Neural Network Approach for Photon-counting Detection The First - - PowerPoint PPT Presentation

▶
neural network approach for photon counting detection the
SMART_READER_LITE
LIVE PREVIEW

Neural Network Approach for Photon-counting Detection The First - - PowerPoint PPT Presentation

Dec 16, 2023 290 likes •555 views

Neural Network Approach for Photon-counting Detection The First Step: PPE Correction Ruibin Feng, Ph.D. Ge Wang, Ph.D. David Rundle Biomedical Imaging Center, Biomedical Imaging Center, JairiNovus Technologies Ltd. CBIS/BME, RPI

slide-1
SLIDE 1

Neural Network Approach for Photon-counting Detection – The First Step: PPE Correction

  • Nov. 19, 2017

Ruibin Feng, Ph.D. Biomedical Imaging Center, CBIS/BME, RPI fengr@rpi.edu David Rundle JairiNovus Technologies Ltd. david.rundle@jairinovus.com Ge Wang, Ph.D. Biomedical Imaging Center, CBIS/BME, RPI wangg@rpi.edu

slide-2
SLIDE 2

Outline

  • X-ray Detectors: EIDs vs. PCDs
  • PCD Data Degradation
  • Trigger Threshold Correction
  • Monte-Carlo Simulation
  • Discussions & Conclusion
slide-3
SLIDE 3

Energy-integrating Detectors (EIDs)

  • Mature technology in all current x-ray scanners
  • Energy integration over the entire x-ray spectrum
slide-4
SLIDE 4

Drawbacks of EIDs

  • Energy-dependent information lost

– Linear attenuation not tissue-type sensitive

  • Data quality degenerated due to the dark current

(electric/Swank noise) – Low SNR

  • Low-energy photons under weighted

– Poor contrast, beam-hardening

slide-5
SLIDE 5

Photon-counting Detectors (PCDs)

  • Voltage cross the threshold counted,

individually and energy-sensitively

  • Multiple energy windows spanning the spectral

dimension for CT imaging

slide-6
SLIDE 6

Advantages of PCDs

  • Spectrally unique contrast

– K-edge and fluorescence imaging, beam-hardening avoidance

  • Low radiation dose

– No electronic noise, balanced photon weights, improved SNR

  • High spatial resolution

– Desirable for radiomics

slide-7
SLIDE 7

PCD Data Degradation

  • Pulse Pileup Effect (PPE)
  • Charge sharing
  • K-escape x-rays
  • Compton scattering
slide-8
SLIDE 8

Pulse Pileup Effect (PPE)

Missed Counts

slide-9
SLIDE 9

Pulse Pileup Effect (PPE)

Distorted Energy

slide-10
SLIDE 10

Pulse Pileup Effect (PPE)

  • PCDs degrade in the performance of detection tasks

when the count rate exceeds 20% of the maximum rate

  • Current compensation/calibration methods are not
  • ptimal and difficult to extend for different

applications – Model must be accurate to describe the detection process – Optimization must be specific to address intended tasks such as material decomposition or contrast estimation

slide-11
SLIDE 11

NN-based Trigger Threshold Correction

Distorted Measurement, DLR = 23.21% 50 100 150 200 Energy Bin 100 200 300 400 500 600 700 800 900 1000 Count Distorted Measurement, DLR = 4.07% 50 100 150 200 Energy Bin 20 40 60 80 100 120 140 160 180 200 Count True Measurement, DLR = 4.07% 50 100 150 200 Energy Bin 50 100 150 200 250 Count True Measurement, DLR = 23.21% 50 100 150 200 Energy Bin 200 400 600 800 1000 1200 1400 Count Distorted Measurement, DLR = 56.03% 50 100 150 200 Energy Bin 200 400 600 800 1000 1200 1400 1600 1800 2000 Count True Measurement, DLR = 56.03% 50 100 150 200 Energy Bin 1000 2000 3000 4000 5000 6000 Count

slide-12
SLIDE 12

Trigger Threshold

  • X-ray tube energy: 120 KeV
  • Normal threshold: < 120 KeV
  • Tigger threshold: > 120 KeV

Signal strength over the trigger threshold indicates whether PPE occurs and how severe it is

slide-13
SLIDE 13

NN-based Correction for PPE

Normal Thresholds Trigger Thresholds

Distorted Measurement Corrected Measurement NN

slide-14
SLIDE 14

50 100 150

Energy

0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045

Probablity Normalized Transmitted Spectrum

Monte-Carlo Simulation

  • X-ray spectrum

– TASMICS

  • 43 Combinations of

Attenuators

– Water, Bone, Blood w. 20% Gd – Thickness T = {20, 30} cm – Bone: T(bone) = {0, 1, 3, 5} cm – 20% Gd: T(Gd) = [0:4:20] cm – T(water) = T - T(bone) - T(Gd)

slide-15
SLIDE 15

1 2 3 4 5 6 7 8 9

Time ( s)

0.2 0.4 0.6 0.8 1

Energy (keV) Normailzed Pulse Shape (deadtime = 1 s)

Unipolar Pulse Bipolar Pulse

  • Pulse Shaper

Unipolar Pulse Bipolar Pulse

  • Detector Type

Paralyzable Nonparalyzable

Monte-Carlo Simulation

slide-16
SLIDE 16

Monte-Carlo Simulation

Training and Testing Datasets

  • 1,000 measurements for each attenuator
  • Dataset 1:

– 36 attenuators – Training, validation, testing = 60%, 20%, 20%

  • Dataset 2:

– 7 attenuators

slide-17
SLIDE 17
  • Deadtime Loss Ratio (DLR)

Paralyzable detector: Nonparalyzable detector:

  • Coefficient of Variation (COV)

Monte-Carlo Simulation

slide-18
SLIDE 18
  • Neural Network Model

Fully-connected NN with 1 hidden layer 512 hidden units Dropout and L2 regularizer

  • Unbiased Estimator

Monte-Carlo Simulation

slide-19
SLIDE 19

100 101 DLR (%) 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 COV Comparison with Detector Measurement Proposed Corrector: Dataset 1 Proposed Corrector: Dataset 2 Detector Measurement: Dataset 1 Detector Measurement: Dataset 2 100 101 DLR (%) 0.01 0.02 0.03 0.04 0.05 COV Comparison with Unbiased Estimator Proposed Corrector: Dataset 1 Proposed Corrector: Dataset 2 Unbiased Estimator: Dataset 1 Unbiased Estimator: Dataset 2

Numerical Results

  • Unipolar Pulse & Paralyzable Detector
slide-20
SLIDE 20

100 101 DLR (%) 0.05 0.1 0.15 0.2 0.25 0.3 0.35 COV Comparison with Detector Measurement Proposed Corrector: Dataset 1 Proposed Corrector: Dataset 2 Detector Measurement: Dataset 1 Detector Measurement: Dataset 2 100 101 DLR (%) 0.01 0.02 0.03 0.04 0.05 COV Comparison with Unbiased Estimator Proposed Corrector: Dataset 1 Proposed Corrector: Dataset 2 Unbiased Estimator: Dataset 1 Unbiased Estimator: Dataset 2

Numerical Results

  • Bipolar Pulse & Paralyzable Detector
slide-21
SLIDE 21

100 101 DLR (%) 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 COV Comparison with Detector Measurement Proposed Corrector: Dataset 1 Proposed Corrector: Dataset 2 Detector Measurement: Dataset 1 Detector Measurement: Dataset 2 100 101 DLR (%) 0.01 0.02 0.03 0.04 0.05 COV Comparison with Unbiased Estimator Proposed Corrector: Dataset 1 Proposed Corrector: Dataset 2 Unbiased Estimator: Dataset 1 Unbiased Estimator: Dataset 2

Numerical Results

  • Unipolar Pulse & Nonparalyzable Detector
slide-22
SLIDE 22

100 101 DLR (%) 0.05 0.1 0.15 0.2 0.25 0.3 0.35 COV Comparison with Detector Measurement Proposed Corrector: Dataset 1 Proposed Corrector: Dataset 2 Detector Measurement: Dataset 1 Detector Measurement: Dataset 2 100 101 DLR (%) 0.01 0.02 0.03 0.04 0.05 COV Comparison with Unbiased Estimator Proposed Corrector: Dataset 1 Proposed Corrector: Dataset 2 Unbiased Estimator: Dataset 1 Unbiased Estimator: Dataset 2

Numerical Results

  • Bipolar Pulse & Nonparalyzable Detector
slide-23
SLIDE 23

Future Plan for PPE Correction

  • Systematic Simulation Study
  • Phantom Experiments
  • Preclinical Testing

How to Collect Unbiased Data? – Perform realistic simulation with professional software tools – Reduce the incident flux for PPE-free data via time integration

slide-24
SLIDE 24

Future Plan for CS Correction

Charge Sharing: one photon is detected by multiple pixels with lower energies

slide-25
SLIDE 25

Conclusion

We have proposed an NN/ML approach to handle PPE and other artifacts in PCD data

  • Extract an optimal relationship between PCD data

before and after degradation of any kind

  • Potentially, the NN/ML approach can outperform the

existing patented methods for PCD data correction, and improve photon-counting CT image reconstruction

slide-26
SLIDE 26

Reference

  • 1. Robert K Swank. Absorption and noise in x-ray phosphors. Journal of Applied Physics, 44(9):4199–4203, 1973.
  • 2. Ju

̈rgen Giersch, Daniel Niederlo ̈hner, and Gisela Anton. The influence of energy weighting on x-ray imaging quality. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 531(1):68–74, 2004.

  • 3. Katsuyuki Taguchi, Eric C Frey, Xiaolan Wang, Jan S Iwanczyk, and William C Barber. An analytical model of the effects of pulse pileup on

the energy spectrum recorded by energy resolved photon counting x-ray detectors. Medical physics, 37(8):3957–3969, 2010.

  • 4. Adam S Wang, Daniel Harrison, Vladimir Lobastov, and J Eric Tkaczyk. Pulse pileup statistics for energy discriminating photon counting x-ray
  • detectors. Medical physics, 38(7):4265–4275, 2011.
  • 5. Katsuyuki Taguchi and Jan S Iwanczyk. Vision 20/20: Single photon counting x-ray detectors in medical imaging. Medical physics, 40(10),

2013.

  • 6. M Zhang, ECFrey, JXu, and KTaguchi. Sinogram domain material decomposition using penalized likelihood method in photon counting x-ray

detector (pcxd) with pulse pileup correction. In Proceedings of the IEEE Nuclear Science Symposium and Medical Imaging Conference, Dresden, Germany, pages M06–409, 2008.

  • 7. S Kappler, S Ho

̈lzer, E Kraft, K Stierstorfer, and T Flohr. Quantum-counting ct in the regime of count-rate paralysis: introduction of the pile- up trigger method. In SPIE Medical Imaging, pages 79610T–79610T. International Society for Optics and Photonics, 2011.

  • 8. Jochen Cammin, Steffen Kappler, Thomas Weidinger, and Katsuyuki Taguchi. Photon-counting ct: modeling and compensating of spectral

distortion effects. In SPIE Medical Imaging, pages 941250–941250. International Society for Optics and Photonics, 2015.

  • 9. Jochen Cammin, Steffen Kappler, Thomas Weidinger, and Katsuyuki Taguchi. Evaluation of models of spectral distortions in photon-counting

detectors for computed tomography. Journal of Medical Imaging, 3(2):023503–023503, 2016. 10.J Punnoose, J Xu, A Sisniega, W Zbijewski, and JH Siewerdsen. spektr 3.0a computational tool for x-ray spectrum modeling and analysis. Medical physics, 43(8):4711–4717, 2016. 11.Katsuyuki Taguchi, Mengxi Zhang, Eric C Frey, Xiaolan Wang, Jan S Iwanczyk, Einar Nygard, Neal E Hartsough, Benjamin MW Tsui, and William C Barber. Modeling the performance of a photon counting x-ray detector for ct: Energy response and pulse pileup effects. Medical physics, 38(2):1089–1102, 2011. 12.Hsieh SS, Pelc NJ. Improving pulse detection in multibin photon-counting detectors. Journal of Medical Imaging, 3(2): 023505-023505, 2016.