DE102007014830B3 - Method for transferring data, from detector to data processing system, involves arranging slip ring between detector and data processing system and data from passing slip ring is compressed by compression method - Google Patents

Abstract

The method involves arranging a slip ring (S) between a detector (D) and a data processing system (R). The data from passing the slip ring is compressed by a compression method. The compression method comprises a step for fixing a compression ratio. The compression ratio is determined as a connection of a processor load by a buffer storage. The data is transferred selectively in a loss-free or lossy manner. An independent claim is also included for a device for transferring data, from a detector to a data processing system.

Description

The The present invention is in the field of medical technology and relates to a method and apparatus for transmitting data, at least a detector towards at least one data processing system, wherein between the detector and the data processing system at least a slip ring is provided as defined by the preambles of the independent claims.

in the It is known in the prior art to use data obtained by means of detectors, which are installed in a rotating system, via a so-called slip ring (slip-ring) to be transmitted to a connected data processing system.

Under A slip ring is a rotary joint system for transmission of electrical power or data. With help a built-in example in a computed tomography slip ring become the X-ray source the other hand, the digital detector data transmitted to a connected data processing system. The data processing system then calculate from the transferred Detector data a known attenuation picture.

One known slip ring has a physical maximum data transfer rate on, that's a transmission in real time then opposes when more data is generated, as a maximum transfer can. Currently commercial Slip rings, for example, have data rates of 100 Mbps up to 2.5 GBit / s.

Around even with larger amounts of data a real-time transmissions to be able to provide So far, the number of used slip rings has been increased until the provided data transfer rate Real-time transmission allows.

These Although the procedure allows a real-time transmission, however, the Specialist endeavors as possible few data streams or in general to use data paths to one as possible high electromagnetic compatibility to ensure.

Furthermore is the installation of additional slip rings with additional Cost, maintenance and risk of failure associated, as far as possible should also be avoided.

The The present invention therefore has the object, in the named systems a continuous real-time transmission to ensure or at least improve, without necessarily one have to provide another slip ring or a wider highway.

These The object is solved by the features of the independent claims, wherein expedient developments by the characteristics of dependent claims are described.

Especially beats the present invention is a method of transmitting data, at least a detector towards at least one data processing system, wherein between the detector and the data processing system at least a slip ring is provided and the data before passing the Slip ring are compressed by a compression method.

Thereby There is a simple way of passing through the slip ring optimally exploited data rate and a real-time transmission in a much wider scope provide.

From the US Pat. No. 6,470,065 B1 For example, a CT system is known in which data is acquired from an X-ray source, the acquired data is subsequently processed in a noise reduction unit, and the captured and noise-reduced data is then compressed in a compression unit before that compressed data is transmitted via a slip ring system to a central processing unit ,

It is about it It should also be pointed out that the average person skilled in the art has a certain Prejudice already has data compression immediately after the detector or to make a data reduction. The average expert prefers data manipulation only after viewing the depth of information, to be able to recognize and determine which Data removed or smoothed Need to become.

advantageously, can the data can be transmitted either lossless or lossy. At a lossless compression can also by a data reduction to be spoken.

As a general rule is understood by a lossless data compression that no recognizable image data lost due to data compression go, so essentially get all the visible image information stay. Optimally, only data is removed or smoothed in the order of magnitude the noise component of the detector are located. In this case is through the inventive method optimally utilized the bandwidth or data rate of the slip ring, without losing relevant image information.

In a preferred embodiment, the lossless compression described loss-free in the mathematical sense, ie the original data can be restored in bit-accurate exact form.

A lossless data compression usually has a compression factor (CF) between 1.4 to 4. At higher compression factors speaks one of an almost lossless data compression (CF between 4 to 10) and lossy data compression (CF between 10 to 100).

In a preferred embodiment a transfer takes place with a compression factor between 1.5 and 3.

Under The compression factor (CF) is the quotient of the scope of Understand output data on the amount of compressed data.

When lossy compression is understood to mean a process some of the data is lost for the purpose of compression. The original data is then usually no longer from the compression result zurückrechenbar.

By the possibility optionally between a lossless data compression and a lossy data compression can be changed by the Slip ring provided data rate or data transfer rate in optimal manner be exploited or can by providing a lossy Compression ensures that there is always a transmission possible is, independent on the type of scanning object and thus the composition of the transferred Dates.

advantageously, For example, the compression method includes a step of determining a degree of compression or factor for the transfer.

As is known, points Each compression algorithm includes at least one variable from which the degree of compression depends is. This at least one variable is set accordingly the degree of compression. To select the degree of compression There are usually two options to disposal. A possibility according to the invention is to attach a cache to the detector, then the degree of compression depending on the load of the cache.

Furthermore it is possible, the degree of compression on the basis of the data processing system incoming data, for example by monitoring the completeness the received data. Such monitoring can, for example through the transmission of checksums or a so-called header monitoring to be provided.

By this dynamic determination of the degree of compression makes it possible to as far as possible, a lossless transmission to provide and only as an alternative and if necessary to one to switch almost lossless or lossy compression. This ensures that that always the maximum possible Image data is transmitted.

A different possibility for controlling the degree of compression for audio data and video data in the coding technique known from the MPEG-2 standard, scribed z. In ISO / IEC 13818-7: 1997 (E): "Coding of Moving Pictures and audio ".

It is to be understood that the determination of the degree of compression in both directions are dynamic, in other words, that of Increased compression both as well as can be made smaller, for example, depending the memory usage of the cache. The determination of the Compression degree is thus based on current system parameters and is variable.

Further it is advantageous if the degree of compression in dependence the maximum data transfer rate the slip ring is set.

As already described, the slip ring provides the transmission path with the lowest data transfer rate in the system, allowing a determination of the degree of compression in dependence the maximum data transfer rate of the slip ring prevents a bottleneck there is also the possibility also assign the slip ring a cache and this to determine the degree of compression. If, for example in a system is known, what extent the expected data volume Also, a fixed degree of compression can be specified.

In a preferred embodiment will the data for the compression method represented by a Golomb / Rice coding or transferred. In this embodiment the output data are in a 16-bit logarithmic form.

Both codings - Golomb and Rice - represent so-called entropy codings. These are known to be optimally suited for data reduction or compression of image data. These methods provide for exponential decaying distribution of the input values for optimal operation and such distribution of image data is easily provided by transmitting difference values or entropy values. In order to determine the difference values to be transmitted, a so-called reference value is subtracted from the detector value, so that all data are present and transmitted as a difference of correspondingly selected reference points or regions. As is known, the difference values can also be represented as positive P values, for example by multiplying all negative values by ((-) 2).

A Golomb / Rice coding divides a data word to be transferred into one Part, the unär and encoded into a remainder that is binary coded. Under a unary (or: Roman) Coding is understood to be a natural one Number as a sequence of ones, followed by a zero. The zero can be used as a separator between the part, the coded unary and the part that is binary is coded serve.

Of the A person skilled in the art knows a large number of different designs a Golomb / Rice coding. By way of example only, in the following two versions cited. DA is already the difference value described above.

Common to the codings is the division of the data word into DA = q + RE.

q represents the integer part of the following quotient q = (DA - 1) / b.

Of the integer part of the quotient q is then coded and transmitted unary.

Of the Rest RE represents the fraction of the quotient cut off after the comma This is then transmitted in binary with a certain number of bits. The representation is the more accurate, the greater the number of bits, with the remainder RE is transmitted or the value b determines how many bits are needed to encode RE without loss. A reduction of this Number of bits in favor of compression is possible.

The Compression variable b and the reference value can also be in the so-called Transfer header uncompressed so that the data processing system is able to to reconstruct the corresponding picture. This happens when b is constant for all Data words is. In the preferred embodiment, b becomes dynamic calculated from the data words themselves.

In a particular embodiment, the quotient q is represented as follows and transmitted unary as described above: q = DA / 2 k ,

In the particular embodiment becomes the binary Portion of the data word then transmitted with k-digits, so that both the binary, as well as the unary Proportion of the data word depends on the parameter k. This can be done set the degree of compression in a particularly simple way, since as can be seen, in this embodiment the degree of compression is directly dependent on the selected k value.

Of the k-value is adjusted dynamically by adding to a whole Number of rounded mean of the logarithm to the base 2 of the last one 16 difference values is determined. This results in an approximately optimal k-value. This will approximate optimal coding of the data DA achieved. If q is in unary form is coded and subsequently the rest RE in binary Form as a k-bit number, this value can be calculated exactly without losses. Only if less than k bit, lossy data compression occurs one. The values are obtained from the buffer.

at Selection of a lossy transmission the determination of a maximum k-value, kmax. Should that, As described above, k values determined to be greater than kmax are from Remainder RE (k - kmax) Bits are stripped, resulting in a value RE 'to be transmitted with kmax bits. The stroking of the bits starts with the 2.bit, if necessary to obtain existing sign information in the 1st bit.

• RE' = 11101011 = 11101 (binär)

For example, the following value to be transmitted is RE 'for k = 8 kmax = 5 and RE = 11101011 (binary): (k - kmax) = (8 - 5) = 3

• RE '= 11101011 = 11101 (binary)

To the transmission to a reconstruction computer of a data processing system is the Remainder RE 'with (k - kmax) bits extended. This will change the deviation between the original value RE and the abridged Value RE 'reduced. This gives a value RE '' which is the following image processing is used.

• RE' = 11101011 = 11101 (binär)
• RE'' = 11100001 (binär)

In the example above, RE 'is extended by 3 bits, resulting in the following value RE'': (k - kmax) = (8 - 5) = 3

• RE '= 11101011 = 11101 (binary)
• RE '' = 11100001 (binary)

• RE = 11101011 (binär) = 235 (dezimal)
• RE'' = 11100001 (binär) = 225 (dezimal)
• Abweichung = 10 (dezimal)

Thus, for the example described here, the following deviation from RE to RE "results, ie from the original value to the reconstruction value:

• RE = 11101011 (binary) = 235 (decimal)
• RE '' = 11100001 (binary) = 225 (decimal)
• Deviation = 10 (decimal)

As The greater the deviation, the less pronounced the deviation to be transferred Difference value is.

Furthermore It should be noted that large difference values suggest that the deviations from the reference are very large and that's why big k values must be used for compression. In a preferred Implementation will reference a whole row of a multicellular Detector uses and then the difference of all lines calculated and coded to the reference line. In case of small deviations over the lines and thus high homogeneity result in only small values that work well with k values in the range of 7-9 compress. Significant deviations from the reference to strong fluctuations, which is usually synonymous with a noise of the data is.

With in other words, the proposed compression has the least effect to areas where the largest image information included is. This has a very beneficial effect on the overall transmission quality out. Because the lossy compression almost only at huge Differences occur and these big differences only at strong rushing data is a loss in this case essentially not disturbing, because these data carry almost no relevant image information.

As can be seen from the equations, thereby the required data rate for transmission a data word restricted, since the binary transmitted Rest RE 'only transmitted with kmax bits becomes. The bigger there kmax chosen becomes, the more accurate is the transferred one Value RE ', the greater but also the needed Data rate for transmission of the data word. kmax has in a preferred embodiment a value between 1 and 16. The largest kmax value in a test setup should be dependent on the signal-to-noise ratio be determined so that guaranteed can be that at maximum kmax Value a substantially lossless data reduction or a almost lossless data compression takes place, whereby it can be seen that at a kmax = 16 and underlying 16-bit data one lossless data reduction is possible.

Also kmax, and thus the degree of compression, becomes dynamic during a measurement customized. Therefor According to the invention, the utilization of the Cached monitors. Should the cache reach a fixed load, kmax is lowered by 1, kmax is reduced by 1 until, until the specified utilization of the buffer is no longer exceeded. In the opposite case, so if a fixed utilization of the cache kmax is increased by 1.

By the introduction of kmax with the dynamic adjustment described above and above described change from lossless data reduction to almost lossless data compression can be done in a surprisingly simple way and Way a constantly optimal utilization of the existing data throughput rate and thus ensures an optimized real-time representation become.

Further beats the present invention in particular a device for transmitting from data from at least one detector to at least one data processing system before, wherein between the detector of the data processing system at least a slip ring is provided and wherein between the detector and the slip ring is provided a compression device.

A there is another solution according to the invention also in a system for transmitting of data, in particular of image data of a medical device, of at least one movable detector of a medical device to a stationary Data processing system, being between the detector of the data processing system at least one slip ring is provided and wherein between the Detector and the slip ring a compression device provided is.

The Advantages of the invention can be achieved by the addition of a compression device, the performs a method described above, particularly simple Provide fashion. In addition, the compression device, as far as the detector itself does not yet comprise a buffer, a to provide such. This makes it possible in the simplest way and way a detector system with a device according to the invention upgrade or to retrofit.

In a preferred embodiment, the detector is a detector of a computer tomograph. As is known, in a computer tomograph, a detector and a radiation source rotate about a measurement object or object. Thus, such a structure requires the use of a slip ring to data from the rotating or movable system towards stationary reconstruction computer.

Of the Detector in a computer tomograph measures the radioactive radiation during a measurement, from the opposite Radiation source is emitted and the measurement object or subject happened.

optimally, is determined by the proposed data compression method or reduction method only the data portion is not transmitted or smoothed, which is on the order of magnitude of the noise component is located. This proportion is at a detector for one Computed tomography is relatively easy to measure and is among others dependent on the following factors, such as From quantum efficiency, radiation drift, decay, afterglow, scintillation efficiency, etc.

Consequently the proposed method and apparatus are optimally suited for the transmission of Data generated at a detector of a computer tomograph and must be transferred to a reconstruction computer because a data reduction or even a data compression without loss can be done by relevant image information.

Further Features and advantages of the invention will become clearer in reading the following, by way of example only, and not by way of limitation a preferred embodiment with reference to the accompanying drawings, in which:

1 shows a measuring section between a computer tomograph and a reconstruction computer.

When embodiment becomes a data transfer from a detector D of a computer tomograph to a so-called Reconstruction computer R described. The detector D registered while the X-rays, after passing through the measuring object or subject in this incident. As a result, it is known the possibility to represent a so-called attenuation profile. While a measurement is taken from the X-ray source and the detector D is rotated 360 ° [DEG] and this leads to Need to use a slip ring S, as a fixed Would screw in cable.

typically, are used as detectors D xenon ionization chambers, semiconductors or Scintillation crystals used.

The Weakening profiles registered by detector D become analog / digital converted and transferred to the reconstruction computer R. The data stream is over the Slip ring S shipped.

The analog / digital conversion takes place immediately after receiving the data, in an adjacent to the detector D or this associated A / D converter. For this purpose and to compensate for minor errors on the highway to be able to the detector D is assigned a smaller buffer. The cache usually has a size to 50 to 100 so-called readings to cache. Under In this case, a detector value is a detector value of a detector channel understand. Depending on the type of detector, the detector D has a different one Variety of detector channels on.

the Detector D is also a compression or compression device according to the invention associated with compressing the digital data from the cache. It is assumed that a compression before the slip ring S should be done. The exact location of the compression device is therefore irrelevant to the present invention.

So far the data rate of the slip ring S is not exhausted, there is a lossless Data compression by means of a Golomb / Rice coding. described becomes how a single data word DA is transmitted.

From the detector raw data is subtracted selected reference data, so that only difference values result. So it will be with the present method no absolute values, but only difference values, transfer so-called entropy values. The reference values are transmitted once uncompressed, since this for the reconstruction of the image data in the reconstruction computer R are necessary.

Further is ensured by the following procedure, that is to be transferred Value DA can only have a positive value. Should DA already have a positive value, DA is multiplied by (+2). If DA has a negative value, DA is multiplied by ((-) 2) and subtracting 1 from this value.

DA is then represented by the integer part of a quotient q and a truncated remainder RE as follows: DA = q + RE

For the quotient q applies: q = DA / 2 k

The integer part of the quotient q then becomes unary n → (n-1) × 1 + 0 coded. In other words, the unary coding represents a natural number n as a result of n-1 ones followed by a 0.

Of the Rest RE represents the truncated portion of quotient q Rest RE becomes binary encoded with a length of k bits. Consequently is on the one hand q, as well as the transmitted one Length of the Remaining RE depends on the variable k. How to recognize, must the variable k is continuously adapted to the DA values, for an optimal reduction ratio to be able to receive the data.

A sufficiently optimal k value results from:

It So the last 16 DA values are read from the cache, in each case the logarithm to base 2 of DA is calculated and then the Average calculated. For better handling of follow-up operations can also be the next one integer rounded value to be used for further calculation.

• DA = 400
• k = 7

somit ergibt sich q = DA/2k = 400/27 = 400/128 = 3,125; For a better understanding, the following example is given:

• DA = 400
• k = 7

thus results q = DA / 2 k = 400/2 7 = 400/128 = 3.125;

• 110

The integer part of q is 3 and gives unary representation:

• 110

• 1111101

The truncated remainder RE of the quotient is 125 and results in a binary representation of 7 bits

• 1111101

This results in the following for the data word to be transmitted: DA = q + RE = 110 + 1111101

As to recognize, sets the procedure described so far lossless data reduction ready. The following will now be a lossy or almost lossless data transmission described.

First, will introduced a variable kmax. In the one listed here embodiment kmax is selected between 1 and 16. The larger this kmax is, the more accurately the detector value is transmitted.

First if the k-value calculated above is greater than kmax is a lossy data compression. The lossy data compression limited thereby referring to the remainder RE of the data word, to a deviation of the true detector value from the transmitted Value as possible to keep low.

As described above, the remainder RE is basically transmitted with k bits. If the calculated k value is now greater than kmax, (k-kmax) bits are deleted from the remainder RE. If, for example, k has the value 8 and kmax has the value 5, 3 places of RE are deleted and a value RE 'is obtained. The value RE 'is transmitted to the reconstruction computer R and again extended by this as many bits as were deleted. For a better understanding, the following example is given: k = 8 Kmax = 5 RE = 11101011 (binary) detector value RE '= 11101011 = 11101 (binary) transfer value RE '' = 11100001 (binary) reconstruction value

As To detect, thus results in a deviation of the detector value from the reconstruction value of 10 (decimal). This deviation works thereby on large transmission values less than on lower transmission values. In other words, the data compression described has only a small Effect on the relevant information content of the attenuation profile.

kmax is during the transmission dynamic adjusted. Therefor the latch associated with the detector D is monitored. Should the cache exceed a specified load, kmax is lowered by 1. Conversely, kmax is increased by 1 as soon as the utilization of the cache set a specific Value falls below. As you can see, this is a steady optimal utilization of the available standing transmission rate.

Although the present invention has been fully described above with reference to a preferred embodiment, it should be apparent to those skilled in the art that various changes are possible within the scope of the appended claims without departing from the inventive concept and the protection claimed. In particular, those skilled in the art will recognize that data compression and reduction may also be provided by other algorithms. In particular, the skilled person, since it is This is preferably image data, preferring a so-called entropy coding, such as a so-called Huffman coding.

The Description of the invention is not limiting with regard to a to understand certain physical realization. For one relevant It is particularly obvious to a person skilled in the art that the invention partially or completely can be executed in software and / or hardware, as well as on a or more physical products - in particular also computer program products - distributed can be.

Claims (9)

A method for transmitting data from at least one detector (D) to at least one data processing system (R), wherein between the detector (D) and the data processing system (R) at least one slip ring (S) is provided that the data before passing of the slip ring (S) are compressed by means of a compression method, the compression method comprising a step for determining a degree of compression, characterized in that the degree of compression is determined as a function of a load of a buffer. Method according to claim 1, characterized in that that the data optionally lossless or lossy transfer become. Method according to claim 1, characterized in that alternatively, the degree of compression depends on the maximum data transfer rate of the Slip ring (S) is determined. Method according to one of claims 1 to 3, characterized that the data for the compression method represented by a Golomb / Rice coding become. Device for transmitting from data from at least one detector (D) to at least one data processing system (R), being between the detector (D) of the data processing system (R) at least one slip ring (S) is provided that between the detector (D) and the slip ring (S) a compression device it is seen that the compression device in different Compression degree is switchable, characterized in that the Compression degree depending is switchable from a load of a buffer. Device according to claim 5, characterized in that that the compression device optionally into a lossless or a lossy mode is switchable. Device according to claim 6, characterized in that the degree of compression alternatively as a function of the maximum data transfer rate of the slip ring (S) is switchable. Device according to one of claims 5 to 7, characterized the compression device comprises a Golomb / Rice routine. Device according to one of claims 5 to 8, characterized the detector (D) is a detector of a computer tomograph.