GB2545641A - A method for detecting motion in a series of image data frames, and providing a corresponding warning to a user - Google Patents

Abstract

Method for detecting movement in a sequence of image data frames, and providing a corresponding warning to a user, comprising the steps of: a) Monitoring motion of the image data within the image frames; b) Extracting properties of motion of the image data within the image frames, c) Notifying a user of excessive motion within the image frames; d) Notifying the user of specific patterns of motion within the image frames; and e) Recommending subsequent quality control actions regarding the detected motion properties. The method may also comprise defining a region of interest (ROI) within the image frames and monitoring and extracting properties of motion of the ROI. Motion may be detected by registering each frame to a reference frame or to a preceding or successive frame and both relative and absolute motion may be determined. The patterns of motion may comprise time periods when motion occurs and periods with no motion, statistics of the distribution of motion or gradual drift motion.

Description

A METHOD FOR DETECTING MOTION IN A SERIES OF IMAGE DATA FRAMES, AND PROVIDING A CORRESPONDING WARNING TO A USER

The present invention provides methods for data filtering and alerting a user to defects in data sets.

In particular, it relates to methods for detecting motion of a region of interest (ROI) in image data sets and alerting a user when the detected motion falls outside of a defined range .

The invention more particularly relates to the detection of motion of a region of interest (ROI) between images in a medical imaging data set.

Definitions, Acronyms, and Abbreviations Region of Interest ROI

Positron Emission Tomography PET

Computed Tomography CT

Magnetic Resonance MR

Single photon emission computed tomography SPECT Nuclear Medicine NM

References [1] T. Oakes, T. Johnstone, K. Ores Walsh, L. Greischar, A. Alexander, A. Fox and R. Davidson, "Comparison of fMRI motion correction software tools," Neuroimage, vol. 28, 2005.

[2] W. T. C. f. Neuroimaging, "SPM - Statistical Parametric

Mapping," 2014. [Online]. Available: http://www.fil.ion.ucl.ac.uk/spm/. [Accessed 2014].

[3] R. Goebel, "Brain Innovation - Home," 2014. [Online]. Available: http://www.brainvoyager.com/. [Accessed 2014].

[4] F. -. FMRIB, "MeFLIRT - FslWiki," 2014. [Online].

Available : http ://fsi.fmrib. ox .ac.uk/fsi/fslwiki/MCFLIRT.

[Accessed 2014] .

[5] T. Kohlberger, M. Sofka, J. Zhang, N. Birckbeck, J. Wetzl, J. Kaftan, J. Declerck and S. Zhou, "Automatic multiorgan segmentation using learning-based segmentation and level set optimization," in Medical Image Computing and Computer-Assisted Intervention (MICCAI), 2011.

[6] T. Kohlberger, M. Sofka, J. Wetzl, J. Zhang, N. Birkbeck, S. Zhou, J. Kaftan and J. Declerck, "Method and System for Multi-Organ Segmentation Using Learning-Based Segmentation and Level Set Optimization". Patent US20120230572A1, September 2012.

Introduction

Medical imaging involves acquiring a set of image frames of a subject over a period of time. During this acquisition, subject movement, cardiac and respiratory motion often introduce blurriness or potentially misleading image artefacts into an image frame set.

There exist methods and algorithms for detection of motion between frames of data in medical image data processing systems .

For the purpose of quality control and for better patient management, users such as clinicians and imaging technologists would like to know: - whether there is motion in the image data frame set or not; - whether the motion can be identified and corrected, or a reacquisition of a new set of image data frames is needed; - when motion occurred and what the duration was. If motion just happened in one or a few frames of the image frame set, the data may be effectively used by excluding the affected frame(s) from the image data frame set.

If the motion happened in a frame with long duration, that is to say, data capture for that frame took place over a relatively long time, the data of the frame may be separated into several frames each of shorter duration to enable identification of the time of the motion more accurately.

It would be advantageous if image data analysis could be carried out in real time, or near-real time, such that algorithm based correction may be attempted while the patient is on site, and if such algorithm based correction still results in unsatisfactory scan quality, the patient could be re-scanned before they leave the site which would avoid the financial, logistical and radiation overhead associated with repeating the image data capture scan at a later date.

Certain known arrangements exist for detecting motion between frames of a data frame set. Several software packages currently exist to detect and correct motion in multi-timepoint images [1], [2], [3], [4].

McFLIRT [4] plots three translation and three rotation parameters and an additional 'mean displacement' parameter on a graph which is displayed to a user. Typical output data is shown in Fig. 2. The "mean displacement" parameter is based on a spherical model to represent the ROI, for instance the brain. A magnitude of the motion of the model is computed. An explanation of this system may be retrieved from: http://www.fmrib. ox.ac.uk/analysis/techrep/tr99mjl/tr99mj1/no de3.html

The McFLIRT software package estimates rigid motion only. The motion is estimated from the whole image frame, not a specific ROI within the data frame. The mean displacement curve plotted in McFLIRT is computed from a spherical model, not an image specific ROI region.

The output, expressed in terms of multiple displayed translation and rotation motion curves may be too complicated for general imaging quality control purposes, where a user is under pressure to make rapid decisions on whether acquired data frame sets are of acceptable quality.

However, the user, typically a clinician, is concerned as to whether motion has occurred in a region of interest (ROI) . Such ROI may for example represent an organ (e.g. brain), a lesion, or a structure (e.g. pons). The user is concerned as to the amplitude of such motion; and the consequence of the motion on the interpretation of the image. The image quality in the presence of motion is a concern for clinical diagnosis and quantification. A method of monitoring the motion and detecting the time of occurrence, the magnitude and characteristics of motion in the ROI is needed by both the physicians and technologists for the quality control purpose of an imaging study.

Conventional motion correction tools lack the functionality to assess and highlight excess or specific patterns of motion in dynamic medical image acquisition. Conventionally, curves representing raw motion data in dynamic medical image acquisition are presented to the user and the analysis of these motion curves is left to the user. No additional information is conventionally extracted and the interpretation of the results is left to the user.

The present invention accordingly provides methods and apparatus as defined in the appended claims.

The above, and further, objects, characteristics and advantages of the present invention will become more apparent from the following description of certain embodiments thereof, in conjunction with the appended drawings, wherein:

Fig. 1 shows graphs of data representing estimated translation and rotation parameters and estimated mean displacement over time, as provided by the McFLIRT program; Fig. 2 shows an outline flow chart of a method of the present invention;

Fig. 3 illustrates examples of ROIs in the brain as extracted by automatic segmentation;

Fig. 4 illustrates an example of a ROI defined on an organ, here a kidney, by semi-automatic segmentation;

Fig. 5 illustrates examples of motion-correction workflows; Fig. 6 shows an example of a graph showing estimated mean motion for a ROI region, illustrating measurement of relative motion to adjacent frames and absolute motion as compared to a baseline frame;

Fig. 7 shows an explanatory example of a gradual drift motion;

Fig. 8A shows an example of estimated motion in a ROI in chronological order;

Fig. 8B shows an example of estimated motion in a ROI in order of magnitude of detected motion; and

Fig. 9 schematically represents a computer-implemented embodiment of the present invention.

The present invention provides methods and apparatus for detecting motion of one or more regions of interest (ROI) in image data frame sets representing sequential image data frames, warning a user if the detected motion exceeds a threshold of acceptable motion , or if unexpected patterns of motion are detected, such as drift motion, and/or filtering out image data frames with unacceptable levels of motion.

To monitor the motion in the image data, there are typically two different approaches available.

The first approach is to rely on external motion tracking devices for the detection and monitoring of motion, for instance at the surface of a body region of a patient. That is to say, monitoring motion of the imaged article -typically a patient.

Another possibility is to detect motion in the captured image frames by analysis of the image data in those frames. This may be achieved, for example, by use of an image-registration-based method. Alternatively, a specific imaging protocol may be used which enables motion detection, such as navigator sequences in MRI.

The present invention provides a method to detect and analyse motion from a dynamic sequence of an image data frame set in order to assess the quality of a set of image data frames, in respect of movement of the ROI within the image data frame set in either rigid or non-rigid fashion.

The detected rigid motion is often represented as translation and rotation along one or a few axes. It can be represented at specific time points or between the frames of multi-frame images, e.g. dynamic images over time or a sequence of gated images such as cardiac or respiratory gated images. The common parameters of rigid motion are three translations in the conventional X, Y, Z directions, and three corresponding rotations. Fig. 1 shows such representations.

In the case of the non-rigid motion, it can be represented by a deformation field, as will be familiar to those skilled in the art.

The multiple motion parameters illustrated in Fig. 1 make it difficult to assess the amplitude, duration and time of occurrence of the motion, and are not easily usable for determining whether a captured image data frame set is of useful quality for evaluation.

The invention may further provide an analysis of detected motion, which may assist a user in determining whether the image data set can be used in its entirety, or whether it can be used after filtering, or whether it is of insufficient quality for use, and must be reacquired, for example.

In an example, illustrated in the outline flow chart of Fig. 2, the method of the present invention may comprise:

At step 82, determining a magnitude of motion of image data representing a specific ROI within image data frames. This may be achieved, for example, by calculating a mean motion measurement of the ROI in each image data frame using either an image based motion correction algorithm or other internal and external motion monitoring methods. The mean motion may be determined by calculating the motion of each voxel within the ROI and taking an average of those values. Preferably in such calculations, the magnitude of the motion is considered: so that all motions are regarded as positive, and total motion is considered, regardless of its direction.

At step 84, the mean motion measurement of the ROI in an image data frame may be calculated relative to a reference frame which may be a baseline frame or a respective time-adjacent frame, either respective preceding or following frames. By calculating relative to a baseline frame, a value of absolute motion will be determined. By calculating relative to a time-adjacent frame, relative motion will be determined.

At step 86, the detected motion within the data frames may be analysed to identify particular patterns of movement, such as a sudden shift or a gradual drifting, and/or to detect motion above a specified threshold.

At step 88, if any patterns of movement are detected, or an amount of movement is detected which reduces the quality of the image data frame set for evaluation, an output may be provided at step 90 to a user in order to warn and assist the user to recognise the motion and its patterns in the image data and to make an appropriate decision, for example to remove image data frames which include excessive average motion measurement of the ROI; to employ a motion-correction algorithm to reduce the effect of the motion in the image data frame or to reacquire the image data. Alternatively, at step 92, an output may be provided to a user to confirm that no action is required.

The output may be in graphical, text, or audible form, or indeed any appropriate form. The outputs may suggest to the user one of the following, as examples: - no action required as motion is not clinically significant - reconstruct the image with motion correction algorithm use only motion-free period of the acquisition to reconstruct the image - arrange a rescan in the worst case scenario.

The present invention provides estimation and display of ROI-specific absolute motion with recognised patterns of motion, which the inventors believe unavailable in any conventional motion detection and correction tools for medical imaging.

Description of the workflow

More explanation, and some specific examples, will be provided below.

At least one region of interest (ROI) must be defined in each frame of image data. The ROI may be an object in the image, or a particular organ/lesion/region. Automatic or manual segmentation algorithms could be used on either single or multi-frame images to define the ROI. A variety of segmentation algorithms could be used. One example is template image registration, where a template image (e.g. AAL brain atlas) is registered with the image and standard ROI regions in the template are transformed into the image. Such an algorithm may be used to define ROIs such as illustrated in Fig. 3. Example algorithms for automatic organ segmentation are described in references [5], [6]. Semi- automated or manual segmentation may alternatively be employed, and an example segmentation resulting from such an algorithm is shown in Fig. 4.

In the present invention, determination of motion of the image data is preferably confined to consideration of the region of interest (ROI) in the image data. Some parts for a patient's body will be subject to different movement than the body as a whole. For example, a ROI representing the heart will show significant movement even though the body as a whole does not move. An automatic or manual ROI segmentation is applied to each of the image data frames to identify the ROI. Fig. 3 illustrates three examples of ROIs defined in a brain image, by automated segmentation. The segmentations defined are, respectively, the whole brain; the cerebellum; the pons. Fig. 4 illustrates an ROI representing an organ, here a kidney.

In the motion magnitude determining step 82, the present invention may be applied to motion in the image data frames which has been detected by any suitable method. For example, the motion may be detected by internal and external motion monitoring methods used during the imaging scan. Internal monitoring methods provide detection of motion by analysis of movement of an ROI region within the image data, while external monitoring methods provide detection of motion by analysis of movement of the subject of the imaging, typically a patient.

Alternatively, in embodiments of the present invention, the motion may be estimated from multi-frame image data such as dynamic or gated CT, MR, PET, SPECT, NM images data sets. A multi-time-frame dynamic image may be generated from the raw imaging data by ordering temporally-adjacent frames of image data.

In the step 84 of calculating mean motion relative to a reference frame, absolute motion of the ROI within the image data may be determined with reference to a baseline image, and relative motion may be determined for each frame by reference to a respective adjacent frame.

Fig. 5 schematically shows methods for motion detection according to the present invention. In group (1), each motion is to the baseline frame, frame 1. This will produce a measure of absolute motion, such as shown by curve 62 in Fig. 6. In group (2), each motion is to the respective adjacent image frame. This may be the preceding frame, the following frame or some combination of the two. The result will indicate relative motion, such as shown by curve 64 in Fig. 6.

Motion of the ROI image between frames may be calculated as the detected motion of the ROI in each frame. The detected motion could be rigid, affine or deformable and may be represented by a transformation matrix or deformation field over time points. That motion may be calculated as translation and rotation in three orthogonal axes, and may be combined in a suitable manner to provide a value for overall motion of the ROI between frames of data.

Within step 84, based on the defined ROI and the motion of that ROI in the respective image frames, a measure of the magnitude of the mean motion in the ROI over time is computed. In the case of rigid or affine registration, this measure could be the transform distance TD between two frames, defined below:

Where: | Ω | is the number of voxels in the ROI, Tt is the transformation matrix applied on frame i, and Tj is the final transformation matrix applied on frame j. To compute the motion to the baseline frame i, Γ, is set to the identity matrix. 7) is the transform matrix between frame j and i.

For deformable registration, the mean of the magnitude of the ROI deformation field could be used as the measure.

Analysis of detected motion is performed in step 86. This may include analysis of the absolute and relative motion of the ROI in the image data frames, to detect image capture periods which contain significant motion or are motion-free, or to detect specific motion patterns, such as gradual drifting. A motion threshold, defining an acceptable degree of mean motion may be predefined, or user-defined in real time. Alternatively, a machine-learning arrangement may be used, for example to store and evaluate user-defined thresholds such that an automated recommendation of threshold value may be produced for future threshold determination steps.

Fig. 6, which will be described in more detail below, illustrates examples of absolute and relative motion, using the calculated mean motion of the ROI in each image data frame. Curve 62 represents absolute motion with reference to a baseline image of frame 1. Curve 64 represents relative motion in each frame with respect to preceding or successive frames, preferably immediately adjacent preceding or successive frames. While most frames are determined to be motion-free, movement above threshold level 66 is detected in frames 4 and 10. The relative motion shown by curve 64 shows the motion occurring only in frames 4 and 10, and motion being zero in other frames, while curve 62 illustrates the cumulative effect of both motions. In this illustration, a threshold value 66 for an acceptable amount of motion has been set at 2mm. The example frames of image data represented in Fig. 6 were captured at ten-second intervals.

Preferably, an automatic analysis is performed with the threshold, from the absolute and/or relative motion detected as described above. For multi-frame imaging, an automatic motion correction algorithm may be used to detect the motion in the ROI image data and to correct for the motion.

With detected motion such as shown in Fig. 6, the frames 4, 10 where motion exceeded the threshold value may be excluded from consideration, while the remaining frames may make up a useable image data set. By removing frames such as 4, 10 where motion is detected, motion-free periods may be defined. The user may be informed of this possibility in step 90 of Fig. 2, or the frames 4, 10 may be excluded automatically.

Detection of specific patterns of movement is carried out in step 88. One specific pattern which may be detected is drift motion. Fig. 7 illustrates the effect of a drift motion: where a small motion is detected between one frame and the next, but the cumulative motion becomes significant, using the calculated mean motion of the ROI in each image data frame. Bars 72 indicate the relative motion between one data frame and the adjacent one, while bars 74 indicate the absolute motion with respect to the reference frame, frame 1. Although the motion between each frame and the next is relatively small, the cumulative motion represented by bars 74 rapidly exceeds the acceptable motion threshold 66.

With detected motion such as shown in Fig. 7, a drift motion may be detected, and a correction algorithm, conventional in itself, may be applied to remove the effects of the drift.

At alternative steps 90 and 92, an output may be provided to the user to warn of defects in the captured image data set and assist in resolving associated problems (step 90), or the user may be informed that no significant defects have been identified and that no action is required (step 92).

Such outputs may be provided to the user in a number of formats. For example, several graphs may be displayed to monitor motion, such as: 1) The magnitude of motion and the time of occurrences 2) Magnitude and duration of the motion in the scan 3) Indication of motion-free intervals 4) Specific motion patterns detected.

The results may be presented as line plots such as shown in Fig. 6 or bar charts such as shown in Figs. 7 and 8A, 8B.

The output to the user may comprise multiple graphs containing different motion information and highlights.

The graphs of Figs. 8A and 8B show examples of magnitude and duration of motion during an image acquisition process. In the example shown by Fig. 8A, the motion is characteristic of respiratory motion when it is plotted in chronological order, and such graphical representation may be presented to the user to enable the user to decide whether the data is of sufficient quality for use. When it is sorted by magnitude, as in Fig. 8B, the user can have a better understanding of the distribution of the motion, for example half of the frames are above 1mm motion.

The following is an example of a threshold-based method for detecting motion in an image data capture period. A motion-free period is detected when the relative motion of each frame in the time period is below a threshold and the difference of absolute motion of the frames in the zone is also below a threshold. This is illustrated in Fig. 6. Gradual drift motion is characterized by a small amount of motion at each frame, but resulting in an overall significant motion drift, and an example of such motion is shown in Fig. 7 .

By using both motion curves, motion due to gradual drift can be excluded, for example by detecting a gradual but continuous increase in total movement as in Fig. 7.

Pattern recognition methods with machine learning technique could also be used to define the threshold 66 for detection of motion between frames. For example, after a user has manually entered a threshold value on several occasions, the present invention may calculate a suitable threshold based on these earlier thresholds, for a similar ROI.

The invention may notify the user of detected movement patterns such as :

Detection of absolute (cumulative) motion as compared to the baseline frame 1 over time exceeding a defined threshold. Image frames with a detected level of motion in excess of the defined threshold are highlighted, along with the time of occurrence and the duration of the corresponding motion, as shown in Fig. 6.

The relative motion as compared to an adjacent time point may exceed the defined threshold. The user may then decide to use only those periods found to be free of motion. Fig. 6 shows an example with motion occurring only twice, during the frame acquisitions at frames 4 and 10.

The results of the analysis may be presented in a graphical format, or a a non-graphical format, such as a text output or an audible output.

An analysis may be automatically carried out and, depending on any specific motion pattern detected, different indications or alarms could be presented to the user in step 90, such as: indicate whether defined quality control criteria are met, depending on whether any detected motion is within predefined limits. suggest using only an identified motion-free phase of the data, which may be possible in case some frames present an unacceptable level of movement yet sufficient counts with acceptable levels of movement are available to provide acceptable image quality for analysis. suggest incorporating a motion correction algorithm in to one or more frames with unacceptable levels of movement. suggest a re-scan to re-capture image data frames if a clinically-acceptable dataset cannot be extracted from the captured data.

In the situation illustrated in Fig. 6, the user could take actions such as: - Split the frames 4 and 10 into acquisitions of shorter time frames and reconstruct, so that the moment when the motion happens, can be more precisely identified. - Remove the frames with motion. From consideration of Fig. 6, the dataset is free of motion apart from frames 4 and 10, so the remaining dataset, excluding frames 4 and 10 may be used for analysis purposes.

In a particularly useful feature of embodiments of the present invention, analysis of the motion present in the image frames may be performed rapidly, for example, while a patient is still present at the imaging location such as a hospital. As a result, if the captured image data is of such poor quality that it cannot be used, the patient may be reimaged before leaving the imaging location. This provides financial, logistical and radiation overhead savings as opposed to having to call the patient back another day for re-imaging.

The invention has been particularly discussed with reference to detection of motion of ROI in image frames, but may be applied to the detection of motion of complete image frames. The invention contributes the detection and classification of movement in such embodiments.

The present invention accordingly provides a method to monitor motion within a multi-frame image data and quantify the mean magnitude of the motion. The invention also enables extraction of properties of the motion over the multiple frames. The invention preferably then informs a user of excessive or specific patterns of motion, and may recommend subsequent quality control actions regarding the detected motion properties. The method of the present invention may comprise the steps of: generating a multi-time-frame dynamic image from the raw imaging data; such as a dynamic medical CT, MRI, PET, SPECT, ultrasound image dataset; segmenting an ROI in multi-frame image dataset, monitoring the motion in multi-time-frame dynamic image data by estimating the motion of said ROI over the frames by registering each frame to a reference frame; computing a statistic representing the motion in the ROI over the frames; processing absolute and relative motion magnitude curves to derive patterns of motion; triggering a warning when the motion pattern is outside expected boundaries of acceptable motion patterns, identifying both location and duration of when the unexpected motion occurred; prompting actions for quality control purpose based on the detected motion pattern.

The method of the present invention may include steps of detecting frame intervals when motion occurs, and motion-free frame intervals; calculating statistics of the distribution of motion such as min, max, percentage and duration etc., or specific motion types such as gradual drift motion, from absolute and relative motion curves using combined threshold based or pattern recognition methods.

Referring to Fig. 9, the invention may be conveniently realized as a computer system suitably programmed with instructions for carrying out the steps of the methods according to the invention.

For example, a central processing unit 4 is able to receive data representative of medical scan data via a port 5 which could be a reader for portable data storage media (e.g. CD-ROM) ; a direct link with apparatus such as a medical scanner (not shown) or a connection to a network.

For example, in an embodiment, the processor performs such steps as a) Defining a region of interest ROI within the image data frames; b) Monitoring motion of the ROI within the image data frames; quantify the mean magnitude of the motion in a region of interest, c) Extracting properties of motion of the ROI within the image data frames, d) Notifying a user of excessive motion within the image data frames; e) Notifying the user of specific patterns of motion within the image data frames; and f) Recommending subsequent quality control actions regarding the detected motion properties.

Software applications loaded on memory 6 are executed to process the image data in random access memory 7. A Man - Machine interface 8 typically includes a keyboard/mouse/screen combination (which allows user input such as initiation of applications) and a screen on which the results of executing the applications are displayed.

While the present invention has been described with reference to a limited number of specific embodiments, given by way of examples only, numerous variations and modifications will be apparent to those skilled in the art, without departing from the scope of the invention as defined in the appended claims.

Claims (13)

Claims

1. A method for detecting motion in a series of image data frames, and providing a corresponding warning to a user, comprising the steps of: a) Defining a region of interest ROI within the image data frames; b) Monitoring and quantifying motion of the ROI within the image data frames; c) Extracting properties of motion of the ROI within the image data frames, d) Notifying a user of excessive motion within the image data frames; e) Notifying the user of specific patterns of motion within the image data frames; and f) Recommending subsequent quality control actions regarding the detected motion properties.

2. A method for detecting motion in a series of image data frames according to claim 1, wherein step (a) comprises the step of segmenting an ROI in multi-frame image dataset.

3. A method for detecting motion in a series of image data frames according to any preceding claim, wherein step (b) comprises the step of monitoring the motion of an imaging scan by generating a multi-time-frame dynamic image from raw imaging data.

4. A method for detecting motion in a series of image data frames, according to any preceding claim, wherein step (c) comprises the steps of estimating the motion of said ROI over the frames by registering each frame to a reference image frame; - computing a statistic representing the motion in the ROI over the frames.

5. A method for detecting motion in a series of image data frames, according to any preceding claim, wherein step (c) comprises the steps of - estimating the motion of said ROI in respective image frames by registering each frame to a respective preceding or successive image frame; - computing a statistic representing the motion in the ROI in each frame.

6. A method for detecting motion in a series of image data frames according to claim 5 when dependent upon claim 4, wherein step (c) comprises the steps of - calculating motion of said ROI over the image data frames with respect to reference baseline image data frame; and calculating a relative motion of the ROI in image data frames with reference to its consecutive preceding or successive image data frame; and processing the absolute and relative motion magnitude values to derive patterns of motion.

7. A method for detecting motion in a series of image data frames according to any preceding claim, wherein step (e) comprises triggering a warning when the motion pattern is outside expected boundaries of acceptable motion patterns.

8. A method for detecting motion in a series of image data frames according to claim 7, wherein step (f) comprises notifying a user of possible actions for quality control purpose based on a detected motion pattern.

9. A method for detecting motion in a series of image data frames according to claim 6, wherein the derived patterns of motion comprise at least one of: - detecting periods of when motion occurs and motion free periods; - detecting statistics of the distribution of motion; - detecting gradual drift motion.

10. A method for detecting motion in a series of image data frames according to claim 1, wherein step (a) comprises use of scan equipment provided with a motion monitoring system to obtain the image motion over a scan period.

11. A method for detecting motion in a series of image data frames, and providing a corresponding warning to a user, comprising the steps of: a) Monitoring and quantifying motion of the image data within the image data frames; b) Extracting properties of motion of the image data within the image data frames, d) Notifying a user of excessive motion within the image data frames; e) Notifying the user of specific patterns of motion within the image data frames; and f) Recommending subsequent quality control actions regarding the detected motion properties.

12. Apparatus for identifying a region of interest in medical imaging data of a subject, comprising a processor adapted to: a) Define a region of interest ROI within the image data frames; b) Monitor and quantify motion of the ROI within the image data frames; c) Extract properties of motion of the ROI within the image data frames, d) Notify a user of excessive motion within the image data frames; e) Notify the user of specific patterns of motion within the image data frames; and f) Recommend subsequent quality control actions regarding the detected motion properties.

13. A media device storing computer program code adapted, when loaded into or run on a computer, to cause the computer to become apparatus, or to carry out a method, according to any preceding claim.