 22

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

22

Duke Pediatric Protocols

…

Li, et al, RSNA 2010 0.0 0.2 0.4 0.6 0.8 1.0 10 20

Risk Dose

Radiation risk index

slide-23
SLIDE 23

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

23

0.0 0.2 0.4 0.6 0.8 1.0 10 20

Risk Dose

Radiation risk index Clinical risk index

0.0 0.2 0.4 0.6 0.8 1.0 10 20

Risk Dose

Radiation risk index Clinical risk index Total risk index

slide-24
SLIDE 24

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

24

0.0 0.2 0.4 0.6 0.8 1.0 10 20

Risk Dose

Radiation risk index Clinical risk index Total risk index Samei et al, RPD, 2018

Risk metrics

48

RADIATION RISK

BEIR VII

Risk Index

NCI - SEER

5-year relative survival rate for all cancers Organ doses CLINICAL RISK

AUC curve

% correct answers

NCI - SEER

5-year relative survival rate for liver cancer (from localized to regional stage)

Detectability index (d’)

MC (XCAT) Virtual lesions

TOTAL RISK INDEX

slide-25
SLIDE 25

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

25

Risk curves

49

  • E. Samei, et al. Journal of Radiological Protection. 2018

Risk curves (FBP – IR)

50

slide-26
SLIDE 26

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

26

0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 1.0 80 kVp 100 kVp 120 kVp 140 kVp 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 1.0 80 kVp 100 kVp 120 kVp 140 kVp 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.0 0.2 0.4 0.6 0.8 1.0 80 kVp 100 kVp 120 kVp 140 kVp 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.0 0.2 0.4 0.6 0.8 1.0 80 kVp 100 kVp 120 kVp 140 kVp

Spatial frequency (/mm) Spatial frequency (/mm) Spatial frequency (/mm) Spatial frequency (/mm)

120 kVp 120 kVp 120 kVp 120 kVp

Task function (AU) Task function (AU) (a) (b) (c) (d)

10 mm 10 mm 10 mm 10 mm

Small feature no iodine Large feature no iodine Small feature with iodine Large feature with iodine

Task functions

1 2 3 4 5 0.5 0.6 0.7 0.8 0.9 1 Eff dose (mSv) AZ small lesion no contrast 1 2 3 4 5 0.5 0.6 0.7 0.8 0.9 1 Eff dose (mSv) AZ Small lesion w/contrast 1 2 3 4 5 0.5 0.6 0.7 0.8 0.9 1 Eff dose (mSv) AZ Large lesion no contrast 1 2 3 4 5 0.5 0.6 0.7 0.8 0.9 1 Eff dose (mSv) AZ Large lesion w/contrast

FBP 80 kVp FBP 100 kVp FBP 120 kVp FBP 140 kVp IRIS 80 kVp IRIS 100 kVp IRIS 120 kVp IRIS 140 kVp

slide-27
SLIDE 27

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

27

1 2 3 4 5 0.5 0.6 0.7 0.8 0.9 1 Eff dose (mSv) AZ small lesion no contrast 1 2 3 4 5 0.5 0.6 0.7 0.8 0.9 1 Eff dose (mSv) AZ Small lesion w/contrast 1 2 3 4 5 0.5 0.6 0.7 0.8 0.9 1 Eff dose (mSv) AZ Large lesion no contrast 1 2 3 4 5 0.5 0.6 0.7 0.8 0.9 1 Eff dose (mSv) AZ Large lesion w/contrast

FBP 80 kVp FBP 100 kVp FBP 120 kVp FBP 140 kVp IRIS 80 kVp IRIS 100 kVp IRIS 120 kVp IRIS 140 kVp

1 2 3 4 5 0.5 0.6 0.7 0.8 0.9 1 Eff dose (mSv) AZ small lesion no contrast 1 2 3 4 5 0.5 0.6 0.7 0.8 0.9 1 Eff dose (mSv) AZ Small lesion w/contrast 1 2 3 4 5 0.5 0.6 0.7 0.8 0.9 1 Eff dose (mSv) AZ Large lesion no contrast 1 2 3 4 5 0.5 0.6 0.7 0.8 0.9 1 Eff dose (mSv) AZ Large lesion w/contrast

FBP 80 kVp FBP 100 kVp FBP 120 kVp FBP 140 kVp IRIS 80 kVp IRIS 100 kVp IRIS 120 kVp IRIS 140 kVp

slide-28
SLIDE 28

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

28

Detectability trends with dose/size

slide-29
SLIDE 29

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

29

Contrast-size-dose

  • ptimization

C T D I ( m G y ) I

  • d

i n e C

  • n

c e n t r a t i

  • n

( m g I / m l )

Precision by outcome

Retrospective quality audit

Apply constraints to case performance in concordance with self, peer, or mandates

slide-30
SLIDE 30

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

30

Retrospective audit

CT Dose Index + size + in vivo noise

5 10 15 20 25 30 35 40 45 50 20 30 40 Patient dose (CTDI - mGy) Effective Diameter (cm)

Scanner 1 Scanner 2 Scanner 3

Quality range: 25-75% based on prior data Dose DRL

slide-31
SLIDE 31

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

31

10 20 30 40 50 60 20 25 30 35 40 45 CTDIvol (mGy) Effective Diameter (mGy)

PE Chest Protocol

VCT - Old SOMATOM Definition 10 20 30 40 50 60 20 25 30 35 40 45 CTDIvol (mGy) Effective Diameter (mGy)

PE Chest Protocol

VCT - Old SOMATOM Definition VCT - New

slide-32
SLIDE 32

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

32

5 10 15 20 25 30 35 40 45 50 20 30 40 Image Noise (HU) Effective Diameter (cm) 5 10 15 20 25 30 35 40 45 50 20 30 40 Patient dose (CTDI - mGy) Effective Diameter (cm)

Scanner 1 Scanner 2 Scanner 3

Dose In vivo Noise

5 10 15 20 25 30 35 40 45 50 20 30 40 Image Noise (HU) Effective Diameter (cm) 5 10 15 20 25 30 35 40 45 50 20 30 40 Patient dose (CTDI - mGy) Effective Diameter (cm)

Scanner 1 Scanner 2 Scanner 3

Dose In vivo Noise

slide-33
SLIDE 33

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

33

From DRL to NRL: Defining a Noise Reference

15 20 25 30 35 40 45 50 20 30 40 patient diameter (cm) 5 10 15 20 25 30 35 20 25 30 35 40 patient diameter (cm)

Chest wo contrast Abdomen pelvis w contrast

5 10 15 20 25 5 10 15 20 Noise [HU] (Adjusted to 5 mm ST) Abdomen Pelvis 21-25 cm, N = 56 Model: DISCOVERY CT750 HD Model: SOMATOM DEFINITION FLASH Recon: B31F , 3 mm Recon: B31F , 5 mm Recon: DETAIL AR40, 2.5 mm Recon: DETAIL AR50, 2.5 mm Recon: I31F\2 , 3 mm Recon: I31F\3 , 3 mm Recon: STANDARD AR40, 2.5 mm Recon: STANDARD SS40, 2.5 mm Recon: STANDARD SS50, 2.5 mm 5 10 15 20 25 5 10 15 20 Noise [HU] (Adjusted to 5 mm ST) Abdomen Pelvis 25-29 cm, N = 238 5 10 15 20 25 5 10 15 20 Noise [HU] (Adjusted to 5 mm ST) Abdomen Pelvis 29-33 cm, N = 322 5 10 15 20 25 5 10 15 20 Noise [HU] (Adjusted to 5 mm ST) Abdomen Pelvis 33-37 cm, N = 266 5 10 15 20 25 CTDIvol [mGy] 5 10 15 20 Noise [HU] (Adjusted to 5 mm ST) Abdomen Pelvis 37-41 cm, N = 134 2 4 6 8 10 12 14 16 18 20 5 10 15 20 Noise [HU] (Adjusted to 5 mm ST) Standard Chest 21-25 cm, N = 63 Model: DISCOVERY CT750 HD Model: SOMATOM DEFINITION FLASH Recon: B31F , 3 mm Recon: B31F , 5 mm Recon: STANDARD AR10, 2.5 mm Recon: STANDARD AR10, 5 mm Recon: STANDARD SS30, 5 mm 2 4 6 8 10 12 14 16 18 20 5 10 15 20 Noise [HU] (Adjusted to 5 mm ST) Standard Chest 25-29 cm, N = 270 2 4 6 8 10 12 14 16 18 20 5 10 15 20 Noise [HU] (Adjusted to 5 mm ST) Standard Chest 29-33 cm, N = 621 2 4 6 8 10 12 14 16 18 20 5 10 15 20 Noise [HU] (Adjusted to 5 mm ST) Standard Chest 33-37 cm, N = 631 2 4 6 8 10 12 14 16 18 20 CTDIvol [mGy] 5 10 15 20 Noise [HU] (Adjusted to 5 mm ST) Standard Chest 37-41 cm, N = 192
slide-34
SLIDE 34

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

34

Multi-dimensional precision

Indication-specific safety & quality constraints

Safety Attribute Safety Attribute Patient Attribute Safety Attribute Quality Attribute Quality Attribute Patient Attribute Quality Attribute

Multi-dimensional precision

  • Noise, resolution,

dose across

– 103,547 total scans – 95 facilities – 3 manufacturers – 30 models

  • The largest study of

its kind in breadth and depth

Smith et al, RSNA 2018

Peer concorda nce Ideal

slide-35
SLIDE 35

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

35

Observer preference studies Observer preference studies

Set of images with one varying metric Acceptable range for detecting small hepatic lesions

slide-36
SLIDE 36

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

36

Radiologists preference

Metric Overall rank order agreement Clinically acceptable range of image quality metrics for abdominal CT examinations Mean Std error Median Mean 95% CI Lower bound 95% CI Upper bound Noise (HU) 0.90 0.06 (17.8, 32.6) (17.8, 32.5) (17.8, 17.8) (28.0, 37.1) Liver parenchyma HU 0.98 0.19 (92.1, 131.9) (97.2, 131.8) (82.8, 111.5) (131.6, 132.1) Clarity 1.00 0.22 (0.45, 0.55) (0.47, 0.55) (0.43, 0.50) (0.55, 0.55)

Chest Size Range NRL NRR DoRL DoRR RRL RRR 21.0-24.9 10.79 7.56 3.31 2.06 0.460 0.065 25.0-28.9 9.79 6.83 5.17 2.53 0.471 0.059 29.0-32.9 10.19 5.68 7.75 3.75 0.477 0.079 33.0-36.9 10.50 6.40 11.13 7.29 0.477 0.096 37.0-40.9 10.64 7.56 15.04 12.43 0.466 0.101 Abd-Pelvis Size Range NRL NRR DoRL DoRR RRL RRR 21.0-24.9 6.94 2.10 5.3 1.59 0.464 0.078 25.0-28.9 7.69 2.40 6.72 2.48 0.456 0.087 29.0-32.9 8.19 2.75 8.85 3.64 0.462 0.097 33.0-36.9 8.39 3.34 11.85 5.00 0.463 0.110 37.0-40.9 7.99 3.89 14.69 7.99 0.453 0.121 Smith et al, RSNA 2018

slide-37
SLIDE 37

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

37

Intra-facility variability (Chest)

Resolution (f50) Noise CTDI

Intra-facility variability (Chest)

Resolution (f50) Noise CTDI

slide-38
SLIDE 38

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

38

Intra-facility variability (Chest)

Resolution (f50) Noise CTDI

Intra-facility variability (Chest)

Resolution (f50) Noise CTDI

slide-39
SLIDE 39

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

39

Intra-facility variability (Chest)

Resolution (f50) Noise CTDI

What are the sources of variability?

  • Fit a regression tree

to dataset and compare the effect size of tree nodes

  • Compare rank
  • rdering importance
  • f parameters
Predictor Importance Estimates: z1 noiseair9Aggregate 0.5 1 1.5 Estimates 10-6 Resolution Offset2 Z2_ctdivol z2_manufacturersmodelname LOC z2_effective_diameter z2_min_pitch PixelSize z2_max_slice_thickness ConvolutionKernel Predictors

Recon Kernel Slice Thickness Pixel Size Pitch Effective Diameter Facility Scanner Model Resolution (f50) Predictor Importance: Noise (Abd-Pelvis Scans) CTDIvol AP patient Mis-centering

slide-40
SLIDE 40

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

40

  • 1. Streamlining metrology
  • 2. Sorting out Incidental vs cumulative dose
  • 3. Series based vs study-based analytics and constraints
  • 4. Meaningful synthaxing the data – data fidelity
  • 5. Meaningful analytics via smart AI
  • 6. Retrospective analytics <=> prospective optimization
  • 7. Resourcing the process
slide-41
SLIDE 41

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

41

  • Precision imaging requires patient-derived/relevant

and pragmatic surrogates of Q and S

  • Dose cannot be managed irrespective of image quality
  • Quality and safety together form the basis of risk
  • ptimization for the patient, directly related to the

very purpose of imaging

  • There is a significant amount of intra-facility, and

inter-facility variability

– Inter-facility variability > Intra-facility variability

  • DRL concept should be extended to Q&S analytics /

constraints

  • Q&S variability can be managed by applying

constraints at 3 levels: machine performance, protocol design, and case performance

slide-42
SLIDE 42

11/21/19 (c) Ehsan Samei, 2019. Use for non- personal purposes by prior permission

  • nly.

42

  • Regions of agreement, reference levels, and

reference ranges can be used as a peer-based guidance tool to increase consistency

  • The biggest factors affecting Q&S factors

– Convolution kernel, slice thickness, pixel size, pitch, size, facility, scanner model, patient centering

Thank You!