CMOS Active Pixel Sensors for Medical Imaging Applications

Brief history of CMOS devices

Overview of the MI-3 consortium

- The MI-3 Sensors

Publications

Useful links

---

The demand for imaging devices with intelligent capabilities has led to an upsurge in demand for CMOS Active Pixel Sensors (APS) in scientific applications. The advantage of CMOS imagers over, for example, CCDs, is that they can be designed with application-specific functions built in, to optimize performance.

Brief history of CMOS devices

The CMOS image sensor was first conceived in 1968 as a silicon array with an active ‘buffer’ transistor per pixel. Fabricating image sensors in Complementary Metal Oxide Semiconductor (CMOS) technology at standard integrated circuit (IC) lines enables the addition of analogue and digital electronics on-chip. Even the simplest CMOS design can perform signal amplification and charge detection in-pixel (Figure 1). This enables direct signal buffering and random access readout of the array. Further electronics can be added for simple functions such as timing and control, or in the more advanced case, to provide a fully digital interface with in-pixel analogue to digital conversion (ADC).

Unfortunately, all this intelligence comes at a price. Including extra capabilities on-chip and in-pixel introduces added complications in terms of noise. Differences in transistor performance produces fixed pattern noise both at the pixel level and per-column. This fixed pattern noise, coupled with a larger pixel size relative to the CCD meant CMOS APS were initially rejected as scientific imaging devices. However, the limitations of the CCD have become more apparent as the demand for high speed, large area, low power and low cost imaging has increased since the 1990s. The performance of CMOS APS has been steadily improving over this period and they are now gaining popularity as they can provide solutions for both scientific applications and industrial imaging. Also, an exciting prospect for the future is the ability to fabricate large scale imagers without loss of performance whose size is limited only by the CMOS wafer scale (20”).

Figure 1: Standard '3-T' pixel layout. Reset transistor 'R' clears the pixel of integrated charge, Source Follow transistor 'SF' amplifies/buffers the signal and Row Select transistor 'RS' selects the row for readout.

 

---

Overview of M-I 3 consortium

UCL is a member of the RC-UK funded Multidimensional Integrated Intelligent Imaging project (M-I 3). The project consists of eleven research centres throughout the U.K. and was founded to advance the capabilities of CMOS Active Pixel Sensors for a variety of scientific applications. The aims of the consortium are to extend the effective spectral response of these devices and to develop on-chip intelligence down to the pixel level to meet the future demands of the scientific community.

In the Radiation Physics group at UCL, our goal is to demonstrate that CMOS active pixel sensors can be employed in a range of medical imaging applications. We carefully characterize each novel CMOS imager that is produced by the M-I 3 consortium to ensure that it is suitable for medical imaging [1]. Performance is evaluated in terms of x-ray linearity, modulation transfer function (MTF), noise power spectrum (NPS) and detective quantum efficiency (DQE). As can be seen from the other pages on this site, a key area of interest for our group is advancing the sensitivity and specificity of mammography. This extends to our work with CMOS APS[2].

One application aims to exploit the intelligence of CMOS APS in x-ray diffraction studies of breast tissue[3]. Angle dispersive x-ray diffraction peaks appear as concentric circles about the transmitted x-ray beam. Coherently scattered photons have interacted with matter so provide material specific information about an object under investigation. The positions of diffraction peaks corresponding to healthy and cancerous tissue could be quickly identified using novel readout features and real time signal processing, only afforded by CMOS APS. A large area CMOS APS would allow the entire diffraction pattern to be visualised simultaneously without the need for an x-ray image intensifier, improving the accuracy of the result. This will be available for testing from March 2008.

Other applications being pursued in the group include Phase Contrast Imaging [4] and perfusion mammography.

---

The MI-3 sensors and example images

 

 

---

Publications

Journal Papers
[1] 'A characterization study of two novel Active Pixel Sensors', S.E. Bohndiek, C.D. Arvanitis, G.J. Royle, A.T. Clark, J.P. Crooks, M.L. Prydderch, R. Turchetta, A. Blue, V. O'Shea, R.D. Speller, Optical Engineering, 46(12), 124003, 2007.

[2] 'Empirical electro-optical and x-ray performance evaluation of CMOS Active Pixel Sensors for low dose, high resolution x-ray medical imaging', C.D. Arvanitis, S.E. Bohndiek, G.J. Royle, A. Blue, H.X. Liang, A. Clark, M. Prydderch, R. Turchetta, R. D. Speller,Medical Physics, 34(12) 4612-25, 2007.

[3] 'A CMOS Active Pixel Sensor system for laboratory-based x-ray diffraction of biological tissue', S.E. Bohndiek, E.J. Cook, C.D. Arvanitis, A. Olivo, G.J. Royle, A.T. Clark, M.L. Prydderch, R. Turchetta, R.D. Speller, Physics in Medicine & Biology, 53, 655-72, 2008.

[4] 'First evidence of phase contrast imagign with laboratory sources and active pixel sensors', A. Olivo, C.D. Arvanitis, S.E. Bohndiek, A. Clarck, M. Prydderch, R. Turchetta, R.D. Speller, Nuclear Instruments and Methods in Physics Research, A581(3), 776-82, 2007.

---

Links

MI-3 Consortium homepage

General introduction to CMOS Image Sensors

Comparison of electronic imagers to film