Download - Microsoft PowerPoint - UBC Talk Sept 30

• Introduction to Medical Imaging– X-ray Computed Tomography (CT)– Nuclear Medicine

• Single Photon Emission Computer Tomography (SPECT)• Positron Emission Tomography (PET)

• VLSI Circuits for CT and SPECT

• Merging CT and SPECT hardware into one imaging technology.

• Highlights from CMOS Emerging Technologies workshop (www.cmoset.com) recently held in Vancouver

Kris Iniewski, Redlen Technologies

Circuits for CT Scanners and SPECT Gamma Cameras

About the Instructor

• Krzysztof (Kris) Iniewski is managing R&D at Redlen Technologies Inc., a start-up company in British Columbia. His research interests are in VLSI circuits for medical imaging.

• He is an editor of “VLSI Circuits for the NanoScale: Communications, Imaging and Sensing”, “Wireless Technologies: Circuits, Systems and Devices”, “Medical Imaging Electronics” and co-author of “Network Infrastructure and Architecture”.

• Dr. Iniewski has held management and research positions at the Universiy of Alberta (2004-2006), PMC-Sierra (1995-2003) and the University of Toronto (1988-1994). He has published over 100 research papers and holds 18 international patents.

• Kris is a Technical Chair for CMOS Emerging Technologies workshop (www.cmoset.com). He can be reached at [email protected]

Motivation for Medical Imaging

http://www.ecse.rpi.edu/censsis/

Healthcare Trends Drive Imaging Growth


Ultrasound Imaging

Visible InfraredMilli-

metre

Micro-

wave

and RF

THz gap

10 15Hz 10 14Hz 10 13Hz 10 12Hz 10 11Hz 10 10Hz

Ultra-

violetX Ray

10 16Hz10 17Hz

MRI

Nuclear medicine

10 18Hz10 19Hz

X Ray Imaging

100keV 10keV

TerahertzImaging

Frequency

Photonics ElectronicsOptical Imaging

MRI Optical Molecular Imaging Modalities

PET Imaging

µµµµ

+-

Variety of Techniques Available


CT vs. Nuclear Medicine (PET/SPECT)

• X-ray• Source/detector geometry• 3D computed tomography (CT)

• Nuclear Medicine (SPECT/PET)• Source of signal from within body

• 2D and 3D imaging

Source

Detector

Source

Detector

FLUOROSCOPY

Electron

Image Intensifier

TV PickupTube

XX--RayRay ANALOG IMAGE

LightTubeTube

MAMMOGRAPHY & RADIOGRAPHY PhosphorPhosphor Film

LightLightXX--RayRay

ANALOG IMAGE

TubeTube

XX--RayRay

XX--RayRay

Detector

TubeTube

XX--RayRayXX--RayRay

DIGITAL DETECTOR - FUTURE

DIGITAL IMAGE

CT/PET/SPECT Trends

X-Ray & Computed Tomography (CT)

Source: http://www.iwr.uni-heidelberg.de/groups/ngg/Tutorial/TutCT_121203_Lauritsch.pdf

CT Scanner - Principle

From Kris Iniewski, “X-ray and Computed Tomography Imaging Principles”, in Medical Imaging Electronics, K. Iniewski (Ed.), Wiley 2009.

CT Scanner - Reality

X-Ray Imaging to CT Imaging

• Standard X-ray’s limitations– 3D structures are collapsed

into 2D images– Low soft-tissue contrast,

great for bones– Not very quantitative

• X-ray CT– Take a large number of x-rays

at multiple angles– Calculate the 3D image

• Similar hardware to ordinary x-ray

• Image of a slice - extendable to 3D

• But, heavy computational load


SPECT Gamma Cameras

SPECT Diagnostic Example

• Left: SPECT scans of the brain of a three year old male near drowning patient shown shortly after the accident s howing decreased brain activity. The patient presented in a persistent vegetative state, and was pronounced blind with sev ere spasticity.

• Right: SPECT scans of the same child taken 9 months later demonstrating increased brain activity and blood fl ow following 120 hyperbaric oxygen treatments. The child was now alert, responsive, laughing, eating and drinking normally, walking, speaking bi-lingually, and had regained normal visi on.

SPECT vs. CT

• Unlike X-ray CT, SPECT produces 3-D images that relate an organ’s function.- better relay of extent of disease - reveals the course of the disease earlier.

• Simple process with immediate results. Less expensive than MRI or PET. Covered by insurance when brain injury is present.

• Unlike X-ray, there is an injection.

• Image quality can be decreased by patient movement (but new CZT based SPECT equipment has dramatically reduced measurement time).

Safety and Biohazards of CT/SPECT

• X-ray/Computer Tomography CT• Ionizing radiation (might be inducing cancer)

• Morbidity associated with contrast agents

• Nuclear Medicine (SPECT/PET)• Ionizing radiation (but very low dose)• Patient injection required

Radiation Detection Principle

Radiation Detector Front End

http://www-physics.lbl.gov/~spieler/Heidelberg_Notes/

Charge Sensitive Amplifier (CSA)

CSA Calibration

Analog Signal Processing Chain


Data Readout

Noise Spectrum


Noise Equivalent Circuit


X-Ray Detector Readout System

Pawel Grybos, “Detector Interface Circuits for X-ray Imaging”, in K. Iniewski (Ed.)Circuits for NanoScale – Communications, Imaging and Sensing, CRC Press 2008

Feedback Configurations


Leakage Current Compensation


DEDIX (Dual Energy Digital Imaging of X-ray)


DEDIX – Single Channel


Example: 8 keV photons from X-ray Tube


Medipix 2 (256x256) – Pixel Readout

X. Llopart, M. Campbell, R. Dinapoli, D. San Segundo, E. Pernigotti, "Medipix2: a 64-k Pixel Readout Chip With 55-µm Square Elements Working in Single Photon Counting Mode,“IEEE Trans. Nucl. Sci., vol. 49, no. 5, 2002, pp. 2279 - 2283.

Medipix2 (256 x 256) – Chip Floorplan

X. Llopart, M. Campbell, R. Dinapoli, D. San Segundo, E. Pernigotti, "Medipix2: a 64-k Pixel Readout Chip With 55-µm Square Elements Working in Single Photon Counting Mode,“IEEE Trans. Nucl. Sci., vol. 49, no. 5, 2002, pp. 2279 - 2283.

Medipix2 Cell Layout

PILATUS Pixel Cell

Brönnimann et al.:"The Pilatus 1M Detector," J. Synchrotron Rad., 13, 2006, 120-130.

Common Circuit Requirements

• Signal amplification (LNA in Ultrasound and MRI, CSA in Nuclear Medicine and X-ray). Fighting noise sources and crosstalk is frequently the main battle in practical systems.

• Signal filtering (signal shaping in Nuclear Medicine). Signal multiplexing (have to deal with hundreds or thousands channels)

• ADC conversion (medical imaging operates at very low input SNR, analog signal processing is a must)

• Power dissipation is typically #1 challenge (difficulties in extracting heat).

• Signal processing of data close to a sensor beneficial, otherwise have to deal with Gb/s of data using a few Watts of power budge (if that).

• Sensor are very specialized and are much more important (and expensive!) than CMOS circuits

Power vs. ENC Trade-off

From Gianluigi De Geronimo, “Low-Noise Electronics for Radiation Sensors”, in MedicalImaging Electronics, K. Iniewski (Ed.), Wiley, 2008.

VA32 Chip (U of Michigan)

Practical Implementation Challenges

• Need to monitor temperature (on chip temperature sensors).

• Must calibrate sensor responses and non-linearities.

• Must deal with noise sources and digital cross-talk. Very difficult to de-bug at the system level.

• Have to deal with channel to channel non-uniformities.

• Would like to dissipate less than 50µW/channel.

• Must implement hundreds of channels per chip, thousands of channels would be even better.

• Would like to be able to self-test the circuit without the sensor stimuli (BIST).

MU

X

CSA Shaping PDChannel #1

CMLI/O

TempSensor

SPI



CtrlLogic

Bias

Practical CMOS Implementations

• FPGA Interface (SPI)

• Temperature Sensing

• Channel to Channel Uniformity

• On-chip Calibration

• Built-In SelftTesting (BIST)

• Low noise switching (CML/LVDS)

Pacific-128 datasheet, www.redlen.com

Pacific-128 Chip (Redlen)

X-Ray and CT Hardware Trends

• The fast front-end electronics for a large array of X-ray sensors should:– amplify and filter small signals from the each sensor element– perform analog to digital conversion– store the data on the integrated circuit in each channel

independently at the same time

• Complexity of the multi-channel mixed-mode integrated circuit implementation lie in the following areas:– power limitation– low level of noise– good matching performance and crosstalk effects

• Current challenges lie in detecting very low Xraydoses (there have been medical reports that CT scans are causing cancer related cell damage at the current doses used!)

CT/SPECT Fusion

• Coupling SPECT with today’s high-powered CT scanners is going to propel the technology into a number of new research and clinical arenas—from in vivo small animal studies to CT angiography in the emergency department.

• New tracers already under testing specifically target cancers of the brain, thyroid, prostate, breast, lung, ovaries, kidneys, and liver, as well as heart and bone diseases and defects.

• With the advent of fusion imaging, nuclear medicine’s potential to diagnose and treat disease will advanced greatly offering numerous opportunities in clinical practice.

CT/SPECT Fusion

• SPECT/CT acquires both scans with the patient in the same position. Specialized registration software then reconstructs the data sets, adjusts for differences in format and scanner geometry, and fuses them into a single image.

• Grafting the high spatial resolution capabilities of today’s high-speed CT scanners with SPECT’shighly accurate definition of disease processes vastly enhances anatomical mapping and localization, moving the new hybrid directly into a wider range of clinical applications.

• Most significantly, CT attenuation correction greatly reduces the problems of distortion and degradation that typically occur with radionuclide-based methods.

CT/SPECT Fusion

• The existing SPECT/CT systems are made with two separated apparatus joined together axially and coaxially.

• Current research aims to enable a clinical system where both apparatus will use the same data acquisition system which is critical to achieve a perfect fusion of anatomical and metabolical images.

Integration (CT) and Counting (SPECT) Electronics

From Edgar Kraft, Ivan Peric, Circuits for Digital X-ray Imaging: Counting and Integration”, in Medical Imaging Electronics, K. Iniewski (Ed.), Wiley 2009.

From Edgar Kraft, Ivan Peric, Circuits for Digital X-ray Imaging: Counting and Integration”, in Medical Imaging Electronics, K. Iniewski (Ed.), Wiley 2009.

Integration (CT) and Counting (SPECT): Measurements

CMOS Emerging Technologies workshop(www.cmoset.com)

• Held in Vancouver, Aug 5-7, 2008. Presentation slides available on-line.

• Related talks:

– Ralph Etienne-Cummings, John Hopkins U, Current Mode Active Pixels Sensors Make Focal-Plane Image Processing

– Jan Thim, Mid Sweden University, CMOS for Color X-Rays– Edoardo Charbon, EPFL, Single-Photon Imaging: the Next

Big Challenges– Karim Karim, U of Waterloo, CMOS Photon Counting Pixel

for Real-time Imaging of Palladium Seeds in Permanent Breast Seed Implantation

• Follow up meetings in Banff (Feb 18-20, 2009) and Vancouver (Sept 23-25, 2009). Speakers, volunteers and session chairs needed. Talk to me if interested, [email protected]