Patent Grant
7199370
This invention relates to the field of positron emission tomography (PET). More particularly, the present invention relates to signal processing devices and methods used in PET front-end electronics.
Positron emission tomography (PET) is a technique used in clinical medicine and biomedical research to create images that show anatomical structures as well as how certain tissues are performing their physiological functions. Radioactive nuclei are introduced into the body as labels on tracer molecules. These nuclei emit positrons which may collide with electrons in the tissue. Each such collision is an annihilation event which may result in two gamma photons. By detecting the gamma photons and processing the result with image processing tools, an image of the activity in the tissue can be produced to display the physiological functions.
In modern PET systems, a scintillation light pulse generated upon the interaction of a photon resulted from; annihilation with a scintillator is collected by photomultiplier tubes (PMT), or avalanche photodiodes (APD), and converted into a charge pulse. Hamamatsu Photonics K.K. Electron Tube Center, Fundamental and Applications of Photomultiplier Tube, J P: Hamamatsu Photonics K.K, 1995, the contents of which are hereby incorporated by reference, provides further details on the PMT. The charge pulse is often amplified and filtered to form a new voltage pulse that has a peak amplitude proportional to the area under the original scintillation light pulse, and hence proportional to the amount of photon energy that is deposited in the scintillator during the interaction. The peak amplitude is then sampled and converted into digital data by use of analog-to-digital converters (ADCs) for subsequent processing. An event time is typically obtained by using constant fraction discriminators (CFDs).
In order to achieve high spatial resolution and a large imaging volume, more and more small scintillators are employed in PET design. Since every scintillator output needs to be separately processed, the number of ADC channels in a modern PET system is rapidly increasing. In addition, as faster scintillators and a 3D imaging mode are more widely used, high-speed ADCs are often desirable. However, a PET system that employs a large number of high-speed ADCs not only consumes a large amount of power, but also is often too expensive for many applications.
Consistent with the present invention, methods and devices may be provided for digitizing gamma ray energy produced in a PET system and characterizing a peak time and a decay time constant without the use of ADCs.
Consistent with the present invention, a method for use in a PET system is performed by digitizing gamma ray energy. The method may include defining a model for a voltage pulse generated by a PET detector; determining a decay time constant, a peak amplitude, and a peak time, as parameters of the voltage pulse model, that are relevant for PET event detection; and computing the determined parameters for the generated voltage pulse.
Also consistent with the present invention, a method for use in a PET system is performed for digitizing gamma ray energy. The method may include defining a voltage pulse generated by a PET detector as a fast linearly rising edge followed by a slower exponential decay and characterized by a decay time constant, a peak amplitude, and a peak time; measuring a plurality of time intervals derived from a received voltage pulse generated by the PET detector based on a plurality of reference voltages; calculating at least the decay time constant, the peak amplitude, and the peak time of the received voltage pulse by using a plurality of time intervals; and outputting results of the calculating in digital format.
Also consistent with the invention, a device for use in PET for digitizing gamma ray energy includes a plurality of comparators, each coupled to receive a PET voltage pulse on a first input, and a first reference voltage on a second input; a plurality of counters, each having at least an enable input, a start input, a stop line, and an output; and a plurality of inverters coupled between outputs of said comparators and start or stop inputs of ones of said counters, the outputs of ones of said comparators coupled to the start or enable inputs of ones of said counters such that said plurality of counters are enabled only during an enable period when an output voltage of one of said comparators that is coupled to receive a lowest reference voltage is positive, and during the enabled period each of said counters starts counting upon a first occurrence of a rising edge of the voltage pulse at its start input and continues counting until a last occurrence of a rising edge at its stop input, and the respective outputs of said counters are digitized time intervals used to determine parameters of the PET voltage pulse for event detection.
Further consistent with the present invention, there is provided a PET system that incorporates the device mentioned above to perform PET.
Also consistent with the present invention, a PET system that uses the method mentioned above may be provided to perform PET.
Also consistent with the present invention, a counter for use in a PET system may be provided to digitize time intervals between a rising edge of a voltage pulse, generated by a PET detector, reaching a reference voltage and the falling edge reaching the reference voltage. The counter may include at least a start input and a stop input, a first register to store a system clock time of a first occurrence of a rising edge received at the start input and to remain unchanged during the entire enabled period; and a second register to store the system clock time when a rising edge occurs at the stop input during the enabled period such that an output of the counter equals a difference between the system clock times stored in the first and second registers.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate several aspects of the invention and together with the description, serve to explain the principle of the invention. In the drawings:
Reference will now be made in detail to exemplary aspects of the invention, examples of which are illustrated in the accompanying drawings. Wherever possible, the same reference numbers will be used throughout the drawings to refer to the same or like parts.
The results of the digital processing by digital subsection 108 can be further transmitted to other systems by a communication subsection 110, or displayed on a console and image reconstruction subsection 112. Communication subsection 110 may be any appropriate type of communication system or device used to transmit the results of the digital processing. Console and image reconstruction subsection 112 may include any appropriate type of console device or computer system used to display the results of the digital processing from Digital subsection 108.
where
The model may then be used to derive three pulse parameters providing information needed in PET event detection, namely, the peak amplitude Vp of the pulse, the decay time constant
Based on this pulse model, these three pulse parameters may be readily calculated from a few time intervals derived from the pulse V(t). In addition, these time intervals can be measured by use of relatively inexpensive comparators and counters.
ti(b)=tp(Vi/Vp), (2)
ti(e)=tp
where ti(b) is the time when V(t) rises above Vi and ti(e) is the time when V(t) falls below Vi. A reference voltage level Vj is chosen such that Vj<Vi<Vp, so that tj(e)>ti(e), and a time interval tij(e)=tj(e)−ti(e)
In step S4, two other reference voltage levels Vk and Vl are chosen such that Vk<Vp, and Vl<Vp. Time intervals tk=tk(e)−tk(b), and tl=tl(e)−tl(b) are also defined. Time tk(b) is the time when V(t) rises above Vk and time tk(e) is the time when V(t) falls below Vk. Time tl(b) is the time when V(t) rises above Vl and time tl(e) is the time when V(t) falls below Vl. In step S5, the two time intervals tk and tl are measured. In step S6, the peak amplitude Vp is calculated.
The following equations can be derived from equations (2) and (3):
tk
tl
Next, equation (6) is subtracted from equation (5) multiplied by (Vl/Vk) to provide:
(s+1)tk−tl=s(tp
where s=Vl/Vk−1. If tp is negligible in equation (7), then
Since the time constant
Since voltage Vl is defined to be Vl<Vp, and tp
In step S7, another reference voltage Vm is defined such that Vm<Vp. A time tm(b) is defined as the time when V(t) rises above Vm. In step S8, a time interval tmi(b)=ti(b)−tm(b) is measured. Finally, in step S9, since Vi=Vp(ti(b)/tp) and Vm=Vp(tm(b)/tp), the peak time tp is calculated as
tp=(Vp/(Vi−Vm))tmi(b), (10)
By using time intervals to derive the three parameters of the voltage pulse, no ADCs and CFDs are needed. Furthermore, when computation speed is of concern, it suffices to obtain only tij(e), ((s+1)tk−tl)/tij(e), and tmi(b). These values may then be used to obtain estimates of
As another non-limiting example, the calculations may be performed in simpler forms when the reference voltages satisfy Vl/Vk=Vi/Vj=2, and Vm=Vi/4. Under these conditions, s=Vl/Vk−1=1 and the calculations in steps S3, S6, and S9 are of forms:
Although only three parameters of the voltage pulse, five reference voltage levels, and nine detection steps are described with respect to the method shown in
A first input of each comparator 204-1 to 204-5 is coupled to receive the voltage pulse V(t). Resistors 202-1 to 202-5 are coupled in series between an input for receiving the reference voltage V0 and ground. In one implementation, the resistance values of resistors 202-1 and 202-2 are twice the values of resistors 202-3, 202-4, and 202-5. A second input of comparator 204-1 is coupled to receive reference voltage V0. The second inputs of comparators 204-2 to 204-5 are coupled to receive reference voltage V0 through one or more of resistors 202-1 to 202-5 as shown in
The inputted voltage pulse V(t) may or may not be pre-amplified. The inputted reference voltage V0 determines the lowest photon energy that can be determined. From the voltage V0, the five reference voltages may be generated by using a simple resistive voltage divider formed by the series connection of resistors 202-1 to 202-5. As a result, comparators 204-1 to 204-5 receive on their second inputs Vl, Vi, Vk, Vj, and Vm, respectively. For achieving more flexible selections of the reference voltages, programmable resistors may also be employed.
Resistors 202-1 to 202-5 may be any resistors, variable or non-variable, programmable or non-programmable. Resistors 202-1 to 202-5 may be selected so that the five reference voltages described with reference to
During the operation of digitizer 200, the rising edge of the output signal of comparator 204-5 signals the occurrence of a detection event. In addition, all of counters 206-1 to 206-4 are enabled only during the period when the output of comparator 204-5 is positive, i.e., when the input voltage V(t) is above Vm. Each of counters 206-1 to 2064 is coupled to operate so that during an enabled period, the counter starts counting upon the first occurrence of a rising edge at its “start” input and continues counting until the last occurrence of a rising edge at its “stop” input. After an event is completed, counters 206-1 to 206-4 are reset to zero by the falling edge of a logic pulse and are then ready to handle the next event. Although each of counters 206-1 to 206-4 is illustrated with a “reset” input, the reset logic is not shown. The reset logic may include any appropriate type of reset logic. Generation of the resetting logic pulse is triggered by the rising edge of the output of comparator 204-5. For example, for LSO scintillators, a pulse duration of about 100 ns should be adequate.
Each of counters 206-1 to 206-4 may be implemented by use of two registers. One of the two registers stores the system clock time of the first occurrence of a rising edge at the counter “start” input and remains unchanged during the entire enabled period. The second register stores the system clock time whenever a rising edge occurs at the counter “stop” input during the enabled period. The counter output then equals the difference between the contents of the two registers. For a noiseless voltage pulse, exactly one rising and one falling edge will be produced at the output of each of comparators 204-1 to 204-5 during an enabled period. Therefore, the content of each counter 206-1 to 206-4 provides discrete approximations to the time intervals tij(e), tl, tk, and tmi(b), respectively. For a noisy voltage pulse, however, multiple rising and falling edges can be produced during one enabled period at the output of one or more of comparators 204-1 to 204-5 and the counters may avoid generating time interval measurements that are much shorter than the actual durations of the intervals. For example, appropriate types of hysteresis comparators may be used for such purposes.
Arithmetic unit 210 calculates the decay time constant
For fast scintillators such as LSO, the values of tj, tl, tk, and ti are expected to be in the range of 40–100 ns. Therefore, a 1 GHz clock rate may be sufficient for generating accurate estimates of the required time intervals. The voltage V0 can be determined during calibration. For example, after obtaining Vp corresponding to 511 keV with a calibration source (e.g., Ge-68) and a sufficiently small initial V0, a new V0 corresponding to for example, 300 keV can then readily be calculated.
The remaining
Only shot noise is considered when obtaining a noisy output pulse for computer simulation. Shot noise is a major noise component of the charge pulse generated by a scintillator/PMT detector. This noise component arises from the stochastic characteristic of the generation of the scintillation photons and of the electron amplification in PMT. Other sources of noise, such as electronic noise and thermal noise, can be reduced by proper electronic design; they were therefore not included in the simulation studies. Dark current of the PMT was also ignored since it is often not important in pulse detection.
When assuming that the voltage pulse V(t) is obtained by directly coupling the PMT output current to a resistor, the shot noise can be obtained as
φ(t)=γN(V(t)/V(tp))
where γ=η/((tp+2
In the simulation, typical values of η=0.2 and F=1.2 for the PMT were used. For the scintillation light pulses, the total number of photons generated upon the deposition of per 1 MeV energy in LSO and BGO were taken to be 2.7×104 and 8.2×103, respectively. The decay time constant and peak time of the voltage pulse depend not only on the type of the scintillator, but also on the bandwidth of the analog portion of the measurement system. The decay time constants of 40 ns and 300 ns for the pulses generated by LSO/PMT and BGO/PMT, respectively, were used. A 10 ns peak time was assumed for both detectors. In practice, the bandwidth of the measurement system, B, should be limited by the bandwidth of the comparators and a B=40 MHz is used for two reasons. First, this bandwidth, considered high for comparators, can support the 10 ns rise time, and the longer 40 ns and 300 ns decay times. Second, relatively inexpensive comparators to provide this bandwidth are available. With these settings, the SNR φ(t) of the resulting voltage pulse was determined with γ≈0.05 for LSO/PMT and γ≈0.007 for BGO/PMT. Once φ(t) was determined, at a given time t a Gaussian noise having a standard deviation V(t)/√φ(t) was added to V(t) to obtain a noisy output pulse. Finally, the occurrence times of the generated pulses were made random with respect to the system clock.
At each photon energy E ranging from 350 keV to 700 keV, 1000 noisy events for LSO/PMT were generated.
Generally speaking, the biases were small and the differences between the results obtained with 3.5 GHz and 1 GHz clock rates were not significant. It is noted that the bias of the estimated photon energy obtained by using the 3.5 GHz clock is slightly larger in magnitude than that obtained by using the 1 GHz clock. In addition, the results are over-estimated at energies below 511 keV, and under-estimated at energies above 511 keV, when the calibration of the photon energy was performed at 511 keV.
The results shown in
Table 1 summarizes experimentally determined biases and full-width-at-half-maxima (FWHMs) of the estimated photon energy, event time, and decay time constant obtained at E=511 keV. These results show that the estimated photon energy obtained are in good agreement-with the actual values. The FWHMs observed for the photon energy correspond to energy resolutions of about 30% and 33% at 511 keV when using 3.5 GHz and 1 GHz clock rates, respectively. The FWHMs in the event time indicate that it is possible to employ a coincidence window of 10 ns and 12 ns without significantly losing true coincidence events, when using 3.5 GHz and 1 GHz clock rates, respectively. Finally, the small FWHMs obtained for the decay time constant may suggest that the scintillator crystal used in the PMT can be effectively identified based on the decay time constants.
Table 2 summarizes the bias and FWHMs of the estimated photon energy, event time, and decay time constant obtained at E=511 keV for BGO/PMT by using the 1 GHz clock. SNR×10 shows the results obtained if the SNR of the generated pulse can be made 10-fold. This higher SNR level may be achieved by employing a PMT of higher quantum efficiency and/or by reducing the bandwidth of the measurement system. With this higher SNR, reasonably good results can be obtained. In addition, BGO/PMT may be more commonly modeled as having 8.5% of light being emitted with a 60 ns decay-time constant and the remaining 91.5% of light being emitted with a 300 ns decay-time constant.
Table 3 shows the results obtained when applying the method consistent with the present invention to pulses generated with dual-exponential decay components modeled as having 8.5% of light being emitted with a 60 ns decay time constant and the remaining 91.5% being emitted with a 300 ns decay time constant with the ×10 SNR level. Even though the method consistent with the present invention is derived from the single-exponential model, the results indicate that good estimates can still be obtained when the SNR of the pulse is adequate.
Other embodiments consistent with the invention will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the invention being indicated by the following claims.
This application is based upon and claims the benefit of priority from prior Provisional Application No. 60/558,709, filed on Apr. 2, 2004, the entire contents of which are incorporated herein by reference.
| Number | Name | Date | Kind |
|---|---|---|---|
| 3961271 | Zlydak et al. | Jun 1976 | A |
| 5378893 | Murray et al. | Jan 1995 | A |
| 6281831 | Shou et al. | Aug 2001 | B1 |
| 6548997 | Bronfer et al. | Apr 2003 | B1 |
| 6901337 | Tanaka et al. | May 2005 | B2 |
| 7045802 | Vernon | May 2006 | B2 |
| Number | Date | Country | |
|---|---|---|---|
| 20050247879 A1 | Nov 2005 | US |
| Number | Date | Country | |
|---|---|---|---|
| 60558709 | Apr 2004 | US |
