WO2017069699A1

WO2017069699A1 - Medical image processing methods and systems for assessing right ventricular function

Info

Publication number: WO2017069699A1
Application number: PCT/SG2016/050495
Authority: WO
Inventors: Liang ZHONG; Xiaodan ZHAO; Shuang LENG; Ru San Tan
Original assignee: Singapore Health Services Pte Ltd
Priority date: 2015-10-20
Filing date: 2016-10-07
Publication date: 2017-04-27
Also published as: WO2017069699A9

Abstract

A medical image processing method of assessing right ventricular function in a subject from medical image data is disclosed. The medical image data comprising a sequence of images of the right ventricular region of the subject during a cardiac cycle for each of a plurality of views, each view of the plurality of views corresponding to a plane defined in relation to features of the right ventricular region. The method comprises: in each view of the plurality of views, tracking reference points in that view and determining the locations of the reference points in each image of the sequence of images corresponding to that view; and calculating a velocity of each reference point using the locations of that reference point in successive images of the sequence of images.

Description

Medical Image Processing Methods and Systems for Assessing Right Ventricular

Function

FIELD OF THE INVENTION

Embodiments of the present invention relate to medical image processing. More specifically, embodiments disclosed herein relate to the processing of medical images of a right ventricular region of a patient to assess right ventricular function in the patient.

BACKGROUND

An understanding of the role of the right ventricle in health and disease has lagged behind that of the left ventricle. Right ventricular (RV) dysfunction is associated with increased morbidity and mortality in patients with pulmonary hypertension (PH), congenital heart disease (CHD), coronary artery disease (CAD), heart failure (HF), and valvular heart diseases. Moreover, RV dysfunction may also affect left ventricular (LV) function, not only by limiting LV preload, but also by adverse systolic and diastolic interaction via the intraventricular septum and the pericardium. Thus, the need for diagnosis of RV dysfunction is highly evident. Accurate evaluation of RV function remains challenging, however, due to the complicated geometry and extreme sensitivity to loading conditions. At present, there is no widely accepted or generally applicable index of RV function. Right heart catheterization (RHC) remains the gold standard for assessment of RV function, but this requires an invasive procedure. In clinical practice, echocardiography is the main diagnostic modality for RV structure and function. Several echocardiography-derived indices, such as peak systolic and diastolic velocities of tricuspid annular (TA) motion, TA plane systolic excursion (TAPSE), and RV index of myocardial performance have emerged as promising parameters of RV function. Transthoracic echocardiography measurements have limitations, however, due to variability in sampling locations and ultrasound beam alignment. Very small changes in beam angle and changes in imaging window can result in dramatically different conclusions about RV size and function. Three-dimensional (3D) echocardiography is one of several emerging modalities for defining cardiac anatomy and function. Available evidence suggests that 3D echocardiography provides better accuracy over two-dimensional (2D) methods for evaluation of LV volume and function. There are still certain limitations to currently available 3D ultrasound methods, however, even with state-of-the-art real-time 3D echocardiography systems. In particular, relatively low image quality and low frame rate may limit everyday clinical use of 3D echocardiography.

Cardiac magnetic resonance (CMR) has been an alternative non-invasive modality of choice for quantitatively assessing RV volume and regional area strain. CMR- based measurements of RV ejection fraction (EF) are strong predictors of clinical outcome over a wide range of HF severity. Very few studies have been undertaken, however, to quantify the motion of tricuspid annulus for RV function assessment using CMR imaging. In Nijveldt, R., T. Germans, G. P. McCann, A. M. Beek, and A. C. van Rossum "Semi-quantitative assessment of right ventricular function in comparison to a 3D volumetric approach: a cardiovascular magnetic resonance study" Eur. Radiol. 18:2399-2405, 2008, the TAPSE and RV fractional shortening were manually evaluated in CMR and compared with volumetric assessment of RV function.

Another approach, described in Marcos, P., W. G. Vick, D. J. Sahn, M. Jerosch- Harold, A. Shurman, and F. H. Sheehan "Correlation of right ventricular ejection fraction and tricuspid annular plane systolic excursion in tetralogy of Fallot by magnetic resonance imaging" Int. J. Cardiovasc. Imaging. 25:263-270, 2009 used a modeling technique to measure the TAPSE at the junction of RV free wall and tricuspid annulus in the apical four-chamber CMR view. In Ito, S., D. B. McElhinney, R. Adams, P. Bhatla, S. Chung, and L. Axel "Preliminary assessment of tricuspid valve annular velocity parameters by cardiac magnetic resonance imaging in adults with a volume-overloaded right ventricle: comparison of unrepaired atrial septal defect and repaired tetralogy of Fallot" Pediatr. Cardiol. 36:1294-1300, 2015, the position of the RV atrioventricular junction (AVJ) was tracked in all CMR images and projected onto a reference line, bisecting the right ventricle between the RV apex and the free wall. The displacement of the RV AVJ along the reference line was measured relative to the position at end-diastole (ED) phase. The RV long-axis displacement has also been estimated by tagging CMR which demonstrated more significant differences between the studied groups.

Despite some recent advances, the 3D assessment of TA motion with CMR remains challenging predominantly due to a paucity of automatic, robust and time-efficient CMR-based methods.

SUMMARY OF THE INVENTION According to a first aspect of the present invention there is provided a medical image processing method of assessing right ventricular function in a subject from medical image data The medical image data comprises a sequence of images of the right ventricular region of the subject during a cardiac cycle for each of a plurality of views, each view of the plurality of views corresponding to a plane defined in relation to features of the right ventricular region. The method comprises: in each view of the plurality of views, tracking reference points in that view and determining the locations of the reference points in each image of the sequence of images corresponding to that view; reconstructing for each of a plurality of successive times, a three dimensional curve representing the boundary of the tricuspid annulus of the subject from the locations of the reference points in the plurality of views; and calculating, for the successive times, the area bounded by the three dimensional curve representing the boundary of the tricuspid annulus of the subject to generate sweep surface area data representing the area swept out by the tricuspid annulus at successive times. In an embodiment the method further comprises numerically differentiating the sweep surface area data to generate sweep surface area velocity data. In an embodiment the method further comprises extracting at least one peak sweep surface area velocity value from the sweep surface velocity data. In an embodiment, wherein at least one peak sweep surface area velocity (SSAV) value is at least one of a positive peak systolic SSAV, a negative peak early diastolic SSAV, and a negative peak late diastolic SSAV. In an embodiment the method further comprises extracting a maximum sweep surface area value from the sweep surface area data. The maximum sweep surface area value may be a maximum sweep surface area in systole value. According to a second aspect of the present invention there is provided a medical image processing method of assessing right ventricular function in a subject from medical image data. The medical image data comprises comprising a sequence of images of the right ventricular region of the subject during a cardiac cycle for each of a plurality of views, each view of the plurality of views corresponding to a plane defined in relation to features of the right ventricular region. The method comprises: in each view of the plurality of views, tracking reference points in that view and determining the locations of the reference points in each image of the sequence of images corresponding to that view; and calculating a velocity of each reference point using the locations of that reference point in successive images of the sequence of images.

In an embodiment, calculating a velocity of each reference point comprises dividing a distance between the reference points between two successive images of the sequence of images by an effective image acquisition time.

In an embodiment, the method further comprises determining a displacement of each reference point. The displacement may be determined by computing a cumulative integral of the velocity of that reference point. In an embodiment the method further comprises extracting peak velocity values for each of the reference points. In an embodiment the method further comprises determining a time to peak velocity for each of the reference points. The method may further comprise determining a standard deviation of the time to peak velocities for the reference points.

In an embodiment the medical image data is cardiovascular magnetic resonance imaging data. In an embodiment tracking reference points comprises: receiving a user selection of a mask region centered on a point in a first image of the sequence of images; performing a search over a search region of a second image of the sequence of images to identify a region of the second image that is a best match to the mask region; and determining the location of the reference point in the second image as the location of the region of the second image that is a best match to the mask region.

In an embodiment performing a search over a search region comprises computing a normalized cross correlation at each location in the search region between the region centered on that location and the mask region.

In an embodiment the views are long-axis views of the right ventricular region of the subject.

In an embodiment the views are one or more of the following: a four-chamber view; a right ventricle outflow view; and a right ventricle two-chamber view.

In an embodiment the reference points are one or more of the following: the right ventricle septal site; the right ventricle lateral site; the right ventricle anterospetal site; the right ventricle posterolateral site; the right ventricle anterior site and the right ventricle posterior site.

According to a third aspect of the present invention there is provided a medical image processing system for assessing right ventricular function in a subject from medical image data. The medical image data comprises a sequence of images of the right ventricular region of the subject during a cardiac cycle for each of a plurality of views, each view of the plurality of views corresponding to a plane defined in relation to features of the right ventricular region. The system comprises: a computer processor and a data storage device, the data storage device having a tracking component; a reconstruction component; and a sweep surface area calculation component comprising non-transitory instructions operative by the processor to: rack reference points in each view of the plurality of views, and determine the locations of the reference points in each image of the sequence of images corresponding to that view; reconstruct for each of a plurality of successive times, a three dimensional curve representing the boundary of the tricuspid annulus of the subject from the locations of the reference points in the plurality of views; and calculate for the successive times, the area bounded by the three dimensional curve representing the boundary of the tricuspid annulus of the subject to generate sweep surface area data representing the area swept out by the tricuspid annulus at successive times.

According to a fourth aspect of the present invention there is provided a medical image processing system for assessing right ventricular function in a subject from medical image data. The medical image data comprises a sequence of images of the right ventricular region of the subject during a cardiac cycle for each of a plurality of views, each view of the plurality of views corresponding to a plane defined in relation to features of the right ventricular region. The system comprises: a computer processor and a data storage device, the data storage device having a tracking component; and a velocity calculation component comprising non-transitory instructions operative by the processor to: track reference points in each view of the plurality of views, and determine the locations of the reference points in each image of the sequence of images corresponding to that view; and calculate a velocity of each reference point using the locations of that reference point in successive images of the sequence of images.

According to a further aspect, there is provided a non-transitory computer-readable medium. The computer-readable medium has stored thereon program instructions for causing at least one processor to perform operations of a method disclosed above.

BRIEF DESCRIPTION OF THE DRAWINGS

In the following, embodiments of the present invention will be described as non- limiting examples with reference to the accompanying drawings in which:

Figure 1 is a block diagram showing a medical image processing system according to an embodiment of the present invention; Figure 2 is a flow chart showing a method of processing medical image data to determine diagnostic markers from displacement or velocity of tricuspid annulus (TA) points according to an embodiment of the present invention; Figure 3 is a flow chart showing a method of processing medical image data to determine diagnostic markers by reconstructing the boundaries of the tricuspid annulus (TA) according to an embodiment of the present invention;

Figures 4a to 4d illustrate an example of medical image data which is processed in embodiments of the present invention;

Figures 5a to 5c illustrate the procedure of applying adaptive template matching in TA tracking on CMR imaging data according to an embodiment of the present invention;

Figures 6a to 6c show examples generated using methods according to embodiment of the present invention of velocity and displacement curves at TA sites for medical image data corresponding to different patients; Figures 7a to 7f show the reconstruction of 3D tricuspid annulus and extraction of SSA and SSAV using medical image data using a method according to an embodiment of the present invention;

Figure 8 is a table showing baseline demographic and clinical characteristics of study subjects;

Figure 9 shows a representative example of TA tracking results using methods according to an embodiment of the present invention; Figure 10 is a table showing average CMR-derived motion parameters based on 3D results according to an embodiment of the present invention; Figures 11a to 11d are scatter plots showing the relationships between diagnostic markers determined using embodiments of the present invention; RV ejection function (RVEF) and parameters determined from right heart catheterization (RHC); Figure 12 is a table showing results of sensitivity, specificity and AUC obtained by applying methods according to the present invention to medical image data generated from CMR-based measurements;

Figure 13 shows a side-by-side comparison of 2 normal subjects, 2 HF patients and 1 rTOF patient obtained using methods according to an embodiment of the present invention;

Figure 14 is a table showing CMR-derived peak SSAV and maximal SSA obtained using methods according to the present invention for different patient diagnostic groups; and

Figure 15 is a table showing results of intra-observer and inter-observer reproducibility analysis, expressed as Pearson's correlation coefficient, Bland-Altman analysis and ICC.

DETAILED DESCRIPTION

Figure 1 shows a medical image processing system according to an embodiment of the present invention. The image processing system 100 comprises a processor 110 which may be referred to as a central processing unit (CPU) that is in communication with memory devices including storage 120 and memory 150. The Processor 110 may be implemented as one or more CPU chips. The memory 150 is implemented as a random access memory (RAM). The medical image processing system further comprises input/output (I/O) devices 160 and network interfaces 170.

The storage 120 typically comprises one or more disk drives and is used for nonvolatile storage of data. The storage 120 stores program components 130 which are loaded into the memory 150 when such programs are selected for execution. The program components comprise a tracking component 131 ; a velocity calculation component 132; a displacement calculation component 133; a reconstruction component 134; a sweep surface area (SSA) calculation component 135; a sweep surface area velocity (SSAV) calculation component 136; and a diagnostic marker calculation component 137. Each of the program modules 130 stored in the storage 120 comprises non-transitory instructions operative by the processor 120 to perform the various operations in the methods described in more detail below. The storage 120 and the memory 150 may be referred to in some contexts as computer readable storage media and / or non-transitory computer readable media. The storage 120 also stores medical image data 140. The medical image data comprises a set of medical images of the RV region of a patient. The medical image data 140 comprises a sequence of images for each of a plurality of views of the RV region of the patient. The medical image data 140 may be in the DICOM format. The medical image data may be obtained through a medical imaging technique such as a computed tomography (CT) scan of the patient, a computed tomography angiography (CTA) scan; an ultrasound scan or a magnetic resonance imaging (MRI) scan of the patient. In the embodiments described below, the medical image data 140 comprises cardiac magnetic resonance (CMR) image data. It is to be understood that the systems and methods described herein could also be applied to images obtained with other imaging modalities.

The I/O devices 160 may include printers; video monitors; liquid crystal displays (LCDs); plasma displays; touch screen displays keyboards; keypads; switches; dials; mice; track balls; voice recognizers; card readers; or other well-known input devices.

The network interfaces 170 may take the form of modems, modem banks, Ethernet cards, universal serial bus (USB) interface cards, serial interfaces, token ring cards, fiber distributed data interface (FDDI) cards, wireless local area network (WLAN) cards, radio transceiver cards that promote radio communications using protocols such as code division multiple access (CDMA), global system for mobile communications (GSM), long-term evolution (LTE), worldwide interoperability for microwave access (WiMAX), near field communications (NFC), radio frequency identity (RFID), and/or other air interface protocol radio transceiver cards, and other well-known network devices. These network interfaces 170 may enable the processor 110 to communicate with the Internet or one or more intranets. With such a network connection, it is contemplated that the processor 110 might receive information from the network, or might output information to the network in the course of performing the below-described method operations. Such information, which is often represented as a sequence of instructions to be executed using processor 110, may be received from and outputted to the network, for example, in the form of a computer data signal embodied in a carrier wave.

The network interfaces 170 may allow the medical image processing system 110 to receive the medical image data 140 from a medical imaging apparatus for processing according an the method described below. As such the network interfaces 170 may act as an image import component which allows the system 100 to receive the medical image data 140. The processor 110 executes instructions, codes, computer programs, scripts which it accesses from hard disk, floppy disk, optical disk (these various disk based systems may all be considered storage 120), flash drive, memory 150, or the network interfaces 170. While only one processor 110 is shown, multiple processors may be present. Thus, while instructions may be discussed as executed by a processor, the instructions may be executed simultaneously, serially, or otherwise executed by one or multiple processors.

It is understood that by programming and/or loading executable instructions onto the medical image processing system 100, at least one of the CPU 110, the memory 150, and the storage 120 are changed, transforming the image processing system in part into a specific purpose machine or system having the novel functionality taught by the present disclosure. It is fundamental to the electrical engineering and software engineering arts that functionality that can be implemented by loading executable software into a computer can be converted to a hardware implementation by well- known design rules.

Although the system 100 is described with reference to a computer, it should be appreciated that the system may be formed by two or more computers in communication with each other that collaborate to perform a task. For example, but not by way of limitation, an application may be partitioned in such a way as to permit concurrent and/or parallel processing of the instructions of the application. Alternatively, the data processed by the application may be partitioned in such a way as to permit concurrent and/or parallel processing of different portions of a data set by the two or more computers. In an embodiment, the functionality disclosed above may be provided by executing the application and/or applications in a cloud computing environment. Cloud computing may comprise providing computing services via a network connection using dynamically scalable computing resources. A cloud computing environment may be established by an enterprise and/or may be hired on an as-needed basis from a third party provider.

Figures 2 and 3 are flow charts illustrating methods of processing medical image data according to embodiments of the present invention. The methods 200 and 300 may be performed on the medical image processing system 100 shown in Figure 1. It should be noted that enumeration of operations is for purposes of clarity and that the operations need not be performed in the order implied by the enumeration.

As described above, the methods 200 and 300 are carried out on medical image data 140. The medical image data 140 may be cardiac magnetic resonance (CMR) scan data. In an exemplary study using methods according to an embodiment of the present invention, CMR scans were performed using balanced turbo field echo sequence (BTFE). All subjects were imaged on a 3T magnetic resonance imaging (MRI) system (Ingenia, Philips Healthcare, Netherlands) with a dStream Torso coil (maximal number of channels 32).

Figures 4a to 4d show an example of medical image data which is processed in embodiments of the present invention.

Figures 4a to 4d show an example of CMR medical image data on which processing according to embodiments of the present invention may be carried out. Figures 4a to 4d show BTFE end-expiratory breath hold cine images were acquired in multi-planar long-axis views. Figure 4a shows a short-axis view; Figure 4b shows a long-axis four-chamber view; Figure 4c shows a long-axis right ventricular (RV) outflow view; and Figure 4d shows a long-axis RV two chamber view. Figure 4a shows the cutting planes through which the long axis views shown in Figures 4b to 4d are obtained. The long-axis four-chamber view shown in Figure 4b was obtained along the plane 410 shown in Figure 4a; the long axis RV outflow view shown in Figure 4c was obtained along the plane 412 shown in Figure 4c; and the long-axis RV two chamber view was obtained along the axis 414 shown in Figure 4a. The RV two-chamber view shown in Figure 4d was piloted by bisecting the right ventricle in the four-chamber view shown in Figure 4b as a slice through tricuspid valve and RV apex. The RV three-chamber (RV inflow/outflow) view shown in Figure 4c was piloted by 3 points: center of pulmonary valve in RV outflow tract (RVOT) view, center of tricuspid valve on RV four-chamber or two-chamber view, and apex on RV four-chamber view shown Figure 4b or two-chamber view shown in Figure 4d.

The following typical sequence parameters were used to obtain sequences of medical images making up the medical image data: TR/TE 3/1 ms, flip angle 45°, slice thickness 8 mm for both short- and long-axis, pixel bandwidth 1797 Hz, field of view 280 - 450 mm, temporal resolution «28 ms, in plane spatial resolution 0.6 mm ^χ 0.6 mm - 1.1 mm ^χ 1.1 mm, frame rate was selected as 30 or 40 frames per cardiac cycle.

The short-axis view in Figure 4a shows the right atrium RA; the left atrium LA; the aorta Ao; and the pulmonary artery PA. The long-axis four chamber view in Figure 4b shows the right atrium RA; the tricuspid valve TV; and the right ventricle RV. The RV septal tricuspid annular point 422 and the RV lateral tricuspid annular point 424, the motion of which is discussed in more detail below, are shown in Figure 4b. The long axis three-chamber view in Figure 4c shows the right atrium RA; the tricuspid valve TV; and the right ventricle RV. The RV anteroseptal tricuspid annular point 432 and the RV posterolateral tricuspid annular point 434, the motion of which is discussed in more detail below, are shown in Figure 4c. The long-axis RV two chamber view in Figure 4d shows the right atrium RA; the tricuspid valve TV; and the right ventricle RV. The RV anterior tricuspid annular point 442 and the RV posterior tricuspid annular point 444, the motion of which is discussed in more detail below, are shown in Figure 4d. The CMR imaging sequences are acquired in clinical practice, and conformed to the DICOM protocol. The following meta-information is recorded for all image sequences: trigger time, image position (3-by-1 vector denoted by ImagePos), image orientation along horizontal (3-by-1 vector denoted by lmageOri_h) and vertical (3-by-1 vector denoted by lmageOri_v) directions, and pixel spacing.

In some embodiments of the present invention TA points are identified throughout the cardiac cycle in multiple CMR planes. Peak velocities and maximal displacements at multiple TA points are extracted and interpreted. Alternatively, the 3D tricuspid annulus is reconstructed as a function of time by 2D projection and interpolation with the TA tracking results. Sweep surface area velocity, a new diagnostic marker, is calculated and analyzed. Detailed discussions on each step are given in the following subsections. Figure 2 shows a method 200 of processing medical image data to determine diagnostic markers from displacement or velocity of tricuspid annulus (TA) points according to an embodiment of the present invention.

In step 202, medical image data is received. The medical image data comprises sequences of images of the RV region of a patient as described above.

In step 204, reference points on the tricuspid annulus (TA) in the medical image data are tracked by the tracking component 131 of the medical image processing system 100. Step 204 may comprise semi-automatic tracking of TA motion in multiple CMR planes. This tracking may be carried out using software developed in the MATLAB environment (MathWorks Inc., MA, USA). The 6 different TA points described above with reference to Figures 4a to 4d may be tracked.

The tracking in step 204 uses the method of template matching which is an algorithm for searching for and finding the location of a template image within a larger image. This larger image is known as the search area. More information on the method of template matching can be found in Gonzalez, R. C, and R. E. Wood: "Digital image processing", Prentice-Hall, Inc. Upper Saddle 440 River, NJ, USA. 2006. Figures 5a to 5c show the procedure of applying adaptive template matching in TA tracking with CMR imaging.

As shown the upper part of Figure 5a, a user selects a template 502 with size (2w+ l)x(2A + l) in an end-diastolic (ED) frame of a CMR imaging sequence. The

ED frame is usually frame 1 of the sequence of frames received as the medical image data. The lower part of Figure 5a shows the region 506 enlarged. As shown in Figure 5a, the template 502 is a rectangular region. The region 502 is selected by visual inspection. Typical values of selected w and h range from 8 to 10 pixels. The search region 504 sharing the same center with the selected template 502 is then automatically generated in frame 2 with size (2w+ l + 2/) x (2/¾ + l + 2/) where / denotes the length of the searched neighborhood. The value of / (set to be 10 pixels in this example) may be selected such that the maximal displacement between TA points in successive frames was close to, but smaller than/ . The template matching was conducted to detect the best match of the template 502 in the search region 504 by sliding the template 502 image over the search image 504 one pixel at a time (left to right, up to down) while computing the normalized cross correlation at each location. As shown in Figure 5b, a resulting correlation image is formed. In Figure 5b, the dark region 508 represents the highest correlation and therefore the best match. The point 508 with the highest correlation coefficient in the resulting correlation image indicates the location of the best match. This point was used to update the template in frame 2.

As shown in Figure 5c, the point with the highest correlation is used to update the location of the template 510 in frame 2 which undergoes the same template matching within the automatically extracted search region 512 in frame 3. In step 206 the displacement calculation component 133 of the medical image processing system 100 determines the distance between the positions of the reference points between adjacent frames. This distance is determined as the Euclidean distance between template positions on two adjacent frames. In step 208, the velocity calculation component 132 of the medical image processing system 100 generates velocity curves for the reference points by dividing the CMR effective image acquisition time by the Euclidean distance between tracked template positions on two adjacent frames. TA displacement curves are then obtained as the cumulative integrals of TA velocity over the cardiac cycle, using the trapezoidal rule.

In step 210, the diagnostic marker calculation component of the medical image processing system 100 determines diagnostic markers from the velocity or displacement curves. The diagnostic markers may be output to a clinician or stored on the medical image processing system 100 for further analysis.

Figures 6a to 6c show examples of generating velocity and displacement curves at TA sites for medical image data corresponding to different patients.

Figure 6a shows results for a 44-year-old female healthy volunteer; Figure 6b shows results for a 51 -year-old male heart failure (HF) patient; and Figure 6c shows results for a 19-year-old male repaired tetralogy of Fallot (rTOF) patient. As shown in each of Figures 6a to 6c, the following TA points were identified and tracked: the RV anterior 602a, 602b, 602c; the RV posterior 604a, 604b, 604c; the RV anteroseptal 606a, 606b, 606c; the RV posterolateral 608a, 608b, 608c; the RV lateral 610a, 610b, 610c; and the RV septal 612a, 612b, 612c. TA velocity curves 620a, 620b, 620c were generated by dividing the CMR effective image acquisition time by the Euclidean distance between tracked template positions on two adjacent frames.

TA displacement curves 630a, 630b, 630c were then obtained as the cumulative integrals of TA velocity over the cardiac cycle, using the trapezoidal rule.

At each of the 6 TA sites, 3 peak velocities were automatically extracted in systole, and in early and late diastole using segmental peak detection. These were (i) positive peak systolic velocity as the tricuspid annulus descends toward the RV apex (Sm) 622a, 622b, 622c; (ii) early diastolic velocity below the baseline as the tricuspid annulus ascends away from the RV apex (Em) 624a, 624b, 624c; and (iii) late diastolic velocity during right atrial contraction (Am) 626a, 626b, 626c. Maximal displacement tricuspid annular plane systolic excursion (TAPSE) 632a 632b 632c in ventricular systole was determined from the TA displacement curve 630a, 630b, 630c. In addition, time to Sm, Em, Am and TAPSE (denoted T-Sm, T-Em, T- Am and T-TAPSE) were recorded and used to calculate RV peak velocity and displacement dyssynchrony indices, defined as the standard deviation (SD) of time to peak velocity and SD of time to maximal displacement, respectively, among the 6 TA points.

The diagnostic markers calculated in step 210 include 6-point mean Sm, Em, Am, TAPSE, and standard deviations of T-Sm, T-Em, T-Am, T-TAPSE.

Figure 3 shows a method 300 of processing medical image data to determine diagnostic markers by reconstructing the boundaries of the tricuspid annulus (TA) according to an embodiment of the present invention. The diagnostic markers obtained by the method 200 shown in Figure 2 may be combined with the diagnostic markers obtained by the method 300 shown in Figure 3, alternatively, diagnostic markers obtained by either the method 200 shown in Figure 2 or the method 300 shown in Figure 3 may be analyzed separately and independently.

In step 302, the image processing apparatus 100 receives medical image data. The medical image data may be CMR data as described above with reference to Figures 4a to 4c.

In step 304, the tracking component 131 of the medical image processing apparatus 100 tracks points on the tricuspid annulus. This may be carried out as described above in relation to step 204 of Figure 2. The semi-automatic tracking system provided spatial coordinates of multiple points located on the tricuspid annulus as a function of CMR frame time. Figures 7a to 7f show the reconstruction of 3D tricuspid annulus and extraction of SSA and SSAV in a 34-year-old female healthy volunteer using a method according to an embodiment of the present invention. As shown in Figure 7a, TA points in 2D coordinate system at a first cardiac frame referred to as cardiac frame 1 are identified.

The TA points then are mapped into a 3D coordinate system with respective image position and image orientation information using the transformation:

Coord_3D = Coord _{2D x} x ImageOri _h + Coord _{2D y} x ImageOri _v + ImagePos ( 1 ) where Coord_3D represents the 3-by-1 coordinate vector in 3D space, scalars Coord_{2D x} and Coord_{2D y} are 2D coordinates in x- and y-axis, ImageOri,, and ImageOri_v are the 3-by-1 image orientation vectors along horizontal and vertical directions, and ImagePos denotes the 3-by-1 image position vector.

The mapped 3D coordinates Coord_3D are projected onto a new 2D Cartesian coordinate plane defined by the centroid of the mapped coordinates and two points in the 3D coordinate system. As shown in Figure 7b, the TA points are mapped onto the new 2D Cartesian coordinate system. The 6 TA points: RV posterolateral 704; RV posterior 706; RV septal 708; RV anteroseptal 710; RV anterior 712; and RV lateral 714 are shown in Figure 7b. In step 306, the boundary of the tricuspid annulus is reconstructed from the tracked TA points by the reconstruction component 134 of the medical image processing system.

The reconstructed TA boundary 720 is shown in Figure 7b. A curve was generated in the 2D Cartesian plane by interpolating the projected coordinates using piecewise cubic Hermite interpolation. The TA boundary 720 shown in Figure 7b was reconstructed by inverse projection of the curve back onto the 3D coordinate system. As shown in Figure 7c, the same procedure was applied for TA reconstruction at each cardiac frame. Figure 7c shows the reconstructed TA boundaries for frames 1 , 3, 5, 7, 9, and 11 , respectively labeled as 720 722 724 726 728 730.

In step 308 the SSA calculation component 135 of the medical image processing system 100 determines sweep surface area (SSA) data for the tricuspid annulus. Following the TA reconstruction by a curve representing the boundary of the tricuspid annulus at each CMR frame as shown in Figure 7c, the SSA that is the area bounded between the corresponding curves is computed as the surface area swept out by the tricuspid annulus at successive CMR frames. This calculation may be carried out by Delaunay triangulation.

Figure 7d shows the calculation of the SSA between successive frames. As shown in Figure 7d, the indicative illustration of sweep surface area calculation SSA-i-3: 742 is the SSA from frame 1 to 3; SSA_3-5: 744 is SSA from frame 3 to 5, SSA_5-7: 746 is SSA from frame 5 to 7, and SSA_5-7: 748 is SSA from frame 7 to 9.

In step 310, the SSAV calculation component 136 of the medical image processing system 100 determines sweep surface area velocity (SSAV) data from the SSA data determined in step 308. In step 310, the rate of TA motion is quantified using SSAV, by taking the first order time derivative of SSA.

In step 312, the diagnostic marker calculation component 137 of the medical image processing system 100 calculates or extracts diagnostic markers from the SSA data and / or the SSAV data.

Figure 7e shows the SSAV curve and extracted peak SSAV values. The extracted values from the SSAV curve 750 are positive peak systolic SSAV (SSSAV) 752, negative peak early diastolic SSAV (ESSAV) 754, and negative peak late diastolic SSAV (ASSAV) 756.

Figure 7f shows the SSA curve and extracted peak SSA value. The extracted value from the SSA curve 760 is maximal SSA (SSA_max) in systole 762. The significance of the extracted diagnostic markers will now be described. The data was analyzed using SPSS (version 17.0, Chicago, IL, USA) and SAS (version 9.3, Cary, NC, USA). Comparisons of demographics, patient characteristics, and CMR measurements between patients and control subjects were performed using independent t tests for normally distributed data, Mann-Whitney U tests for non- normally distributed data, and Fisher's exact test for categorical data. Mean age differed substantially among some study groups, which might influence cardiac function measurements. Therefore, each CMR-based TA motion parameter was adjusted for age using one-way analysis of covariance. An F test was used to test the omnibus hypothesis of equality among diagnosis groups of age-adjusted least- squares means, which was then followed up with post-hoc comparisons of HF, rTOF, PH and HCM age-adjusted means to the normal mean. A p value < 0.05 was considered statistically significant.

Intra- and inter-observer variability in CMR-derived TA velocities was assessed by Pearson's r correlation, Bland-Altman analysis and intra-class correlation coefficient (ICC) using data from 6 randomly chosen subjects (2 normal controls, 2 HF patients and 2 rTOF patients). Inter-observer variability was assessed by comparing measurements made by two independent observers. Intra-observer variability was assessed from repeated measurements, 3 days apart, on the same 6 cases by the same observer.

Baseline demographic and clinical characteristics of enrolled subjects are summarized in the table shown in Figure 8.

Figure 8 is a table showing baseline demographic and clinical characteristics of study subjects. Data are represented as mean ± SD. BSA: body surface area; DP: diastolic pressure; SP: systolic pressure; LV: left ventricular; EDV: end-diastolic volume; ESV: end-systolic volume; SV: stroke volume; EF: ejection fraction; RV: right ventricular; HF: heart failure; rTOF: repaired tetralogy of fallot; PH: pulmonary hypertension; HCM: hypertrophic cardiomyopathy. *denotes statistically significant difference between normal controls and patients (p < 0.05). a Fisher's exact test. The data shown in the table of Figure 8 shows HF patients had significantly lower LVEF and LV stroke volume (SV) and higher LV end-diastolic volume (EDV), LV end-systolic volume (ESV) and LV mass compared to normal controls. Significant RV dilation in rTOF patients was indicated on CMR by large values of RVEDV, RVESV and RVSV. PH patients had significantly higher RVEDV and RVESV. Patients with HF, rTOF, and PH exhibited significantly reduced RVEF compared to controls.

During the study, semi-automatic TA tracking required approximately 3 minutes per subject (6 TA points), including initialization, tracking and motion parameter extraction. Velocity and displacement curves were obtained for all subjects.

Figure 9 shows a representative example of TA tracking results using methods according to an embodiment of the present invention. The results shown in Figure 9 are for a four-chamber view in a 44-year-old female healthy volunteer.

The sequence of frames at the top of Figure 9 shows RV septal tracking. The template 902 is tracked through frames at times of 0ms 910; 364ms 920; 754ms 930; and 1014ms 940. The sequence of frames at the bottom of Figure 9 shows RV lateral tracking. The template 904 is tracked through frames at times of 0ms 950; 364ms 960; 754ms 970; and 1014ms 980.

The sequences of frames correspond to start-of-systole 910 950, end-systole (ES) 920 960, diastasis 930 970, and end-diastolic (ED) 940 980.

Referring now to Figure 6, velocity and displacement curves at 6 TA sites are shown in Figure 6a. Three peaks (one positive and two negative) are present for each velocity curve and correspond to the peak velocities during systole (Sm), early diastolic rapid filling (Em), and late diastolic atrial contraction (Am). The extracted values of 6-point mean Sm, Em and Am were 10.4, 11.6, and 11.0 cm s-1 , respectively. The mean value for TAPSE derived from the displacement curves was 21.8 mm. The respective SDs for times to peak velocity and displacement (SD T-Sm, SD T-Em, SD T-Am and SD T-TAPSE) were 18.1 , 16.8, 12.8 and 18.9 ms. Moreover, one can observe that the tricuspid annulus is nearly stationary during the diastolic slow filling phase as indicated by a zone of very low velocity and constant displacement. Figures 6b and 6c show results for a 51 -year-old male HF patient and a 19-year-old male rTOF patient. Figure 10 is a table showing average CMR-derived motion parameters based on 3D results. Values are means ± SD. CMR: cardiac magnetic resonance; TA: tricuspid annular; Sm: peak TA systolic velocity; Em: peak TA velocity during early diastolic filling; Am: peak TA velocity during atrial contraction; TAPSE: maximal displacement; T-Sm: time to Sm; T-Em: time to Em; T-Am: time to Am; T-TAPSE: time to TAPSE; SD: standard deviation; HF: heart failure; rTOF: repaired tetralogy of Fallot; PH: pulmonary hypertension; HCM: hypertrophic cardiomyopathy. Values marked with "$" are adjusted for age.

The table shown in Figure 10 gives the 3D CMR-derived motion parameters with differentiation by subject group. The 6-point mean results show that patients with HF and PH had significantly reduced Sm (4.2 ± 1.3/5.0 ± 1.0 vs. 9.7 + 1.7 cm s-1 , p < 0.05), Em (3.4 ± 1.0/3.7 ± 1.0 vs. 8.3 ± 3.0 cm s-1 , p < 0.05), Am (4.3 ± 2.4/6.0 ± 2.1 vs. 10.1 ± 2.6 cm s-1 , p < 0.05) and TAPSE (7.9 ± 2.7/8.8 ± 1.8 vs. 17.6 ± 2.4 mm, p < 0.05) and significantly larger SD for time to TAPSE (51.3 ± 19.7/84.3 ± 20.3 vs. 24.1 ± 6.0 ms, p < 0.05) in comparison with normal controls.

The group of rTOF patients exhibited significantly lower 6-point mean Sm (7.5 ± 1.1 vs. 9.7 ± 1.7 cm s-1 , p < 0.05), Am (6.6 ± 2.1 vs. 10.1 ± 2.6 cm s-1 , p < 0.05) and TAPSE (13.6 ± 2.3 vs. 17.6 ± 2.4 mm, p < 0.05) compared to normal controls.

Patients with HCM had comparable Sm (9.5 ± 1.9 vs. 9.7 ± 1.7 cm s-1 , p = 0.846) but relatively lower Em (6.1 ± 0.8 vs. 8.3 ± 3.0 cm s-1 , p = 0.062) and Am (8.3 ± 1.1 vs. 10.1 ± 2.6 cm s-1 , p = 0.137) compared to those in the control group. Figures 11 a to 11 d are scatter plots showing the relationships between diagnostic markers determined using embodiments of the present invention; RV ejection function (RVEF) and parameters determined from right heart catheterization (RHC). RHC was performed at rest using standard techniques for continuous measurement of RV chamber pressure. The RV pressure was then differentiated with respect to time for determination of maximal rate of increase during systole (RV dP/dtmax). The RV dP/dtmax was normalized to instantaneous pressure (IP) at which dP/dtmax occurred (dP/dtmax/IP). Peak ECG R wave was used as end-diastolic timing marker. Figure 1 1a shows the relationship between CMR-based mean Sm and RVEF. As shown in Figure 1 1a, good overall correlation was found between CMR-derived 6- point average peak systolic TA velocity Sm and RVEF (r = 0.675 p < 0.01 ).

Figure 1 1 b shows the relationship between CMR-based mean TAPSE and RVEF. As shown in Figure 11 b, good overall correlation was found between the CMR-derived 6-point mean TAPSE and RVEF (r = 0.651 , p < 0.01 ).

The utility of CMR-based measurements for diagnosing disease states (HF, rTOF and PH) versus normal controls is demonstrated in the table shown in Figure 12, where area under the receiver-operating characteristic (ROC) curve (AUC) for the CMR-based 6-point mean Sm and TAPSE, and RVEF were 0.957, 0.981 and 0.871 , respectively.

Figure 12 is a table showing results of sensitivity, specificity and AUC using CMR- based measurements. In Figure 12, the following abbreviations are used: CMR: cardiac magnetic resonance; Sm: peak TA systolic velocity; TAPSE: maximal displacement; TA: tricuspid annular; RV: right ventricular; EF: ejection fraction; AUC: area under the ROC curve. The CMR-derived Sm and TAPSE in the patient cohort with PH were based on an average of 4 TA points, since only four-chamber and RV two-chamber CMR images were available for this group of patients.

Figure 11c shows the relationship between CMR-based mean Sm and dP/dt_max/IP from RHC. Figure 11d shows the relationship between CMR-based mean TAPSE and dP/dtmax/IP from RHC. As shown in Figures 11c and 11d, both Sm and TAPSE from CMR correlated positively with dP/dWIP from RHC (Sm: r = 0.621 , p < 0.01 ; TAPSE: r = 0.648, p < 0.01 ). Figures 7e and 7f described above show the SSAV and SSA curves for a 34-year- old female healthy volunteer. The extracted global values of SSSAV. ESSAV, ASSAV. and SSAmax were 173.9, 186.6, 186.7 cm² s \ and 33.1 cm², respectively. Figure 13 shows a side-by-side comparison of 2 normal subjects, 2 HF patients and 1 rTOF patient obtained using methods according to an embodiment of the present invention.

The left hand column of Figure 13 shows 3D TA reconstruction for cardiac frames 1 , 3, 5, 7, 9, 11 , 13, and 15. The middle column of Figure 13 shows extraction of SSAmax. The right hand column of Figure 13 shows extraction of SSSAV, ESSAV and ASSAV in two normal subjects (N1 , N2), two HF patients (HF1 , HF2) and one rTOF patient (rTOF). Figure 14 is a table showing CMR-derived peak SSAV and maximal SSA for different patient diagnostic groups. The values shown in the table are means ± SD. The following abbreviations are used: CMR: cardiac magnetic resonance; SSAV: sweep surface area velocity; SSA: sweep surface area; SSSAV: peak systolic SSAV; ESSAV: peak SSAV during early diastolic filling; ASSAV-" peak SSAV during atrial contraction; SSA_max: maximal SSA; HF: heart failure; rTOF: repaired tetralogy of Fallot; PH:

pulmonary hypertension; HCM: hypertrophic cardiomyopathy. Values marked with a "$" symbol are Adjusted for age.

The CMR derived peak SSAV and maximal SSA values for each subject group presented in the table shown in Figure 13 indicate that patients with HF had significantly lower SSSAV (53.6 ± 13.1 vs. 129.7 ± 44.2 cm² s^"\ p < 0.05), ESSAV (49.9 ± 13.5 vs. 129.1 ± 54.2 cm² s \ p < 0.05), A_SSAV (59.7 ± 35.4 vs. 157.3 ± 96.2 cm² s \ p < 0.05) and SSA_max (1 1.3 ± 4.1 vs. 24.4 ± 7.2 cm², p < 0.05) when compared to normal controls. Patients with PH had similar significant reduction in SSA and SSAV measurements with the single exception of ASSAV which did not reach statistical significance (75.9 ± 17.3 vs. 157.3 ± 96.2 cm² s^"\ p = 0.061 ). Patients in the rTOF group had significantly reduced SSSAV (77.3 ± 21.4 vs. 129.7 ± 44.2 cm² s^"1, p < 0.05), ASSAV (71.1 ± 18.4 vs. 157.3 ± 96.2 cm² s^' p < 0.05) and SSA_MAX (16.4 ± 3.0 vs. 24.4 ± 7.2 cm², p < 0.05) in comparison to the controls.

Figure 15 is a table showing results of intra-observer and inter-observer reproducibility analysis, expressed as Pearson's correlation coefficient, Bland-Altman analysis and ICC. In Figure 15, the follow abbreviations are used: ICC: Intra-class correlation coefficient; CI: confidence interval; Sm: peak TA systolic velocity; Em: peak TA velocity during early diastolic filling; Am: peak TA velocity during atrial contraction; HF: heart failure; rTOF: repaired tetralogy of Fallot.

As demonstrated by Figure 15, the CMR derived measurements demonstrated excellent consistency as reflected by Pearson's correlation (ranges: intra-observer, 0.995 - 0.982; inter-observer, 0.997 - 0.975) and ICC (ranges: intra-observer, 0.995 - 0.977; inter-observer, 0.997 - 0.975) with no significant bias and narrow limits of agreement for both intra-observer and inter-observer measurements.

The present study unveiled the relationship between CMR imaging and TA motion assessment. Currently, clinicians use M-mode echocardiography and tissue Doppler imaging (TDI) to measure regional TA velocity, displacement and timing at the RV free wall level. The angle dependency is a serious limitation for all Doppler-based techniques, however, including Doppler-derived myocardial velocities. It is also not unusual to have some variation in peak velocity amplitude and timing by subtle changes in the sample volume positioning. The number of RV segments that can be assessed by TDI is limited, and only measurements of septum-to-RV free wall dyssynchrony are feasible. Moreover, attainment of standard RV views may be difficult in practice because the RV free wall may not be optimally imaged from typical echocardiographic windows due to its anterior location and lung interference. In contrast, the present study adopted CMR imaging to evaluate the 3D TA motion for RV systolic and diastolic function assessment. Compared to echocardiography, CMR has high spatial resolution, excellent tissue characterization, and free spatial orientation of imaging planes. The proposed CMR-based procedure is automated, reproducible, and efficient. As described above, comprehensive assessment of regional TA motion in 3D may provide novel insights into RV pathophysiologic mechanism and function. Hence, multiple CMR planes (four-chamber, RV three-chamber, RV two-chamber) were selected to cover the tricuspid annulus across its septal, anterior, free wall, and inferior components. The long-axis four-chamber view is a basic planar representation for studying the right ventricle and is routinely obtained in conventional cardiac MRI protocols. The right two- and three-chamber views are specific planes recommended when there is clinical suspicion of RV involvement. Further insights into the 3D TA motion can be acquired from the in-depth analysis in terms of velocity distribution and time differences in peak systolic and diastolic velocities along the tricuspid annulus.

Accurate and time efficient tracking of TA motion in selected CMR planes is important in the subsequent extraction of motion parameters. The general motion of TA points observed in each CMR plane involves three components: translation, rotation and deformation. In fact, the major component of TA motion can be described as translational motion with superimposed slow-varying rotational and deformational motion. Hence, a simple - yet effective and reliable - tracking method incorporating adaptive template matching was adopted in this study. The translational TA motion between consecutive frames was determined using basic template matching, and the tracking errors due to all the non-translational transformations were minimized by iterative updates of template throughout the cardiac cycle. As described above, two indices (SSA and SSAV) were proposed to further quantify and assess 3D TA motion. These indices measure the "swept area change" globally and regionally along the reconstructed tricuspid annulus. The present study has demonstrated a potential CMR-based methodology for evaluating not only the TA configuration but also its dynamic behavior and trajectory during the cardiac cycle. By considering the 3D TA motion as a whole (in both radial and longitudinal directions), the proposed method and parameters provide good morphological visualization and functional description of the tricuspid annulus. The present study provides a novel CMR-based RV diagnostic solution and a set of markers comprising of the following:

• Regional and global peak systolic and diastolic velocities and ratio {Sm (SSSAV), Em (ESSAV), Am (ASSAV), Em/Am (E_SSAV/ASSAV)}; and maximal displacement {TAPSE (SSAmax)}-

• Time to regional peak velocities {T-Sm, T-Em, T-Am} and maximal displacements {T-TAPSE}; and standard deviation for these regional timing values {SD T-Sm, SD T-Em, SD T-Am, SD T-TAPSE}.

The former are related to the RV function in terms of ventricular pumping ability and cardiac output, and the latter quantify RV dyssynchrony.

In the present study, we observed significant differences pertaining to the TA motion among the enrolled patient groups.

First, the peak systolic (Sm) and diastolic (Em and Am) velocities, as well as maximal displacement (TAPSE), were significantly reduced in HF patients in comparison with normal controls. These outcomes were consistent with earlier studies and support the hypothesis that the RV dysfunction may develop in association with LV dysfunction via multiple mechanisms. Patients in the rTOF group had reduced (albeit not statistically significant) TA velocity Em during early diastolic filling and a significantly lower peak systolic velocity Sm, late diastolic TA velocity Am during atrial contraction, and TAPSE in systole as compared with normal controls. These findings were in line with prior studies that have consistently found systolic and diastolic abnormalities in patients with rTOF. Significant decreases in both systolic (Sm) and diastolic velocities (Em, Am) were observed in patients with PH. In the presence of RV pressure overload, as occurs in PH patients, the RV function decline is related more to loss of RV transversal displacement than to longitudinal shortening. The current CMR-based method is advantageous in this respect it is an angle independent technique, thus avoiding the limitations of TDI related to translational cardiac motion. The diastolic function of the right ventricle in patients with HCM tended to be impaired, with reduced Em and Am velocities, which agreed with earlier work suggesting that ventricular interdependence and increased chamber stiffness may constitute the possible mechanisms of RV diastolic dysfunction.

Second, patients with HF, rTOF and PH had greater intra-ventricular contractile timing dyssynchrony, which was quantified by differences in motion timing among 6 TA sites from 3 CMR views as compared with controls. Similar results have been presented in [37-39] demonstrating dyssynchronous TA motion in these patient groups. It is well established that cardiac ^synchronization therapy (CRT) can improve cardiac function and enhance functional capacity in selected HF patients. Studies have indicated that the beneficial effects of CRT are related mainly to the reduction of mechanical dyssynchrony within the left and/or right ventricle with subsequent improvement in pumping efficiency. Hence, assessment of ventricular mechanical dyssynchrony from 3D CMR may be useful for future guidance of CRT. The assessment of systolic RV dysfunction provides significant diagnostic, therapeutic, and prognostic information in patients with congenital and acquired heart disease such as rTOF and congestive HF. RVEF is a widely adopted clinical parameter for systolic RV function assessment. The present study investigated TA motion parameters derived from CMR and observed good overall correlation between RVEF and the 6-point average Sm and TAPSE for all enrolled subjects. Moreover, the results described above have demonstrated better diagnostic accuracy of Sm and TAPSE by CMR as compared to RVEF.

Systolic HF (SHF) and diastolic HF (DHF) are the two clinical subsets of HF syndrome most frequently encountered in clinical practice. RVEF is relatively non- informative in diagnosing DHF, since patients with DHF typically exhibit ventricular EF within the normal range. The present study suggests that TA peak systolic velocity Sm and maximal displacement TAPSE may be more sensitive for detecting systolic dysfunction than RVEF in the HF patient group.

The CMR-based evaluation of TA motion in a regional context may also be informative in diagnosing deteriorated regional myocardial motion and abnormalities in the TA contractile pattern in rTOF group. The study described above provides an informative evaluation showing how the TA motion related measurements compare with information provided by the gold standard method - RHC - in patients with PH who had indication for invasive procedure. The results suggest that the systolic TA motion parameters derived by CMR imaging can be used as useful surrogate marker of RV function.

The evaluation of TA motion is further extended based on CMR-based SSAV and SSA, which reflect radial, longitudinal and circumferential changes of the tricuspid annulus. The much smaller peak SSAV values and maximal SSA shown in the table of Figure 14 implied a slow and irregular motion of tricuspid annulus in the HF and rTOF patient groups. Results in the present study suggest that SSA and SSAV of the 3D tricuspid annulus are novel and valuable clinical indicators for RV assessment in patients with diverse heart diseases. The 3D TA reconstruction and SSAV extraction are clinically useful in terms of (1 ) quantifying RV systolic and diastolic function, (2) assessing tricuspid valve function, (3) monitoring ventricular remodeling in patients after heart attack, (4) enhancing the indication of CRT, and (5) evaluating the effectiveness of medical/surgical therapy in patients.

Whilst the foregoing description has described exemplary embodiments, it will be understood by those skilled in the art that many variations of the embodiments can be made within the scope of the present invention as defined by the appended claims.

Claims

1. A medical image processing method of assessing right ventricular function in a subject from medical image data, the medical image data comprising a sequence of images of the right ventricular region of the subject during a cardiac cycle for each of a plurality of views, each view of the plurality of views corresponding to a plane defined in relation to features of the right ventricular region, the method comprising: in each view of the plurality of views, tracking reference points in that view and determining the locations of the reference points in each image of the sequence of images corresponding to that view;

reconstructing for each of a plurality of successive times, a three dimensional curve representing the boundary of the tricuspid annulus of the subject from the locations of the reference points in the plurality of views; and

calculating, for the successive times, the area bounded by the three dimensional curve representing the boundary of the tricuspid annulus of the subject to generate sweep surface area data representing the area swept out by the tricuspid annulus at successive times.

2. A method according to claim 1 , further comprising numerically differentiating the sweep surface area data to generate sweep surface area velocity data.

3. A method according to claim 2, further comprising extracting at least one peak sweep surface area velocity value from the sweep surface velocity data.

4. A method according to claim 3, wherein at least one peak sweep surface area velocity (SSAV) value is at least one of a positive peak systolic SSAV, a negative peak early diastolic SSAV, and a negative peak late diastolic SSAV.

5. A method according to any one of claims 1 to 4, further comprising extracting a maximum sweep surface area value from the sweep surface area data.

6. A method according to claim 5, wherein the maximum sweep surface area value is a maximum sweep surface area in systole value.

7. A medical image processing method of assessing right ventricular function in a subject from medical image data, the medical image data comprising a sequence of images of the right ventricular region of the subject during a cardiac cycle for each of a plurality of views, each view of the plurality of views corresponding to a plane defined in relation to features of the right ventricular region, the method comprising: in each view of the plurality of views, tracking reference points in that view and determining the locations of the reference points in each image of the sequence of images corresponding to that view; and

calculating a velocity of each reference point using the locations of that reference point in successive images of the sequence of images.

8. A method according to claim 7, wherein calculating a velocity of each reference point comprises dividing a distance between the reference point between two successive images of the sequence of images by an effective image acquisition time.

9. A method according to any one of claims 7 to 8, further comprising determining a displacement of each reference point.

10. A method according to claim 9, wherein determining a displacement of each reference point comprises computing a cumulative integral of the velocity of that reference point.

11. A method according to any one of claims 7 to 10, further comprising extracting peak velocity values for each of the reference points.

12. A method according to claim 11 , further comprising determining a time to peak velocity for each of the reference points.

13. A method according to claim 12, further comprising determining a standard deviation of the time to peak velocities for the reference points.

14. A method according to any preceding claim, wherein the medical image data is cardiovascular magnetic resonance imaging data.

15. A method according to any preceding claim, wherein tracking reference points comprises

receiving a user selection of a mask region centered on a point in a first image of the sequence of images;

performing a search over a search region of a second image of the sequence of images to identify a region of the second image that is a best match to the mask region; and

determining the location of the reference point in the second image as the location of the region of the second image that is a best match to the mask region.

16. A method according to claim 15, wherein performing a search over a search region comprises computing a normalized cross correlation at each location in the search region between the region centered around that location and the mask region.

17. A method according to any preceding claim, wherein the views are long-axis views of the right ventricular region of the subject.

18. A method according to claim 17, wherein the views are one or more of the following: a four-chamber view; a right ventricle outflow view; and a right ventricle two-chamber view.

19. A method according to any preceding claim, wherein the reference points are one or more of the following: the right ventricle septal site; the right ventricle lateral site; the right ventricle anterospetal site; the right ventricle posterolateral site; the right ventricle anterior site and the right ventricle posterior site.

20. A medical image processing system for assessing right ventricular function in a subject from medical image data, the medical image data comprising a sequence of images of the right ventricular region of the subject during a cardiac cycle for each of a plurality of views, each view of the plurality of views corresponding to a plane defined in relation to features of the right ventricular region, the system comprising: a computer processor and a data storage device, the data storage device having a tracking component; a reconstruction component; and a sweep surface area calculation component comprising non-transitory instructions operative by the processor to:

track reference points in each view of the plurality of views, and determine the locations of the reference points in each image of the sequence of images corresponding to that view;

reconstruct for each of a plurality of successive times, a three dimensional curve representing the boundary of the tricuspid annulus of the subject from the locations of the reference points in the plurality of views; and

calculate for the successive times, the area bounded by the three dimensional curve representing the boundary of the tricuspid annulus of the subject to generate sweep surface area data representing the area swept out by the tricuspid annulus at successive times.

21. A medical image processing system according to claim 20, the data storage device further comprising a sweep surface area velocity calculation component comprising non-transitory instructions operative by the processor to numerically differentiate the sweep surface area data to generate sweep surface area velocity data.

22. A medical image processing system according to claim 21 wherein the sweep surface area velocity calculation component further comprises non-transitory instructions operative by the processor to extract at least one peak sweep surface area velocity value from the sweep surface velocity data.

23. A medical image processing system according to claim 22, wherein the at least one peak sweep surface area velocity (SSAV) value is at least one of a positive peak systolic SSAV, a negative peak early diastolic SSAV, and a negative peak late diastolic SSAV.

24. A medical image processing system according to any one of claims 20 to 23, wherein the sweep surface area calculation component further comprises non- transitory instructions operative by the processor to extract a maximum sweep surface area value from the sweep surface area data.

25. A medical image processing system according to claim 31 , wherein the maximum sweep surface area value is a maximum sweep surface area in systole value.

26. A medical image processing system for assessing right ventricular function in a subject from medical image data, the medical image data comprising a sequence of images of the right ventricular region of the subject during a cardiac cycle for each of a plurality of views, each view of the plurality of views corresponding to a plane defined in relation to features of the right ventricular region,

the system comprising:

a computer processor and a data storage device, the data storage device having a tracking component; and a velocity calculation component comprising non- transitory instructions operative by the processor to:

track reference points in each view of the plurality of views, and determine the locations of the reference points in each image of the sequence of images corresponding to that view; and

calculate a velocity of each reference point using the locations of that reference point in successive images of the sequence of images.

27. A medical image processing system according to claim 26, wherein the velocity calculation component further comprises non-transitory instructions operative by the processor to calculate a velocity of each reference point by dividing a distance between the reference point between two successive images of the sequence of images by an effective image acquisition time.

28. A medical image processing system according to any one of claims 26 to 27, the data storage device further comprising a displacement calculation component comprising non-transitory instructions operative by the processor to determine a displacement of each reference point.

29. A medical image processing system according to claim 28, wherein the displacement calculation component comprises non-transitory instructions operative by the processor to determine a displacement of each reference point comprises computing a cumulative integral of the velocity of that reference point.

30. A medical image processing system according to any one of claims 26 to 29, the data storage device further comprising a diagnostic marker calculation component comprising non-transitory instructions operative by the processor to extract peak velocity values for each of the reference points.

31. A medical image processing system according to claim 30, wherein the diagnostic marker calculation component further comprises non-transitory instructions operative by the processor to determine a time to peak velocity for each of the reference points.

32. A medical image processing system according to claim 31 wherein the diagnostic marker calculation component further comprises non-transitory instructions operative by the processor to determine a standard deviation of the time to peak velocities for the reference points.

33. A medical image processing system according to any one of claims 20 to 32, wherein the medical image data is cardiovascular magnetic resonance imaging data.

34. A medical image processing system according to any one of claims 20 to 33 wherein the tracking component is further comprises non-transitory instructions operative by the processor to :

receive a user selection of a mask region centered on a point in a first image of the sequence of images;

perform a search over a search region of a second image of the sequence of images to identify a region of the second image that is a best match to the mask region; and

determine the location of the reference point in the second image as the location of the region of the second image that is a best match to the mask region.

35. A medical image processing system according to claim 34 wherein the tracking component further comprises non-transitory instructions operative by the processor to perform a search over the search region by computing a normalized cross correlation at each location in the search region between the region centered around that location and the mask region.

36. A medical image processing system according to any one of claims 20 to 35, wherein the views are long-axis views of the right ventricular region of the subject.

37. A medical image processing system according to claim 36, wherein the views are one or more of the following: a four-chamber view; a right ventricle outflow view; and a right ventricle two-chamber view.

38. A medical image processing system according to any one of claims 20 to 37, wherein the reference points are one or more of the following: the right ventricle septal site; the right ventricle lateral site; the right ventricle anterospetal site; the right ventricle posterolateral site; the right ventricle anterior site and the right ventricle posterior site.

39. A non-transitory computer readable medium carrying computer executable instructions which when executed on at least one processor cause the at least one processor to carry out a method according to any one of claim 1 to 19.