DE102019202830A1 - Method for evaluating an image quality of a volume image data set - Google Patents

Abstract

The invention relates to a method for evaluating an image quality of a volume image data set depicting at least part of the thorax of a patient, the volume image data set being based on projection measurement data recorded by means of a computed tomography device and comprising a three-dimensional matrix of image pixel values, having the steps of receiving the volume image data set by means of a first Interface, - determining, based on the volume image data set and by means of a computing unit, a first probability of occurrence of a first artifact that is caused by the movement of the patient's heart, and / or a second probability of occurrence of a second artifact in the volume image data set that is caused by the movement of the thorax is caused by the patient's breathing, evaluation of the image quality of the volume image data set by means of the computing unit based on the determined first probability t and / or based on the second probability, and output of the assessment of the image quality by means of a second interface.

Description

The invention relates to a method and a device for evaluating an image quality of a volume image data set depicting at least part of the thorax of a patient, a computed tomography device comprising a device for evaluating an image quality, as well as a computer program product and a computer-readable storage medium.

Medical X-ray applications are an essential part of everyday clinical practice. X-ray imaging by means of a computed tomography device (CT device) makes an important contribution. The X-ray radiation used is generated by means of an X-ray tube, also called an X-ray source. The X-ray tube is arranged on a turntable which, when the CT device is in operation, performs a rotational movement around a system axis, also called the patient axis or z-axis, along which the patient's body region to be imaged is positioned. During the rotational movement of the X-ray tube, projection measurement data are recorded for a plurality of angular directions by means of the X-ray detector. The projection measurement data is a projection or a plurality of projections which contain information about the attenuation of the radiation by the examination object, for example a patient, for the respective angular direction. Various methods for scanning an examination object with a CT device are known. For example, circular scans, sequential circular scans with a sequential advance of the patient along the system axis or spiral scans with a continuous advance of the patient table are used. A three-dimensional volume image data set or two-dimensional slice image data sets of the examination subject can then be reconstructed from the projection measurement data, for example by means of a so-called filtered back projection or iterative reconstruction method.

One application of computed tomography imaging is imaging the thorax of a patient, and in particular cardiac imaging with a CT device. For high-quality CT recordings of the thorax it is a prerequisite that the patient holds his breath as much as possible while recording the projection measurement data so that the thorax does not move while recording the projection measurement data. The quality of the patient's breath holding is linked to the quality of the training, the breathing command and the patient's willingness to cooperate. In addition, cardiac imaging also requires a good resting state of the heart. The heart rate can usually be adjusted with the help of medication, but the heart's rhythm can only be adjusted to a limited extent. This represents a particular challenge when the final volume image data set is composed of a plurality of partial image data sets which originate from a plurality of successive cardiac cycles. This is the case in particular when the extent of the X-ray detector used is not sufficient to scan the heart as a whole.

This initial situation can often lead to artifacts in the volume image data set in CT images of the thorax or explicitly of the heart. In the event that the image datasets are not properly checked by an experienced radiologist or experienced medical technician directly after they have been taken, this can result in the patient being sent home even in the case of images with artifacts, even though the images are only of limited use.

Due to the strong influence of the patient, a fundamental solution to the problem is in retrospect, i.e. after recording the volume image data set, only possible with difficulty. There are approaches and solutions to correct the recorded images by registering the individual parts and thus to counteract the triggering causes such as breathing or an irregular heartbeat. Other solutions are based on very complex, hardware-driven solutions which minimize the influence of the above factors and thus try to avoid the occurrence of artifacts from the outset. For example, by using the largest possible detector in the direction of the longitudinal axis of the patient (area detector system), the largest possible area can be recorded at the same time, so that the recording times can be kept as short as possible or a string of partial image data sets from different cardiac cycles can be avoided . However, this requires expensive detector sizes of at least 14 cm or more.

In many cases, a check of the image data sets in order to guarantee a sufficiently high image quality after the projection measurement data have been recorded cannot be avoided.

The object of the invention is therefore the possibility of providing an improved check of a volume image data set depicting at least a part of the thorax of a patient with regard to its image quality.

The object is achieved by the features of the independent patent claims. Further advantageous and in part inventive embodiments and developments of the invention are set out in the subclaims and the following description.

The solution to the object according to the invention is described below both in relation to the claimed method and in relation to the claimed device. Features, advantages or alternative embodiments mentioned here are also to be transferred to the other claimed subjects and vice versa. In other words, the present claims (which are directed, for example, to a device) can also be developed with the features that are described or claimed in connection with a method. The corresponding functional features of the method are formed by corresponding objective modules.

The invention relates to a method for evaluating an image quality of a volume image data set depicting at least part of the thorax of a patient, the volume image data set being based on projection measurement data recorded by means of a computed tomography device and comprising a three-dimensional matrix of image pixel values. The method according to the invention comprises the steps of receiving, first determining or second determining, evaluating and outputting.

In the receiving step, the volume image data set is received by means of a first interface. The three-dimensional matrix of image pixel values of the volume image data set allows a three-dimensional, in particular three-dimensional, spatial representation of the examination area of the patient measured by means of the projection measurement data. The volume image data set can also be represented as a plurality of slice image data sets. A slice image data set comprises in each case a slice of the volume image data set at a position along a marked axis, in particular the z-axis. A slice image data record allows a two-dimensional, in particular spatially two-dimensional, representation of the respective slice. In particular, the volume image data record depicts at least part of the patient's thorax. In particular, it can also display the patient's heart completely or at least partially.

Based on the volume image data set, the computing unit determines the first probability for the occurrence of the first artifact in the step of the first determination, or the second probability for the occurrence of the second artifact is determined in the step of the second determination. The second determination can take place independently of the first determination or a second probability can also be determined independently of a first probability. A temporal sequence should also not be expressed by the expressions “first determination” and “second determination” or “the first probability” and “the second probability”. In this context, the wording only serves to differentiate between the steps and the determined probabilities.

The first artifact can also be called a heart artifact in the following, for example. The second artifact can also be called breathing artifact below, for example. The first or second artifact can each result from the fact that a change in the position or the state of an anatomical structure occurred during the recording of the projection measurement data. In the case of the first artifact, a changed condition of the heart due to the heartbeat and, in the case of the second artifact, a movement of the thorax or the internal structures due to breathing is the cause. For example, the heart is imaged in a first slice image data set in a different phase of the cardiac cycle than in a slice image data set adjacent to it. This can in particular result in inconsistencies occurring in the volume image data set along the z-axis and a reproduction of the imaged area for further diagnosis is impaired. In the first ascertaining step, a probability is ascertained that a heart artifact will occur in a representation of the volume image data set and impair it. In the second determination step, a probability is determined that a respiratory artifact is present in the volume image data set or that the artifact occurs in a representation of the volume image data set.

The first and second determination can be carried out alternatively to one another. In particularly preferred embodiments, however, both the step of the first determination and the step of the second determination are carried out. This enables a particularly meaningful and differentiated assessment of the image quality and, if necessary, a more targeted derivation of further measures.

The first or second determination of the first or second probability can include determining a parameter or a specific numerical value, for example based on the image pixel values of the volume image data set or the slice image data sets from which the volume image data set consists, the parameter or the numerical value with the occurrence of the first or second artifact connected, ie correlated, is. In this case, a parameter on which the first probability is determined can be a different parameter than the parameter on which the second probability is determined. In particular, the first probability can be determined based on a different image section than the image section which is used to determine the second probability. The latter can represent a particularly simple possibility of distinguishing between a first artifact and a second artifact. That is to say, this can in particular represent a simple possibility of assigning an underlying cause to an artefact that occurs.

Based on the determined first probability and / or based on the determined second probability, the image quality of the volume image data set is then assessed in the step of evaluating. The image quality can at least be assessed to the effect of whether the appearance of a first or a second artifact in the representation of the volume image data set is to be expected for a viewer. The assessment can reflect or estimate a level of impairment due to an artifact occurring in the representation of the volume image data set or for a diagnosis based on the volume image data set. An assessment based on the first and / or second probability can include classifying one or more essentially continuous numerical values or continuous numerical values, which represent the first or second probability, into different categories or grades, so that a simplified assessment of the result by clinical Personnel is enabled. The link between a degree of impairment and the determined first probability or the determined second probability can, for example, be established empirically beforehand through preliminary examinations based on existing volume image data sets and an assessment by experienced clinical staff.

A level of impairment or a classified category determined in the evaluation step can be linked to an instruction and / or a control command. The control command can include a control of a CT device. This can include, for example, repetition of recordings, adaptation of recording or reconstruction parameters or the like.

The evaluation is then output using a second interface in the output step. The evaluation can in particular be output to an output unit which a user, i. E. Allows clinical staff to view or access the assessment. This can be in the form of a representation unit, i. a display or monitor, or a voice output or similar. Based on the assessment, for example, a further procedure with regard to optimal diagnosis and care of the patient can be determined. The evaluation can also be output to a control unit of a computed tomography device or to a reconstruction unit of a computed tomography device, in particular if the evaluation is linked to an instruction and / or a control command.

Furthermore, all the steps described above can be carried out automatically. “Automatically” in the context of the present application means that the respective step runs independently by means of the claimed device, and essentially no operator interaction is necessary for the respective step. The operator only has to confirm calculated results or carry out intermediate steps. For example, the operator can select an image section. In further variants of the invention with steps carried out “fully automatically”, no operator interaction at all is necessary to carry out these steps. In particular, all steps of the claimed method can be carried out “fully automatically”. Regardless of whether the individual steps are carried out “automatically” or “fully automatically”, the method according to the invention can be part of a workflow that additionally requires interaction from an operator. The interaction with the operator can consist of the operator manually selecting a recording protocol and / or a clinical question, for example from a menu presented by means of a screen. An automatic implementation has the advantageous effect that an evaluation of the volume image data set can be available quickly and at all times, in particular also directly after recording the projection measurement data.

Advantageously, the method according to the invention can circumvent the need for experienced clinical staff to always be available for an initial assessment of the volume image data set, which can lead to waiting times for a patient or to a strong time commitment of the clinical staff, or the consequence that a patient is discharged without at least one Volume image data set is available with sufficient quality for further diagnosis. In contrast, clinical staff with less experience can be advantageously supported in assessing the present volume image data set and in deriving measures. This is particularly advantageous in the case of volume image data sets with very complex image information, for example with a further dependency on a time component, ie if, for example, the volume image data set can also be displayed as a function of the phase of the cardiac cycle. The use of the method described here for evaluating the image quality can thus advantageously contribute to relieving the burden on clinical staff and to an efficient workflow in clinical operation.

In an advantageous variant of the method, the method according to the invention also includes the steps of deriving an instruction and a further output step.

In the deriving step, an instruction based on the assessment of the image quality is derived by means of the computing unit. Deriving here means that an instruction is determined or one or more suggestions for an instruction are determined based on the assessment. The instruction can correspond in particular to a proposal for the clinical staff or a strategy for a further procedure, so that if possible a volume image data set with a sufficiently high quality can be provided for a further diagnosis or diagnosis based thereon. The derived strategy based on the evaluation can also include that the volume image data set has to be accepted in the present form, since the prospect of improvement is assessed as low even if projection measurement data are recorded again. The derivation of the instruction can in particular be carried out automatically.

In the next step of output, the instruction is output. The handling instruction is output in particular to an output unit which enables a user to access the handling instruction. It is also conceivable that an implementation of the instruction, if possible, takes place automatically or after authorization by the clinical staff. The instruction can then accordingly also be output to a control unit or to a reconstruction unit. This can relate, for example, to a new reconstruction with improved reconstruction parameters, the setting of recording parameters or a repetition of recordings.

The deriving and issuing of instructions can advantageously accelerate decision-making processes. Inexperienced users in particular can be advantageously supported with the help of options for action.

• - Erneutes Aufnehmen von Projektionsmessdaten,
• - Anpassen eines Aufnahmeparameters für ein erneutes Aufnehmen von Projektionsmessdaten, oder
• - Anpassen eines Rekonstruktionsparameters für eine Rekonstruktion des Volumenbilddatensatzes basierend auf den bereits aufgenommenen Projektionsmessdaten oder basierend auf erneut aufgenommenen Projektionsmessdaten.

In an advantageous variant of the method, the specific instruction includes at least one instruction from the following group of instructions:

• - Re-recording of projection measurement data,
• - Adjustment of an acquisition parameter for a new acquisition of projection measurement data, or
• Adaptation of a reconstruction parameter for a reconstruction of the volume image data set based on the projection measurement data already recorded or based on projection measurement data recorded again.

The application of an instruction from this group to an instruction can advantageously contribute to ensuring a sufficiently high image quality for a further diagnosis. In addition, the group of instructions can also include other recommendations for action.

In an advantageous embodiment of the method, the method includes the step of determining the first probability and also has the step of first determining. In this case, a first image area is determined based on the volume image data set, which comprises at least a partial area of the patient's heart, and the determination of the first probability is based on the determined first image area.
Additionally or alternatively, the method can include the step of determining the second probability and, for this purpose, also have the step of second determining. A second image area is determined based on the volume image data set, which area comprises at least a partial area of the part of the patient's thorax comprised by the volume image data set, and the determination of the second probability is based on the determined second image area.

The first image area can also be called a heart image area, for example. The second image area can also be called the thorax image area, for example. The first specific image area preferably differs from the second specific image area. The first image area preferably essentially comprises the heart. The second image area preferably includes at least one image section outside the imaged heart of the patient. In this case, too, it is pointed out that the expressions “first” and “second” do not necessarily express a dependency on one another or a time sequence, but rather serve in particular to differentiate.

The first and / or second image area can be determined automatically or fully automatically. The determination of the first or second image area can include an interaction of a user. For example, a user can define a maximum extent based on the volume image data set or the slice image data sets. For example, the operator can set a seed point for segmenting an anatomical structure, based on which an image area can be determined. It is also conceivable for the heart or other anatomical structures to be recognized fully automatically based on the volume image data set and for a first or second image area to be determined based thereon. For example, characteristic anatomical structures in the volume image data set or the slice image data sets can be automatically detected and an image area can be determined based thereon. The image areas can be determined based on a segmentation of anatomical structures, for example the heart. For example, a region-oriented, area-based or an edge-based segmentation algorithm is used for segmentation. In the automated determination of the first or second image area, a machine learning method is also conceivable.

A machine learning process or an artificial intelligence system can be used to describe a system for the artificial generation of knowledge from experience. The use of such a system can include a recognition of patterns and regularities in the training data. After the training phase, the artificial intelligence system can extract features or relationships from previously unknown data, for example. For example, a machine learning method is a neural network (“neural network”), for example a deep neural network (“deep neural network”) and / or a convolutional neural network (“convolutional neural network”, CNN). Another method of machine learning can also be used. An artificial intelligence system can be used in the first or second determination in order to automatically recognize anatomical structures, based on which the first or second image area can be determined. An artificial intelligence system can be used in the first or second determination in order to automatically determine or suggest particularly suitable image sections for the first and second determination of the first or second probability. By using an artificial intelligence system, all relevant influencing variables can be taken into account for the first or second determination, including those for which a user cannot estimate a relationship. Such a machine learning method can, for example, be trained with a large number of image areas defined manually by a user.

The determination of a specific first image area for determining the first probability and / or the determination of a specific second image area for determining the second probability can advantageously and in a simple and improved manner make it possible to assign an occurring first artifact to a first cause and / or an occurring one assign second artifact to a second cause. This is particularly the case when a first or second artifact occurs essentially locally in the first or second image area. A heart artifact, for example, has an effect in particular on the image area of the imaged heart.

In a variant of the method according to the invention, the volume image data set is based on projection measurement data recorded during at least two different cardiac cycles of the patient's heart and can be displayed as a plurality of slice image data sets which originate from the at least two different cardiac cycles and which each include a two-dimensional matrix of image pixel values .

In this case, the volume image data set can be described as being composed of a plurality of blocks, ie partial image data sets, of slice image data sets, each block being based on the recording of projection measurement data during a specific cardiac cycle. The blocks can each comprise a plurality of slice image data sets. They can also comprise only one slice image data set, ie one slice of the volume image data set. The number of slice image data records per cardiac cycle block can be determined by the X-ray detector used in the CT device, in particular by its extension in the form of pixel lines in the direction of the system axis. Such a procedure is particularly with associated with the use of X-ray detectors, the extent of which along the system axis is not sufficient to fully accommodate a patient's heart. X-ray detectors with smaller dimensions in the direction of the system axis are particularly cheaper.

In an advantageous variant of the method based on volume image data composed of a plurality of blocks of slice image data sets, when determining the first probability or the second probability, a number of boundary slice image data sets is determined, the number of boundary slice image data sets corresponding to the subset of slice image data sets which are in each case a border between two adjacent slice image data sets in the volume image data set which originate from two different cardiac cycles.

This means that in this variant of the method, those slice image data sets are determined which lie on a boundary between a block of slice image data sets of a cardiac cycle and an adjacent block of slice image data sets of a following (or preceding) heart cycle. In particular, the first determination of the first probability and / or the second determination of the second probability is then based on the determined boundary slice image data sets. A first or second artifact usually occurs more frequently at such boundaries between two blocks of slice image data records. The analysis of the boundary slice image data sets thus represents a particularly efficient option for evaluating the volume image data set.

In an advantageous embodiment, when determining the first probability or the second probability for each boundary between two adjacent slice image data sets in the volume image data set, which originate from two different cardiac cycles, an individual probability value for the occurrence of the first artifact or for the occurrence of the second artifact is used certainly. The individual probability value determined for each boundary is based on two boundary slice image data records that are adjacent in the volume image data record.

An individual probability value is determined for each boundary between a block of slice image data sets of a cardiac cycle and an adjacent block of slice image data sets of a following or preceding cardiac cycle. With a plurality of transitions, i.e. Limits, a plurality of individual probability values is determined accordingly.

Since an individual probability value is determined for each transition, an assessment based on an individual limit and thus differentiated is advantageously possible. A meaningful evaluation of the entire volume image data set can also be made possible by including all or a subset of the determined individual probability values.

According to an advantageous embodiment, a threshold image data record is generated for each of the number of boundary slice image data sets based on the application of at least one threshold value to the image pixel values of a respective boundary slice image data set, the individual probability value determined for each boundary then being based on the threshold image data sets.

A threshold image data record is generated at least for each boundary slice image data record. In addition, further threshold image data sets can also be generated based on further slice image data sets of the volume image data set. The formation of a threshold image data record can include that a calculation rule is applied depending on whether a previously defined threshold value is exceeded or not reached by a respective image pixel value. For example, an image pixel value can be replaced by another value based on this. A plurality of threshold values can also be applied to the image pixel values in order to generate a threshold value image data set.

By forming the threshold image data set, for example any outliers in the image pixel values that result only from incorrect behavior of the X-ray detector or noise values can be excluded.

Structures imaged in the slice image data sets, which, for example, predominantly assume image pixel values above or below a threshold value, can also be explicitly excluded or included from the analysis. The formation of threshold image data sets can therefore advantageously enable a more robust and improved first or second determination of the first or second probability.

In a further advantageous variant of the method according to the invention, the individual probability value determined for each boundary is based on the pixel-by-pixel difference between at least two locally corresponding subsets of image pixel values of two adjacent boundary slice image data sets, or, if threshold image data sets of the boundary slice image data sets have been generated, on the pixel-by-pixel difference between at least two locally corresponding subsets of image pixel values of two threshold image data sets associated with two adjacent boundary slice image data sets.

In other words, an individual probability value is based in this case either on a pair of boundary slice image data sets or on threshold value image data sets formed therefrom. Locally corresponding subsets means that the subset of image pixels having the image pixel values in each slice image data set comprises the same spatial coordinates within a respective slice image data set. At least one respective individual probability value for the first probability can be based on the first specific image area. At least one respective individual probability value for the second probability can be based on the second specific image area. In addition, a respective individual probability value can also be based on further subsets of image pixel values.

The formation of a pixel-by-pixel difference comprises, for example, that in each case a difference between two spatially corresponding image pixel values, i.e. is formed between two image pixels having the same position coordinates within two slice image data sets that are adjacent in the z-direction. The differences can also be quantified in other ways, for example by means of a quotient formation of the image pixel values or the like. Large, abrupt or increased differences between the boundary layer image data sets or the threshold value image data sets formed therefrom can advantageously indicate an increased probability of the occurrence of the first or second artifact. In contrast, in the case of artifact-free image data, only minor differences are generally expected between two directly adjacent slice image data sets, since the imaged slices are usually very fine, i.e. are very thin compared to the depicted anatomical structures of the patient. The determination of the individual probability value determined for each limit on the pixel-by-pixel difference between at least two spatially corresponding subsets represents in particular a particularly simple possibility of determining the occurrence of an artifact.

The subset of image pixel values for the first determination of the first probability differs from the subset which is used for the second determination of the second probability. As a result, it is possible in a very simple manner to distinguish the causes of occurring differences in the image pixel values or to assign these to a first or second artifact.

According to an advantageous embodiment of the method according to the invention, the individual probability value determined for each boundary is additionally based on a further locally corresponding subset of image pixel values of a slice image data set adjacent to one of the two boundary slice image data sets, which is not part of the number of boundary slice image data sets.

This can mean that in addition to the boundary slice image data sets, further slice image data sets are also used to determine the respective individual probability values. These can be used as a kind of normalization. For example, the differences found between the boundary image data sets, i.e. at a transition between two blocks from different cardiac cycles, are set in relation to differences which occur between two slice image data sets that originate from the same cardiac cycle. This means that the differences occurring at a boundary between two blocks can be put in relation to the expected differences within a block. The expected differences can be caused, for example, by slight differences in the depicted anatomical structures from layer to layer. This advantageously allows a more robust determination of the first or second probability.

In a further variant of the method according to the invention, the method comprises the step of determining a heart image area based on the volume image data set or the method comprises the step of determining a thorax image area based on the volume image data set, and the respective subsets of image pixel values, the basis for determining the Individual probability values are determined by the heart image area or the thorax image area determined as part of the method.

According to an advantageous embodiment of the method, each individual probability value is also based on a value of the image noise of a slice image data set. In particularly advantageous refinements, each individual probability value is based on a value of the image noise of at least one of the boundary slice image data sets on which the respective individual probability value is based.

By including an image noise value in the determination of the individual probability value, image pixel values can be excluded from the analysis which are in the area of the image noise and therefore cannot provide any relevant information. Any specific differences between image pixel values can also be related to an image noise value. The image noise value can represent a reference value for expected differences. This can advantageously lead to a more robust and clearer result when determining the respective individual probability value.

In an advantageous embodiment of the method according to the invention, when determining the first probability or the second probability, an overall probability value is determined for the volume image data set based on the individual probability value determined for each limit.

An overall probability value advantageously allows a global assessment of the image quality of the volume image data set and thus a simplified evaluation of the volume image data set for clinical staff. The overall probability value can be based, for example, on a (weighted) sum, a mean value or also a maximum value of the individual probability values.

Both the individual and the overall probability value can be included in the assessment of the image quality in the assessment step in order to enable an assessment of the image quality that is as differentiated as possible. However, the assessment can also be provided solely on the basis of the overall probability.

• - einer ersten Wahrscheinlichkeit eines Auftretens eines ersten Artefakts, welches durch die Bewegung des Herzens des Patienten hervorgerufen wird, und/oder
• - einer zweiten Wahrscheinlichkeit eines Auftretens eines zweiten Artefakts in dem Volumenbilddatensatz, welches durch die Bewegung des Thorax durch die Atmung des Patienten hervorgerufen wird,

Die Recheneinheit ist weiterhin ausgebildet zum Bewerten der Bildqualität des Volumenbilddatensatzes basierend auf der ermittelten ersten Wahrscheinlichkeit und/oder basierend auf der zweiten Wahrscheinlichkeit. Die Vorrichtung weist außerdem eine zweite Schnittstelle, ausgebildet zum Ausgeben der Bewertung der Bildqualität, auf.The invention also relates to a device for evaluating an image quality of a volume image data set depicting at least part of the thorax of a patient, which is based on projection measurement data recorded by means of a computed tomography device and comprises a three-dimensional matrix of image pixel values, having a first interface designed to receive the volume image data set and a computing unit designed to determine, based on the volume image data set,

• a first probability of the occurrence of a first artifact which is caused by the movement of the patient's heart, and / or
• - a second probability of occurrence of a second artifact in the volume image data set, which is caused by the movement of the thorax by the patient's breathing,

The computing unit is also designed to evaluate the image quality of the volume image data set based on the determined first probability and / or based on the second probability. The device also has a second interface designed to output the assessment of the image quality.

A device according to the invention for evaluating an image quality can in particular be designed to carry out the previously described methods according to the invention and their aspects. The device for evaluating an image quality can be designed to carry out the method and its aspects in that the interfaces and the computing unit are designed to carry out the corresponding method steps.

The invention also relates to a computed tomography device comprising a device for evaluating an image quality.

In particular, the invention relates to a computer program product with a computer program which can be loaded directly into a memory of a device for evaluating an image quality, with program sections in order to carry out all the steps of the method for evaluating an image according to one of claims 1 to 13 when the program sections are to evaluate an image quality.

The invention also relates to a computer-readable storage medium on which program sections readable and executable by a device for evaluating an image quality are stored in order to carry out all steps of the method for evaluating an image quality according to one of claims 1 to 13 when the program sections from the device for evaluating an image quality.

A largely software-based implementation has the advantage that devices that have already been used for evaluating an image quality can easily be retrofitted with a software update in order to work in the manner according to the invention. In addition to the computer program, such a computer program product can optionally include additional components such as documentation and / or additional components, as well as hardware components such as hardware keys (dongles etc.) for using the software.

• 1 einen schematischen Ablauf einer Variante eines Verfahren zur Bewertung einer Bildqualität eines Volumenbilddatensatzes,
• 2 eine schematische Darstellung eines Volumenbilddatensatzes mit einem ersten und zweiten Bildbereich,
• 3 eine schematische Darstellung eines Volumenbilddatensatzes aufweisend eine Mehrzahl an Schichtbilddatensätzen,
• 4 einen schematischen Ablauf einer weiteren Variante eines Verfahren zur Bewertung einer Bildqualität eines Volumenbilddatensatzes,
• 5 eine schematische Darstellung einer Vorrichtung zur Bewertung einer Bildqualität eines Volumenbilddatensatzes,
• 6 eine schematische Darstellung eines Computertomographie-Geräts aufweisend eine Vorrichtung zur Bewertung einer Bildqualität eines Volumenbilddatensatzes.

In the following, the invention is explained on the basis of exemplary embodiments with reference to the accompanying figures. The representation in the figures is schematic, greatly simplified and not necessarily true to scale. Show it:

• 1 a schematic sequence of a variant of a method for evaluating an image quality of a volume image data set,
• 2 a schematic representation of a volume image data set with a first and second image area,
• 3 a schematic representation of a volume image data set having a plurality of slice image data sets,
• 4th a schematic sequence of a further variant of a method for evaluating an image quality of a volume image data set,
• 5 a schematic representation of a device for evaluating an image quality of a volume image data set,
• 6th a schematic representation of a computed tomography device having a device for evaluating an image quality of a volume image data set.

1 shows a variant of a method for evaluating an image quality of a volume image data set in a schematic block diagram. The method shown includes the steps of receiving S1 , the first determination S2a or the second determination S2b , of evaluating S3 and spending S4 .

In the step of receiving S1 becomes a volume image data set VBD by means of a first interface ST1 receive. The volume image data set is based on a computed tomography device (CT device) 1 recorded projection measurement data. An image pixel of the three-dimensional matrix of the volume image data set VBD each having an image pixel value is generally also referred to as a voxel. A computed tomography intensity value can be assigned to each image pixel of the three-dimensional volume image data set. For example, the image pixel values are the image pixels of the three-dimensional volume image data set VBD expressed in Hounsfield units. The volume image data set VBD can also be used as a plurality of slice image data sets I. , also marker data sets I. called, which are a two-dimensional representation of the computed tomography intensity values of a slice of the examination area in each case at a slice position along a marked axis, in the case described in particular the system axis 8th (in the following also z-axis).

According to the method shown for evaluating the image quality of the volume image data set VBD becomes in particular at least a part of the thorax of a patient 13 pictured. He can also do the heart 31 of the patient 13 include completely or at least partially. The volume image data set VBD can in particular in the context of a cardiac CT scan of a patient 13 have been recorded.

In the first discovery step S2a is based on the volume image data set VBD and by means of a computing unit RH , a first probability of occurrence of a first artifact caused by the movement of the heart 31 of the patient 13 is caused. In the variant of the method shown, the step of the second determination can alternatively also be used S2b based on the volume image data set VBD and by means of the computing unit RH a second probability of a second artifact occurring in the volume image data set VBD which is determined by the movement of the thorax through the patient's breathing 13 is caused. On the basis of the determined first or second probability, it is determined whether in a representation of the volume image data set VBD the first or the second artifact occurs or is to be expected.

The first or second detection S2a , S2b of the first or second probability, for example, the determination of a parameter or a specific numerical value, for example based on the image pixel values of the volume image data set VBD or the layplan data sets I. of the volume image data set VBD include, wherein the parameter or the numerical value is correlated with the occurrence of the first or second artifact. A correlation or dependency of the numerical value or the characteristic variable with the occurrence of the first or second artifact can be empirically established or examined, for example, in preliminary examinations on volume image data sets that have already been recorded.

When one occurs due to the movement of the heart or due to the breathing of a patient 13 caused artifact, it can happen, for example, that neighboring slice image data sets I. in the direction of the system axis 8th , ie in the direction of the scanning direction or the table feed when recording the projection measurement data, do not match. This can result in the representation of the volume image data set VBD for example, an unexpected, unnatural, step-like offset of anatomical structures in the scanning direction, ie in the direction of the system axis 8th , surrender. On the one hand, this can be caused by the fact that the heart is in a different state, ie a different phase of the cardiac cycle, than in an adjacent slice when recording a slice. Likewise, displacements between slices of the volume image data set caused by breathing can occur if the patient does not completely hold his breath while recording the projection measurement data. The first or second determination can be based on unexpected differences between adjacent slices, ie for example unexpectedly large differences in the image pixel values along the system axis 8th or unexpected shifts of mapped anatomical structures in the volume image data set VBD or between adjacent slice image data sets I. to recognize. That is to say, an above-mentioned parameter or numerical value can refer to differences between adjacent slice image data sets I. in the direction of the system axis 8th based. Specifically, for example, differences between image pixel values adjacent in the direction of the scanning direction can be formed and processed further in order to analyze the occurrence of unexpected, sudden changes in the image pixel values along the system axis. Another possibility of determining the first or second probability can be to use edge segmentation or volume segmentation to detect unexpected shifts in the edges or in the form of anatomical structures in the direction of the system axis in the volume image data set and to use this to detect a value or a parameter which is associated with correlated to the occurrence of the artifact.

The first investigation S2a the first probability of a heart artifact can in particular be based on an image section of the volume image data set VBD or the layplan data sets I. based on what the heart 31 at least partially included. The first determination is preferably based S2a the first probability of a heart artifact in particular on an image section of the volume image data set VBD or the layplan data sets I. based, which is essentially just the heart 31 of the patient 13 or part of the heart 31 includes. As a result, the occurrence of a heart artifact can be analyzed in a particularly suitable manner without the first determination being made without effects from mapped structures outside the heart S2a affect the first probability. This can be done in 1 The method shown for evaluating the image quality also includes the step of first determining S7a a first image area HB based on the volume image data set VBD exhibit. The first image area HB then corresponds to the image section mentioned above. A subsequent first determination S2a the first probability can then be based on the specific first image area HB based.

The second determination S2b the second probability of a breathing artifact can in particular be based on a second image section of the volume image data set VBD or the layplan data sets I. based. This image section includes, for example, the patient's lungs 13 at least in part. To this end, the process sequence can have a second determination step S7b a second image area TB can be determined based on the volume image data set which comprises at least a partial area of the part of the patient's thorax comprised by the volume image data set, and wherein the second determination S2b the second probability on the determined second image area TB based. Preferably, for the second determination S2b the second probability, an image section can be selected which shows a part of the thorax but not the area of the heart 31 of the patient 13 , includes. As a result, an artifact due to a breathing movement can be analyzed particularly advantageously separately from a possibly additionally occurring cardiac artifact.

To illustrate this, in 2 exemplary a certain first image area HB , ie a heart image area, and a certain second image area TB , ie a thorax image area indicated. 2 shows schematically a volume image data set VBD , which, as indicated by the dashed lines, into slice image data sets I. , in the direction of the z-axis, which is the direction of the system axis 8th of the CT machine 1 corresponds, can be divided. The slice image data sets I. a volume image data set consisting of a number of N slice image data sets VBD can also be used in the following description by means of a Index i can be described in an identifiable manner where it is useful for understanding. I _j then describes the i-th slice image data set I. of the volume image data set VBD in the direction of the z-axis, where i = {1, ..., N}.

In the volume image data set shown VBD is the heart 31 of a patient 13 which is from the volume image data set VBD is depicted, outlined. The remaining structures of the volume image dataset VBD included part of the patient's thorax 13 are not included in the illustration for the sake of clarity. Based on the volume image data set VBD is an exemplary first image area HB illustrated, indicated by the illustrated outline of the heart 31 enclosing, three-dimensional box. In addition, is an exemplary second image area TB outlined, represented by the dashed, second, three-dimensional box. In particular, the second image area comprises TB advantageously an area that contains image details outside of the heart 31 includes. On the right side of the 2 is also a two-dimensional slice image data set I. of the volume image data set VBD schematically outlined, which depicts a slice plane perpendicular to the z-axis of the volume image data set as a two-dimensional slice image. The three-dimensional first or second image area HB , TB in the volume image data set VBD is in the slice image data set I. each on a two-dimensional image area HB , TB transferable. The specific first or second, three-dimensional image area is cuboid or rectangular in the case shown. However, other shapes are also conceivable, for example essentially cylindrical or the like.

The first determination S7a of the first image area HB or the second determining S7b of the second image area TB can be selecting an image volume in the three-dimensional volume image data set VBD or a selection of a two-dimensional image region in one or more slice image data sets I. , in which the volume image data set VBD can be divided, include. When you first determine S7a of the first image area HB or at the second determination S7b of the second image area TB A suitable algorithm can be used which, for example, defines a right-angled three-dimensional box, or a box of some other shape, in the volume image data set, which box encloses a specific anatomical area. In this case, a purely automated, ie fully automated, determination of the image area or at least a proposal for a respective image area by means of the computing unit RH conceivable. For example, the heart becomes 31 in the volume image data set or in the slice image data sets I. automatically recognized its dimensions or the occupied volume by means of a segmentation algorithm in the volume image data set VBD or in the slice image data sets I. extracted and based on this the heart image area HB certainly. For example, an area of the lungs or another anatomical structure in the thorax can also be automatically recognized and the second image area based thereon TB to be determined. Likewise, through an interaction of a user in the volume image data set or in a slice image data set, a volume, a maximum extent or corner points of an image section can be determined, based on which the first or second image area is determined.

According to the in 1 The method shown for evaluating the image quality is then based on the ascertained first probability or based on the ascertained second probability, the image quality of the volume image data set VBD rated. The image quality will at least be assessed in terms of whether the occurrence of a first or a second artifact in the representation of the volume image data set VBD is to be expected for a viewer. The assessment of the image quality can also include an expected extent or level of impairment by an artifact occurring in the representation of the volume image data set VBD or for a diagnosis based on the volume image data set VBD to reproduce or estimate. The assessment can include whether the present volume image data set VBD is or is not suitable for a diagnosis based on it. The assessment can include whether an adaptation of a recording condition in the event of a potential renewed recording of projection measurement data can be expected to improve image quality. The assessment can also include other ratings of the image quality.

For example, when determining the first or second probability, a parameter or a specific numerical value is based on the volume image data set VBD by means of the computing unit and in particular automatically calculated which one occurs with the occurrence of the first or the occurrence of the second artifact in the volume image data set VBD correlated. For the evaluation, this parameter or this numerical value is then classified on the basis of previously established comparison values, the comparison values having a degree of impairment of the representation of the volume image data set VBD are linked. The link between the degree of impairment and the comparison values can be established empirically / manually through preliminary examinations using existing volume image data sets and a subjective assessment by experienced clinical staff. The comparison values can be present, for example, in the form of a table stored by means of a memory unit, which can be queried automatically by means of the computing unit and are used for evaluating the image quality based on the first or second probability.

The evaluation can in particular be translated into a control command and / or an instruction. That is, a classification of the image quality can be linked to an instruction and / or a control command.

In the step of spending S4 is then the evaluation using a second interface ST2 issued. In the example shown, the evaluation is carried out in particular to an output unit AE are output, which enables a user, ie clinical staff, to see or access the assessment. This can be implemented in the form of a display unit, ie a display or monitor, or a voice output. Based on this, a further procedure with regard to an optimal diagnosis and care of the patient can be determined. To this extent, for example, a very strong impairment of the image quality of the volume image data set VBD has been evaluated, the user can, for example, specify that projection measures should be taken up again. In addition, the determined first or second probability can also be used for a detailed analysis of the image quality of the volume image data set VBD are issued. The evaluation can also be output to a control unit of a computed tomography device or to a reconstruction unit of a computed tomography device, in particular if the evaluation is linked to an instruction and / or a control command for the CT device.

The method shown for evaluating image quality can also include the steps of recording S11 and reconstructing S12 include, in the step of receiving S11 Projection measurement data of the examination area comprising at least part of the patient's thorax 13 with the help of the CT machine 1 be included. In the step of reconstructing S12 these absorption data or projections collected in this way can be added to the slice image data sets using appropriate reconstruction methods I. or to the volume image data set VBD offset.

The method shown for assessing image quality can also include the steps of deriving S5 an instruction by means of the computing unit RH and spending S6 the handling instruction by means of an interface. In this case, the instruction is automated by means of the computing unit RH certainly. For this purpose, for example, instructions can be stored in a memory unit MU be stored and linked to the assessment or in connection with a degree of impairment, ie a degree of impairment, within the scope of the assessment. The handling instruction is output in particular to an output unit AE which enables a user to access the instruction. In this case, this can be a display unit, for example a display. The instruction corresponds in particular to a proposal or a strategy for a further procedure, so that if possible a volume image data set with a sufficiently high quality can be provided for a further diagnosis or diagnosis based thereon. The strategy based on the evaluation can also include that the volume image data set VBS must be accepted in the present form, since the prospect of an improvement is assessed to be low even if projection measurement data are recorded again.

An instruction which can be included in the described method can be a renewed recording of projection measurement data. A renewed recording can then be particularly useful insofar as the volume image data set VBD is rated as unsatisfactory for a further diagnosis and it is estimated that avoiding the artifact when re-admission is likely. For example, a pure breathing artifact can be avoided if the patient holds his breath completely for the renewed intake. Further operating instructions can be the adjustment of a recording parameter for a new recording of projection measurement data. For example, this can be an improved adjustment of the course of the recordings to the breathing behavior or the cardiac curves of the patient, or the adaptation of a reconstruction parameter for a reconstruction of the volume image data set based on the projection measurement data already recorded, for example with an improved correction algorithm, or based on newly recorded projection measurement data .

The method can be used particularly advantageously when the volume image data set under consideration VBD on during at least two different cardiac cycles of the heart 31 of the patient 13 recorded projection measurement data and as a plurality of slice image data sets I. can be shown which originate from the at least two different cardiac cycles. That means the slice image data sets I. of the volume image data set VBD are based on during a plurality of cardiac cycles recorded projection measurement data. Each of the slice image data sets I. comprises, as also described above, in each case a two-dimensional matrix of image pixel values. The volume image data set VBD can then be described as being composed of a plurality of blocks of slice image data sets, each block being based in each case on the recording of projection measurement data during a cardiac cycle. This means that a first block can originate from a first cardiac cycle c = 1, a second block can originate from a second cardiac cycle c = 2 following the first cardiac cycle, and so on 3 is a schematic of such a volume image data set VBD shown, which in a number of N slice image data sets I. is divided. In the illustration shown are the first four slice image data sets I. with i = 1,2,3,4 assigned to a first cardiac cycle c = 1. The projection measurement data on which these slice image data sets are based were recorded in a first cardiac cycle c = 1. The slice image data sets with i = 5,6,7,8 are assigned to a second cardiac cycle c = 2. The projection measurement data on which these slice image data sets are based were accordingly recorded in a second cardiac cycle c = 2, etc. The volume image data set shown here schematically and only as an example VBD is thus made up of blocks of 4 slice image data sets each I. put together, each block coming from a different cardiac cycle. The cardiac cycles do not have to follow one another directly in time, ie there can be further cardiac cycles between the cardiac cycles during which projection measurement data were recorded. This representation is purely schematic. The number of 4 slice image data records per block was chosen only for reasons of clarity. As a rule, such a block comprises more than four slice image data sets.

For an optimal representation of the heart 31 by means of such a volume image data set VBD is the heart 31 shown in each block in the same phase of the respective cardiac cycle. In particular, this requires good timing of the recording cycles with the patient's heart rhythm 13 advantageous during the recording of the projection measurement data. But especially if the patient's heartbeat is irregular 13 discrepancies can arise here, which lead to the blocks of slice image data records I. cannot be combined to form an optimal representation and artifacts, so-called “stack” artifacts, occur in particular at the boundaries G between the blocks. That is to say, such a heart artifact occurs when the heart is imaged in a slice of the volume image data record in a different state relative to the adjacent slice, ie in a different phase of the cardiac cycle. Such an artifact thus occurs in particular at the boundaries G between blocks of slice image data sets I. from two different heart cycles. Alternatively or in addition, however, a breathing movement of the patient can also be used 13 lead to the fact that the volume image data set is likewise not optimally represented VBD across the limits G is possible.

4th shows a schematic sequence of a variant of the method for evaluating an image quality of a volume image data set VBD , which in a particularly advantageous manner both the step of the first determination S2a the first probability as well as the second determination S2b the second probability includes. Previously, in a step of the first determination S7a of the first image area HB based on the volume image data set VBD and in a step of the second determining S7b the second image area TB based on the volume image data set VBD certainly. The determination of the first probability or the second probability is then based on the determined first image area HB or the specific second image area TB .

The volume image data set VBD in this example there is in particular a volume image data record VBD which occurs during at least two different cardiac cycles of the heart 31 of the patient 13 recorded projection measurement data and as a plurality of slice image data sets I. can be shown which originate from the at least two different cardiac cycles. That is to say, the volume image data record considered in the method variant described below VBD corresponds to one as in 3 schematically depicted volume image data set composed of partial image data sets VBD .

mit i = {1,...,N} und c = {1,...,C} wobei jeder einzelne Schichtbilddatensatz $I_i^c\equiv I_i^c\left(x,y\right)$

jeweils eine zweidimensionale Bildpixelmatrix mit Bildpixelwerten repräsentiert und wobei jeder Bildpixelwert den Computertomographie-Intensitätswert an den OrtsKoordinaten (x,y) widergibt. Jeder Schichtbilddatensatz I der Anzahl an N Schichtbilddatensätzen wäre bereits durch den Index i vollständig beschrieben, so dass ohne Weiteres der Index c im Folgenden, falls nicht zum Verständnis notwendig, ausgespart werden kann.The slice image data sets I. of such a volume image data set VBD , which from a number of N slice image data sets I. which are based on a number of C cardiac cycles can be described identifiable by means of the indices i and c. That is, the volume image data set VBD consists of N slice image data sets ${\mathrm{I.}}_i^c$

with i = {1, ..., N} and c = {1, ..., C} where each individual slice image data set ${\mathrm{I.}}_i^c\equiv {\mathrm{I.}}_i^c\left(x,y\right)$

each represents a two-dimensional image pixel matrix with image pixel values, and each image pixel value reproduces the computed tomography intensity value at the location coordinates (x, y). Each slice image data set I. the number of N slice image data records would already be fully described by the index i, so that the index c can easily be left out in the following, if not necessary for understanding.

wobei davon jeweils nur der erste Schichtbilddatensatz eines jeweiligen Paars an Grenz-Schichtbilddatensätzen BI umfasst ist. Das heißt, eine Grenze G liegt jeweils zwischen einem Paar an Grenz-Schichtbilddatensätzen BI bestehend aus den Schichtbilddatensätzen I_j und I_j+1 vor, wobei der Index j durch obigen Zusammenhang bestimmt ist. Angewendet auf den in 4 schematisch dargestellten Volumenbilddatensatz gilt für den Index j, welcher jeweils herangezogen werden kann, um eine Grenze G, d.h. einen Übergang zwischen zwei Herzzyklen, zu identifizieren: j = {4,8,12}. According to an advantageous embodiment of the method for evaluating the image quality, the first determination S2a the first probability or the second determination S2b a number of boundary slice image data sets BI is determined from the second probability, the number of boundary slice image data sets BI of the subset of slice image data sets I. which corresponds to a boundary G between two in the volume image data set VBD neighboring slice image data sets I. which come from two different cardiac cycles. In the schematic representation of the volume image data set VBD in 3 is the subset of boundary slice image data sets BI of the volume image data set shown VBD highlighted for clarity. The boundary slice image data sets BI each lie at a transition, ie at a boundary G, between slice image data sets I. from two different cardiac cycles. The boundary slice image data records BI always appear in pairs. The boundary slice image data records BI include in particular the subset of slice image data records I. for which applies $${\mathrm{I.}}_j^c\in \left\{\forall \left({\mathrm{I.}}_j^a,{\mathrm{I.}}_{j+1}^b\right)mit\left|a+b\right|=1\right\}$$

only the first slice image data set of a respective pair of boundary slice image data sets BI being included. That is, a boundary G is in each case between a pair of boundary slice image data sets BI consisting of the slice image data sets I _j and I _{j + 1} , the index j being determined by the above relationship. Applied to the in 4th The volume image data set shown schematically applies to the index j, which can be used in each case to identify a boundary G, ie a transition between two cardiac cycles: j = {4, 8, 12}.

According to a preferred embodiment of the method, for each transition, ie for each boundary G, between two in the volume image data set VBD neighboring slice image data sets I. which originate from two different cardiac cycles when first determined S2a the first probability is an individual probability value for the occurrence of the first artifact and for the second determination S2b each of the second probability determines a single probability value for the occurrence of the second artifact. The two in the volume image data set VBD neighboring slice image data sets I. which originate from two mutually different cardiac cycles each correspond to a pair of boundary slice image data sets BI.

wobei in diesem Fall zwei Schwellwerte, bezeichnet als T^α und T^β, auf den Schichtbilddatensatz I_j bzw. dessen Bildpixelwerte angewandt werden, um den dazugehörigen Schwellwertbilddatensatz J_j zu erzeugen. In diesem Fall werden für den Schwellwertbilddatensatz J_j alle Bildpixelwerte des Schichtbilddatensatzes I_j, welche kleiner sind als T^α, durch den Wert von T^α ersetzt. Ebenso werden alle Bildpixelwerte des Schichtbilddatensatzes I_j, welche größer sind als T^β im Schwellwertbilddatensatz J_j durch den Wert von T^β ersetzt. Es können in alternativen Ausführungen auch andere Vorschriften zur Bildung eines Schwellwertbilddatensatzes angewandt werden.According to an advantageous embodiment, for the first determination S2a the first probability and the second determining S2b the second probability threshold image data sets based on the slice image data sets I. educated. At least one threshold image data record is generated for each of the number of boundary slice image data sets BI based on the application of at least one threshold value to the image pixel values of a respective boundary slice image data set BI, the individual probability value determined for each boundary then being based on the threshold image data sets.
Based on a slice image data set I _j , a threshold image data set J _j can be determined by, for example $$J_j=min\left(max\left({\mathrm{I.}}_j,T^{\alpha }\right),T^{\beta }\right)$$

in this case two threshold values, designated T ^α and T ^β , are applied to the slice image data set I _j or its image pixel values in order to generate the associated threshold value image data set J _j . In this case, for the threshold image data set J _j, all image pixel values of the slice image data set I _j which are smaller than T ^{α are} replaced by the value of T ^α . Likewise, all image pixel values of the slice image data set I _j which are greater than T ^β in the threshold value image data set J _{j are} replaced by the value of T ^β . In alternative embodiments, other rules for forming a threshold image data set can also be used.

There can be N threshold image data sets based on the N slice image data sets I. of the volume image data set VBD be generated. Within the scope of the method, it is also possible only for the subset of boundary slice image data sets BI or another subset of slice image data sets of the volume image data set VBD Threshold image data records are formed. In other embodiment variants, however, the first or second probability can also be ascertained without the formation of threshold image data records.

According to an advantageous embodiment of the method, the individual probability value determined for each boundary G is also based on the pixel-by-pixel difference between at least two spatially corresponding subsets of image pixel values of two adjacent boundary slice image data sets BI, ie a pair of boundary slice image data sets BI. Alternatively, the individual probability value determined for each boundary is based on the pixel-by-pixel difference between at least two spatially corresponding subsets of image pixel values of two threshold image data sets belonging to two adjacent boundary slice image data sets BI.

In the exemplary variant of the method carried out here, the locally corresponding subsets in the slice image data sets are each defined by the first image area HB or the determined second image area TB given. Forming a pixel-by-pixel difference then includes, for example, forming a difference between two spatially corresponding image pixel values, ie between two image pixels that have the same spatial coordinates (x, y) within two slice image data sets that are adjacent in the direction of the system axis. The differences can also be quantified in other ways, for example by means of a quotient formation of the image pixel values or the like.

A specific possibility of evaluating the pixel-by-pixel difference between two locally corresponding subsets of image pixel values is described below. For example, in order to determine the individual probability value determined for each limit for the first artifact or for the second artifact, in each case a weighting factor W is determined which is based on the pixel-by-pixel differences. Such a weighting factor W _{j can be} given for a boundary G or G _j between a pair of boundary layer image data sets BI comprising the layer image data sets I _j and I _{j + 1} $$\begin{array}{cc}{\mathrm{W.}}_j\left(k\right)={\displaystyle \sum _{x,y}{\left(max\left(\left|J_{j+k-1}-J_{j+k}\right|-{\mathrm{R.}}_j\mathrm{,\; 0}\right)\times N_j\right)}^2,}& wObeik=1\end{array}$$

In this example, the weighting factor is based on threshold image data sets J formed from the boundary slice image data sets BI. In other embodiments, the slice image data sets corresponding thereto can also be used I. can be used by yourself. In addition, the weighting factor is based on a pixel mask N _j and a value of the image noise R _j .

The pixel mask N _j is based in the example shown, if the first probability is determined, on that in the step of the first determination S7a certain first image area HB or, if the second probability is determined, on that in the step of the second determination S7b certain second image area TB .
For a slice image data set I _j or a threshold value image data set J _j , the pixel mask N _{j can be} based on the heart image area HB for example be defined as $$N_j\left(x,y\right)=\{\begin{array}{rr}\hfill 1& \hfill \left(x,y\right)\in H\mathrm{B.}\\ \hfill 0& \hfill sOnst\end{array}$$

The application of such a pixel mask N _j (x, y) in the above expression of the weighting factor then only includes that subset of image pixel values of a slice image data set I. or its threshold image data set J for the calculation, its image pixel coordinates (x, y) within the first image area HB , ie of the specific heart image area.

A pixel mask for the second determination S2b the second probability based on the chest image area TB can be done in a similar way using the second image area TB be defined. A preferred pixel mask N _j (x, y) for the second determination S2b the second probability can also be defined as $$N_j\left(x,y\right)=\{\begin{array}{cc}1& \left(x,y\right)\in T\mathrm{B.}\cap \neg \left(\left(x,y\right)\in H\mathrm{B.}\right)\\ & \begin{array}{cc}0& sOnst\end{array}\end{array}$$

In this advantageous definition of a pixel mask for determining the second probability based both on the first image area HB as well as the second image area TB In particular, those image pixels are included which are from the thorax image area TB are included, on the other hand those image pixels are excluded, which are also within the heart image area HB lie.

It is conceivable that, within the framework of the method, the cases are also taken into account when there is no explicit first image area HB or second image area TB is defined in a slice image data set or volume image data set. For example, there is no defined thorax image area TB before, the entire slice image viewed can be used as a thorax image area TB , ie as the second image area TB , be determined. For example, there is no explicit heart image area HB before, the heart image area can HB in particular, for example, be determined as empty.

In the concrete example set out above for the calculation of the weighting factor W, which can be used to determine a respective individual probability value, the differences between the threshold image _data sets are also set in relation to the image noise value R _j . In the specific case of the above example of the weighting factor W, the differences are only included in the weighting factor W if the calculated differences between the threshold image data sets J _{j + k-1} and J _{k + k are} greater than the image noise value R _j . In an exemplary embodiment, the value of the image noise can be determined by $${\mathrm{R.}}_j=median\left(\left\{\left|{\mathrm{I.}}_j\left(x,y\right)-{\mathrm{I.}}_j\left(x-\mathrm{1,}y\right)\right||N_j>0\right\}\right)$$

R _j is based on differences between neighboring pixels within a respective boundary slice image data set BI of the boundary under consideration, which of the respective pixel mask N _{j is} based on the first image area HB or the second image area TB , are included. However, other definitions for the value of the image noise R _{j are also} conceivable.

The respective individual probability value for a boundary G _j between a pair of boundary layer image data sets BI having the layer image data sets I _j and I _{j + 1} can then be calculated as a function of the weighting factor W _j (k = 1). The greater the incoming, summed up differences in the image pixel values of the locally corresponding subsets, the greater the respective weighting factor. The value of the weighting factor W can thus provide an indication of the probability of the occurrence of a first artifact or a second artifact, depending on whether a first image area or a second image area is included as an image pixel mask N _j .

According to a further advantageous embodiment, the individual probability value determined for each boundary G is additionally based on a further locally corresponding subset of image pixel values of a slice image data set NI adjacent to one of the two boundary slice image data sets BI, which is not part of the number of boundary slice image data sets BI. In 3 such slice image data sets NI adjacent to a respective boundary slice image data set BI are highlighted as dotted lines by way of example. For this purpose, threshold image data sets J based on these slice image data sets can also be formed, the image pixel values of which are based on the image pixel values of the adjacent slice image data sets NI. The individual probability value determined for each limit G can then additionally be based on a further locally corresponding subset of image pixel values of the threshold image data sets J belonging to the adjacent slice image data sets NI.

For example, the individual probability value is based not only on a pixel-by-pixel difference between at least two spatially corresponding subsets of image pixel values of two adjacent boundary slice image data sets BI or threshold image data sets J formed therefrom, but also on a pixel-by-pixel difference of locally corresponding subsets of image pixel values between one boundary slice image data set BI and one adjacent slice image data record NI, which originates from the same cardiac cycle, or threshold image data records formed therefrom. The differences between two boundary slice image data sets BI which originate from different cardiac cycles can thus be set in relation to differences between two slice image data sets which originate from the same heart cycle. This can advantageously be used as a type of normalization which shows differences occurring at a boundary G or across a boundary between two cardiac cycles in relation to in the volume image data set VBD expected differences between two slice image data sets. Expected differences between two slice image data sets can result, for example, from anatomical differences in the object.

mit einem Skalierungsfaktor S. Der Skalierungsfaktor S kann zuvor anhand von Voruntersuchungen bestimmt sein und in einer Speichereinheit MU abrufbar abgelegt sein.For example, for a boundary G between a pair of boundary layer image data sets BI having the layer image data sets I _j and I _{j + 1, in} addition to a weighting factor W _j (k = 1), a weighting factor W _j (k = 0) or W _j (k = 2) according to the expression given above for the weighting factor W _j . For example, an individual probability value of a respective limit G _j can then be defined as $$P_j=min\left(\mathrm{S.}\times \frac{{\mathrm{W.}}_j\left(1\right)}{max\left({\mathrm{W.}}_j\left(0\right),{\mathrm{W.}}_j\left(2\right)\right)}\mathrm{,1}\right)$$

with a scaling factor S. The scaling factor S can be determined beforehand on the basis of preliminary examinations and in a storage unit MU be stored in a retrievable manner.

In a further advantageous embodiment, during the first determination S2a the first probability or the second determination S2b the second probability is an overall probability value for the volume image data set VBD is determined based on the individual probability value determined for each limit G.

One possibility of an overall probability value GP based on the respective individual probability values can be defined as $$GP=\text{Max}\left(\left\{P_j|{\mathrm{F.}}_j>Q\right\}\right)$$

mit der jeweiligen Pixelmaske N_j basierend auf dem ersten bzw. zweiten Bildbereich.F _{j represents} the portion of the image pixels of the two-dimensional matrix of image pixel values of the j-th slice image data set, which is used as a subset in the determination of the individual probability values. The factor F _{j of} a j-th slice image data set can be defined as $${\mathrm{F.}}_j=\frac{{\displaystyle {\sum }_{x,y}{N}_j}}{{\displaystyle {\sum }_{x,\mathrm{<figure-callout\; id="1"\; label="y"\; filenames="DE102019202830A1\_0014.png"\; state="\{\{state\}\}">y</figure-callout>}}1}}$$

with the respective pixel mask N _j based on the first or second image area.

The factor Q in the possibility of determining the overall probability value GP specifically explained above is a previously established value which has proven to be useful in preliminary investigations, for example experimentally, and in a storage unit MU can be retrieved.

Other calculations of an overall probability value are also conceivable. For example, an averaging of the individual probability values or a weighted or unweighted summation of the individual probability values or the like can be used to obtain a global overall probability value of the volume image data set VBD to determine.

In the in 4th In the method illustrated by way of example, both the first probability of the occurrence of a heart artifact and the second probability of the occurrence of a respiratory artifact are determined. The first determination of the first probability and the second determination of the second probability can essentially be carried out in the same way, for example by means of the above-described relationships for the weighting factor W, the individual probability value P or the overall probability value GP, but with threshold values T. , defined values S, Q and the specific image areas or pixel masks N _j based thereon for determining the first probability and determining the second probability.

In the in 4th The method illustrated by way of example is then assessed based on both the determined first probability and the determined second probability. The assessment can then include the impairment of the volume image data set VBD to be judged by a first and a second artifact. The assessment can then include both the extent of a heart artifact and the extent of a respiratory artifact in the volume image data set VBD to be classified according to impairment level.

5 shows a schematic embodiment of a device for evaluating an image quality of a volume image data set depicting at least a part of the thorax of a patient. The device has a first interface ST1 on. The first interface ST1 is designed to receive the volume image data set. The volume image data set is based on projection measurement data recorded by means of a computed tomography device and comprises a three-dimensional matrix of image pixel values. The first interface ST1 can use the volume image data set VBD for example from a reconstruction unit received which, based on the projection measurement data, has reconstructed the volume image data set by means of a suitable algorithm, for example the filtered back projection.
The device SYS also has a computing unit RH which is designed to first determine based on the volume image data set VBD , a first probability of occurrence of a first artifact caused by the movement of the heart 31 of the patient 13 is caused and / or is designed for the second determination, based on the volume image data set VBD , a second probability of occurrence of a second artifact in the volume image data set VBD which is caused by the movement of the thorax through the patient's breathing 13 is caused.
The arithmetic unit RH is also designed to evaluate the image quality of the volume image data set VBD based on the determined first probability and / or based on the second probability.
The device SYS furthermore has a second interface ST2 which is designed to output the evaluation of the image quality.

The device shown here SYS is designed in particular to carry out the method for evaluating the image quality and its aspects, for example in the form of the exemplary embodiments described, by the first and second interfaces ST1 , ST2 and the arithmetic unit RH are designed to carry out the respective steps of the method.

At the device SYS it can in particular be a computer, a microcontroller or an integrated circuit. Alternatively it can be with the device SYS be a real or virtual network of computers (an English technical term for a real network is "cluster", an English technical term for a virtual network is "cloud"). A unit of account RH can have hardware elements or software elements, for example a microprocessor or a so-called FPGA (English acronym for “Field Programmable Gate Array”).

In the 5 shown device SYS also has a storage unit MU on. This can be implemented as a non-permanent main memory (Random Access Memory, RAM for short) or as permanent mass storage (hard disk, USB stick, SD card, solid state disk).

At an interface ST1 , ST2 it can be a hardware or software interface (e.g. PCI bus, USB or Firewire).

Optimally, the device SYS furthermore comprise an input and output unit, an input and output unit at least one input unit EE and / or at least one output unit AE includes. By means of the output unit AE in the form of a representation unit, for example a display, a determined probability, an assessment, can be displayed for a user. An input unit enables, for example, the manual interaction of a user, for example the confirmation of a specific first or second image area HB , TB .

The device can also be used with a CT machine 1 be connected. The device SYS can be connected to the CT device via a network 1 be connected. In an alternative embodiment, the device can also be used as part of the CT device 1 be trained. The device can be designed to generate a control signal for the CT device optionally based on the evaluation.

A network can be a local network (a technical term is "Local Area Network", "LAN" for short) or a large-scale network (a technical term is "Wide Area Network", "WAN" for short). An example of a local network is an intranet, an example of a large-scale network is the Internet. The network can in particular also be implemented wirelessly, in particular as a WLAN (for “wireless LAN”, in English the abbreviation “WiFi” is common) or as a Bluetooth connection. The network can also be designed as a combination of the examples mentioned. Furthermore, the communication between the device SYS and a CT machine 1 can also be done offline, for example by exchanging data carriers.

The 6th shows a computed tomography device 1 with an x-ray tube 26th . The CT machine 1 has a gantry 20th , a tunnel-shaped opening 9 and a storage device 10 on. The gantry 20th has the support frame 21st and the slewing ring 24 on. The slewing ring 24 is on the support frame by means of a pivot bearing device 21st around the system axis 8th relative to the support frame 21st rotatably arranged.
Into the tunnel-shaped opening 9 is the patient 13 insertable. In the tunnel-shaped opening 9 is an area to be mapped, especially the thorx, of the patient 13 positionable in such a way that the X-rays 27 from the X-ray tube 26th to the area to be imaged and, after an interaction with the area to be imaged, to the X-ray detector 28 can get. The storage device 10 indicates the storage base 11 and the storage plate 12 for positioning the patient 13 on. The storage base 11 and the storage plate 12 can also be called a patient table. The storage plate 12 is such relative to the storage pedestal 11 movable on the storage pedestal 11 arranged that the storage plate 12 in a longitudinal direction of the storage plate 12 , in particular essentially along the system axis 8th into the tunnel-shaped opening 9 is insertable, ie the advance of the patient along the system axis 8th is made possible.

The CT machine 1 is for recording projection measurement data based on X-rays 27 educated. The X-ray tube 26th is on the turntable 24 arranged and for the emission of X-rays 27 educated. The X-ray detector 28 is for the detection of X-rays 27 educated. The X-rays 27 can from the x-ray tube 26th to the area of the object to be mapped 13 arrive and after an interaction with the area to be imaged on the X-ray detector 28 hit. In this way, measurement data of the area can be acquired in the form of projection measurement data.

The computed tomography device comprises a control unit SE, which is designed to control the recording of the projection measurement data. The computed tomography device 1 also comprises a device according to the invention SYS for evaluating an image quality of a volume image data set VBD imaging at least part of the patient's thorax 13 . The device SYS has a reconstruction unit RK. The reconstruction unit RK is designed to use the three-dimensional volume image data set based on projection measurement data VBD or the slice image data sets I. to create. Alternatively, it is also possible that the reconstruction unit is separate from the device SYS are designed to evaluate the image quality. The computed tomography device 1 also has a storage unit MU and an output unit AE in the form of a display unit DE. The display unit DE can in particular be designed to display the volume image data set, a slice image data set, the assessment and / or the determined first or second probability or to display it for a user.

Although the invention has been illustrated in more detail by the exemplary embodiments, the invention is not restricted by the disclosed examples and other variations can be derived therefrom by the person skilled in the art without departing from the scope of protection of the invention.

Claims (18)

A method for evaluating the image quality of a volume image data set (VBD) depicting at least part of the thorax of a patient (13), the volume image data set (VBD) being based on projection measurement data recorded by means of a computed tomography device (1) and comprising a three-dimensional matrix of image pixel values, having the steps - Receiving (S1) the volume image data set (VBD) by means of a first interface (ST1), - First determination (S2a) of a first probability of occurrence of a first artifact, which is caused by the movement of the heart (31) of the patient (13), by means of a computing unit (RH) and based on the volume image data set (VBD), and / or Second determination (S2b) of a second probability of occurrence of a second artifact in the volume image data set (VBD), which is caused by the movement of the thorax by the breathing of the patient (13), by means of the computing unit (RH) and based on the volume image data set (VBD) ), - Assessing (S3) the image quality of the volume image data set (VBD) by means of the computing unit (RH) based on the first probability determined and / or based on the second probability, and - Output (S4) of the assessment of the image quality by means of a second interface (ST2). Procedure according to Claim 1 , also having the steps - deriving (S5) an instruction based on the assessment of the image quality by means of the computing unit (RH), and - further outputting (S6) the instruction by means of an interface. Procedure according to Claim 2 , wherein the specific handling instruction comprises at least one instruction of the group of handling instructions: - renewed recording of projection measurement data, - adjustment of a recording parameter for a new recording of projection measurement data, or - Adaptation of a reconstruction parameter for a reconstruction of a further volume image data set based on the projection measurement data already recorded or based on projection measurement data recorded again. Method according to one of the Claims 1 to 3 , comprising the step of determining (S2a) the first probability and also having the step - first determining (S7a) a first image area (HB) based on the volume image data set (VBD), which contains at least a partial area of the heart (31) of the patient (13 ), and wherein the determination of the first probability is based on the determined first image area (HB), and / or comprising the step of determining (S2b) the second probability and also having the step - second determination (S7b) of a second image area (TB) ) based on the volume image data set (VBD), which comprises at least a partial area of the part of the thorax of the patient (13) comprised by the volume image data set (VBD), and wherein the determination of the second probability is based on the determined second image area (TB). Method according to one of the Claims 1 to 4th , wherein the volume image data set (VBD) is based on projection measurement data recorded during at least two different cardiac cycles of the heart (31) of the patient (13) and can be displayed as a plurality of slice image data sets (I) which originate from the at least two different cardiac cycles, and which each comprise a two-dimensional matrix of image pixel values. Procedure according to Claim 5 , wherein a number of boundary slice image data sets (BI) is determined in the first determination (S2a) of the first probability or in the second determination (S2b) of the second probability, the number of boundary slice image data records (BI) of the subset of slice image data records (I) which each adjoin a boundary (G) between two slice image data sets (I) which are adjacent in the volume image data set (VBD) and which originate from different cardiac cycles. Procedure according to Claim 6 , with the first determination (S2a) of the first probability or with the second determination (S2b) of the second probability one individual for each boundary (G) between two adjacent slice image data sets (I) in the volume image data set (VBD), which originate from different cardiac cycles The probability value for the occurrence of the first artifact or for the occurrence of the second artifact is determined, and the individual probability value determined for each border (G) is based on two border slice image data sets (BI) adjacent in the volume image data set (VBD). Procedure according to Claim 7 , wherein at least for each of the number of boundary slice image data sets (BI) a threshold image data record is generated based on applying at least one threshold value to the image pixel values of a respective boundary slice image data set (BI), and the individual probability value determined for each boundary (G) based on the threshold image data sets. Method according to one of the Claims 7 or 8th The individual probability value determined for each boundary (G) is based on the pixel-wise difference between at least two locally corresponding subsets of image pixel values of two adjacent boundary slice image data sets (BI), or - on the pixel-by-pixel difference between at least two spatially corresponding subsets of image pixel values two threshold image data sets belonging to two adjacent boundary slice image data sets (BI). Procedure according to Claim 9 , wherein the individual probability value determined for each boundary (G) is additionally based on a further locally corresponding subset of image pixel values of a slice image data set (NI) adjacent to one of the two boundary slice image data sets (BI) which is not part of the number of boundary slice image data sets ( BI) is. Procedure according to Claim 9 or 10 , comprising the step of first determining (S7a) the first image area (HB) or comprising the step of second determining (S7b) the second image area (TB) based on the volume image data set (VBD), and wherein the respective subsets of image pixel values by the first Image area (HB) or by the second image area (TB) are determined. Method according to one of the Claims 7 to 11 wherein each individual probability value is based on a value of the image noise of at least one of the boundary slice image data sets (BI) on which the respective individual probability value is based. Method according to one of the Claims 7 to 12 In the first determination (S2a) of the first probability or in the second determination (S2b) of the second probability, an overall probability value for the volume image data set (VBD) is determined based on the individual probability value determined for each limit (G). Device (SYS) for evaluating an image quality of a volume image data set (VBD) depicting at least part of the thorax of a patient (13), which is based on projection measurement data (PM) recorded by means of a computed tomography device (1) and comprises a three-dimensional matrix of image pixel values - A first interface (ST1), designed to receive the volume image data set (VBD), - A computing unit (RH) designed to first determine, based on the volume image data set (VBD), a first probability of the occurrence of a first artifact that is caused by the movement of the heart (31) of the patient (13), and / or is designed for the second determination, based on the volume image data set (VBD), a second probability of the occurrence of a second artifact in the volume image data set (VBD), which is caused by the movement of the thorax through the breathing of the patient (13), - wherein the computing unit is further designed to evaluate the image quality of the volume image data set (VBD) based on the determined first probability and / or based on the second probability, and - A second interface (ST2) designed to output the assessment of the image quality. Device (SYS) for evaluating an image quality according to Claim 14 also trained a method of Claims 1 to 13 execute Computed tomography device (1) comprising a device (SYS) for evaluating an image quality of a volume image data set Claim 14 or 15th . Computer program product with a computer program which can be loaded directly into a memory (MU) of a device (SYS) for evaluating an image quality, with program sections to carry out all steps of the method (S) for evaluating an image quality according to one of the Claims 1 to 13 when the program sections are executed by the device (SYS) for evaluating an image quality. Computer-readable storage medium on which program sections readable and executable by a device (SYS) for evaluating an image quality are stored, in order to carry out all steps of the method (S) for evaluating an image quality according to one of the Claims 1 to 13 when the program sections are executed by the device (SYS) for evaluating an image quality.

Images