WO2008063081A2

WO2008063081A2 - A method for visualising a three-dimensional image of a human body part

Info

Publication number: WO2008063081A2
Application number: PCT/NO2007/000413
Authority: WO
Inventors: Andreas Abildgaard
Original assignee: Oslo Universitetssykehus Hf
Priority date: 2006-11-23
Filing date: 2007-11-22
Publication date: 2008-05-29
Also published as: WO2008063081A3

Abstract

The present invention relates to a method for visualising a set of voxel data on a 3D screen. The method comprises providing a set of voxel data comprising spatial information for each voxel point, and, for each voxel point, a corresponding set of voxel values. Secondly, there is selecting a second volume (2V) from the first volume (IV), the second volume being a sub-volume of the first volume. Subsequently, there is applied, on the sub-set of voxel data within the second volume (2V), a substantially continuous and spatially non-uniform modulation function (MF) a voxel value along a viewing direction, the applied modulation function (MF) being a direct function of the spatial coordinate along the viewing direction. Finally, there is generated an image on the 3D screen along the viewing direction from the sub-set of modulated voxel data within the second volume (2V). The method provides a relatively simple and effective method for searching for abnormalities or visualizing the anatomy in a human body part in a quite intuitive manner due to the non-uniform modulation of the second volume. According to preliminary tests performed by the inventor, the invention also provides an improved detection and identification of abnormal tissue structures and/or forms, such as pulmonary nodules, and various pathological changes in solid tissues, skeleton and vessels.

Description

A method for visualising a three-dimensional image of a human body part

Field of the invention

The present invention relates to a method for visualising a set of voxel data on a screen, the screen being capable of providing a three-dimensional image (3D), the set of voxel data representing a three dimensional image of at least a part of a human. The invention also relates to a corresponding visualisation system, comprising a screen and a connected processing means. The invention further relates to a computer program product for implementing the method on a visualisation system with a processing unit.

Background of the invention

Medical imaging systems such as X-ray computerized axial tomography (CT) and magnetic resonance imaging are capable of producing exact cross-sectional image data that express the physical properties related to electron and nuclear density, respectively, of the human body. Reconstruction of three-dimensional (3D) images using collections of parallel two- dimensional images representing cross- sectional data has been applied in the medical field for some time.

Various rendering techniques are applied for three-dimensional display of medical images. One such technique is surface rendering. A limitation of this technique is that it does not adequately visualize the tissue within solid organs; it is optimised for visualisation of surfaces and boundaries.

Another rendering technique is volume-rendering. Instead of overlaying surfaces using a complex model of three-dimensional data, volume-rendering relies on the assumption that three-dimensional objects are composed of basic volumetric building blocks, so-called "voxels". The voxels are three-dimensional analogs to the more familiar two-dimensional pixels (abbreviation for picture element) used for normal screens, and similarly a voxel has a spatial coordinate and associated voxel values like signal intensity, transparency, assigned colour, and so forth. By using a set of voxel data it is possible by volume rendering to provide an image of the body part of the patient under examination by various volume rendering techniques such as image ordering (also called ray casting), object ordering (also called compositing or splatting), and domain rendering (by e.g. Fourier transformation). Associated projection types for obtaining an image on a two- dimensional screen include maximum intensity projection (MIP), minimum intensity projection (MIN-IP), iso-surface projection, and transparency projections.

Unfortunately, these volume rendering techniques have to be specially designed and tailored to the specific type of tissue and/or the searched abnormality in the tissue. Thus, appropriate volume rendering settings or projection types for detection of pulmonary nodules in CT images or analysis of fractures in skeletal CT images may differ significantly.

In applications where the performance of these rendering techniques in diagnostic imaging can be measured, it has been shown that presently available techniques have limitations. One such application is the detection of pulmonary nodules in CT images. The use of MIP reconstruction can improve the detection of small pulmonary noduli, but even with this technique a number of nodules remain undetected (Naidich 1993, Gruden 2002). In general, there is an inherent trade- off between the possibility of seeing depthness of especially elongated structures in the tissue and the problem with over-projection of normal structures on focal pathology. For solid tissues in particular, e.g. brain and liver tissue, it is relatively difficult to access the correspondingly large amount of data, which makes it quite time-consuming in terms of man-power to use medical imaging techniques for such tissues.

Realizing that end-users of medical imaging systems are not trained for exploiting the more refined details of volume rendering techniques, and thereby obtain the associated advantages, there is a need for a simple and effective volume rendering tool for medical imaging.

Recently, true 3D displays for visualisation of radiological images have been introduced. Some of the 3D display devices provide a true three-dimensional visualisation based on rapid cycling between multiple spatially different oriented views, see for instance WO 2007/119064 and WO 2005/112474 to Setred AS. These display devices are "true" 3D displays, unlike the stereoscopic displays that have existed for several decades. The stereoscopic displays require the use of special goggles ("glasses") with polarisation or special colours, whereas these modern autostereoscopic 3D display devices function without such requirements. In addition, these improved 3D displays provide true 3D visualisation where the view changes e.g. when the observer moves his head horizontally.

The use of 3D modelling of radiological image data sets for 2D display is typically applied for vascular structures such as arteries and veins, skeletal structures, the biliary tree and nerve paths in the central nervous system. These anatomical structures are suited for modelling on a 2D screen because they represent long, often branching structures that interconnect and have defined surfaces, surrounded by zones that are not of interest and are shown only as "empty space" between the structures of interest. The well defined borders between the structures of interest and the surrounding space are fundamental for the common techniques used for visualisation of 3D models on 2D screens. The most important depth cues in the models are:

1) The use of artificial lightening and shadowing, and

2) Structure continuation or discontinuation (continuation in the front of the model, discontinuation of structures further behind).

Use of these techniques requires that the anatomical data contains defined borders between structures of interest and the surrounding space. Therefore, 3D models are not much used for 2D visualisation of continuous, solid tissues such as the abdominal organs. The condensing of signal from a solid tissue onto one single 2D image causes a signal mixture that is very difficult to interpret because of the lack of depth cues. The cues mentioned cannot be used for solid tissues, because there are no relevant borders in the solid tissue that can be depicted meaningfully with lightening effects or surface continuity.

In Academic Radiology, 12, 12, December 2005, 1512-1520, Wang et al., disclosed a method for lung nodule inspection based on generated stereoscopic images from CT scans. However, from their distance-based MIP analysis there was only obtained an increase in contrast of nodules relative to background in some cases, and there was not reported any significant improvement of readings with the 3D images as compared to conventional 2D MIP. Thus, the reference does not document a clinical usefulness of the method.

Hence, an improved method for visualising a three-dimensional image of a human body part would be advantageous, and in particular a more efficient and/or reliable method that would improve the detection rate for pathology would be advantageous.

Summary of the invention

Accordingly, the invention preferably seeks to mitigate, alleviate or eliminate one or more of the above mentioned disadvantages singly or in any combination. In particular, it may be seen as an object of the present invention to provide a method that solves the above mentioned problems of the prior art related to three-dimensional visualisation of the human body or parts of the human body.

This object and several other objects are obtained in a first aspect of the invention by providing a method for visualising a set of voxel data on a screen, the screen being capable of providing a three-dimensional image (3D), the set of voxel data representing a three dimensional image of at least a part of a human body in a first volume (IV), the method comprising;

- providing a set of voxel data comprising spatial information for each voxel point, and, for each voxel point, a corresponding set of voxel values, - selecting a second volume (2V) from the first volume (IV), the second volume being a sub-volume of the first volume,

- applying, on the sub-set of voxel data within the second volume (2V), a substantially continuous and spatially non-uniform modulation function (MF) on one or more type(s) of voxel values along a viewing direction, the applied modulation function (MF) being a direct function of the spatial coordinate along the viewing direction, and

- generating an image on the screen along the viewing direction from the sub-set of modulated voxel data within the second volume (2V). The invention is particularly, but not exclusively, advantageous for obtaining a relatively simple and effective method for searching for abnormalities or visualizing the anatomy in a human body part in a quite intuitive manner due to the non-uniform modulation of the second volume. According to preliminary tests performed by the inventor, the present invention also provides an improved detection and identification of abnormal tissue structures and/or forms, such as pulmonary nodules, and various pathological changes in solid tissues, skeleton and vessels.

The improved depth perception and/or lowering of the amount of over-projection provided by the present invention may result in a better ability to detect focal pathology and other abnormalities than was hitherto possible with rendering methods within medical imaging on three-dimensional (3D) screens.

Additionally, diagnostic procedures may be more efficient with the present invention as the non-uniform modulation within the second volume provides the user with an improved over-view of the part of the tissue being examined. In addition, use of the current invention provides the user with visual clues to the position of the centre of the second volume relative to the first or total volume being studied, thereby improving the perception of the location of focal pathology along the viewing direction and also the accuracy of 3D navigation within the first volume.

The nonlinear modulation of signal intensity along the viewing direction may be used to highlight a central zone in the second volume (2V), thus providing visual clues to the specific location of the second volume within the first volume (IV). In the following this will also be denoted a so-called "focus zone". This focus zone may be made narrow, and thus provide a detailed anatomical depiction of a localised part of the second volume (2V). This may be advantageous for analysis of pathology observed in the data. The second volume (2V) may also be termed a "subvolume" within the context of the present invention. Particularly, the second volume (2V) may be relatively narrow i.e. if the second volume is spatially positioned within a short distance. In that case, the second volume (2V) may be termed a "slab". Similarly, the first volume (IV) may also be termed a "total volume" in the context of the present invention, though the first volume (IV) may still be a subset of an even larger volume.

Stereoscopic displays /3D displays that rely on the use of special goggles (polarized or coloured) have been used for visualisation of radiological images in anatomical education, surgical planning , and during surgical procedures (See for instance Mitchell P, Wilkinson ID, Griffiths PD, Linsley K, Jakubowski J. A stereoscope for image-guided surgery. Br J Neurosurg. 2002 Jun;16(3):261-6).

It has also been shown that visualisation of 3D anatomical models on ordinary 2D displays are useful during surgery. In neurosurgery such visualisation has been advantageously combined with various techniques for positional support of the surgery (neuronavigation), (see for instance Schmid-Elsaesser R, Muacevic A, Holtmannspόtter M, UhI E, Steiger HJ Neuronavigation based on CT angiography for surgery of intracranial aneurysms: primary experience with unruptured aneurysms. Minim Invasive Neurosurg. 2003 Oct;46(5):269). It can be advantageous to co-register a video recording of the operative field with a rendering of the underlying anatomy based on preoperative radiological imaging. (Gildenberg PL₇ Labuz J. Use of a volumetric target for image-guided surgery. Neurosurgery. 2006 Sep;59(3):651-9).

Presently, no functioning 3D display that can be used without special goggles in a sterile operating theatre are generally available. Therefore, it is not feasible at present to do a clinical comparison of the current invention for visualisation on a 3D screen compared to currently used models and 2D screens. Based on the fact that surgeons find stereoscopic / 3D displays advantageous compared to 2D displays for understanding radiological anatomy, it is to be expected that this will also be the case for preoperative visualisation when this option becomes possible.

In this type of preoperative visualisation mentioned above, the anatomical volume that shall be displayed is a subvolume of the total anatomical volume that has been covered by the preoperative radiological investigation.

Some properties that characterise the second volumes i.e. the subvolumes made with the present invention can be expected to be advantageous in this type of preoperative visualisation: 1) With the signal modulation used in the present invention, the depicted subvolume can be made larger along the viewing direction) than with the commonly employed techniques (MIP, Volume Rendering) for tissue rendering, because the signal modulation prevents the signal overflow that otherwise will tend to occur with large subvolumes.

2) The possibility for navigation (automatic or manual movement of the subvolume (second volume) relative to the total anatomical volume (first volume)) is crucial for preoperative use, because the surgeon needs instant adjustment of the position and angulation of the displayed subvolume (second volume, 2V) relative to the total volume (first volume, IV). The focal zone that can be obtained in the centre of subvolumes (second volume, 2V) made with the present invention allows for exact manual navigation through the total volume (first volume, IV).

In one embodiment, the first volume (IV), or parts thereof, may be displayed on the screen together with the image generated from the second volume (2V) in order to improve e.g. navigation around in the image data.

In an embodiment, the second volume (2V) may be displaced, either rotationally or translationally, by a viewer along the viewing direction. Thus, the viewer may initiate or directly control a displacement of the second volume. In particularly, the viewer may perform a rotation of the second volume (2V) around an axis positioned in central position in the second volume (2V), or alternatively around an axis positioned in a central position relative the three dimensional (3D) screen seen by the viewer. Preferably, the second volume (2V) may be displaced by a viewer in a substantially continuous manner along the viewing direction. Additionally, the size of the second volume (2V) may be substantially unchanged during displacement by the viewer. The so-called focus zone mentioned above is advantageous for this navigation, because it provides a visual clue to the position of the rotational axis when the second volume (2V) is rotated with respect to the first volume (IV). Another advantage provided by this embodiment may be the possibility to visualise a large volume without signal overflow. Experience by the present inventor indicates that the size of the second volume (2V) that can be meaningfully visualised can be in the order of 2 to 3 times the size of a meaningful MIP subvolume. The parts of the second volume that has e.g. a lower signal than the focus zone will be displayed less distinctly, but still aids the viewer in the appreciation of the anatomy and the location of this part of the second volume (2V) i.e. the focus zone.

The type of voxel value being modulated may be selected from: signal intensity, transparency, and colour value. The non-uniform modulation function (MF) may have a maximum point or a minimum point along the viewing direction. Possibly, the maximum point or the minimum point can define a central position of the second volume if the modulation function is symmetric around such a central position. More particularly, the maximum point or the minimum point may not be positioned at an end or a front portion of the second volume; preferably the maximum point or the minimum point may be positioned within a central portion constituting 20%, 30%, 40% or 50% of the second volume (2V).

Preferably, the non-uniform modulation function (MF) may have a substantially symmetric shape around the maximum point or the minimum point because the present inventor has found that this provides the best "focus" zone. More preferably, wherein the non-uniform modulation function (MF) has a maximum point along the viewing direction, and the modulation function is constantly decaying away from the maximum point. The maximum/minimum point may be an absolute maximum/minimum.

The performed modulation may be a so-called "soft" modulation meaning that the modulation is relatively slowly changing the relevant voxel value within the second volume (2V). More specifically, the ratio of the derivative of the modulation function (MF) to the modulation function (MF) may be maximum 5%, maximum 10%, maximum 20%, maximum 30%, maximum 40%, or maximum 50 %. Alternatively or additionally, the 2^nd derivative of the modulation function may be constantly or monotonously increasing/decreasing. Some preferred shapes of the non-uniform modulation function may be shapes chosen from: a bell shape, a Gaussian shape, or a parabolic shape. Other similar mathematical function may readily be found to be suitable for use within the context of the present invention.

In an alternative or additional embodiment, the modulation function (MF) may be adapted to perform a local averaging process within the second volume by smoothing the one or more voxel value(s) within the second volume. This is particularly beneficial for imaging of solid or dense tissue.

Typically, the set of voxel data may comprise a plurality of slices, where each slice represents a cross-sectional portion of the part of the human body being examined. However, other imaging techniques are possible i.e. various projection methods not based on slices segmentation may be applied within the context of the present invention. The second volume (2V) may then comprise at least 10 slices, preferably at least 20 slices, or even more preferably at least 30 slices. Alternatively, the second volume (2V) may comprise a maximum of 50 slices, preferably a maximum of 80 slices, or even more preferably a maximum of 100 slices.

A slice may represent a cross-sectional portion with a thickness of approximately 2 mm, preferably a thickness of approximately 1 mm, or even more preferably a thickness of approximately 0.5 mm.

The part of the human body being examined may contain substantial amounts of solid tissue. More specifically, solid tissue may be operationally defined as tissue where every or nearly every voxel point should be studied in order to examine the solid tissue appropriately. The present invention in particular provides an efficient tool for visualising solid tissue due to the improved depth perception and/or the lowering of the amount of over-projection. Advantageously, the part of the human body may therefore be an organ parenchyma selected from: liver, pancreas, spleen, and kidney. Alternatively, the part of the human body may be selected from: the musculoskeletal system, the central nervous system, the cardiovascular system, urinary tract, lymphatic system, and bile ducts.

The image may be generated on the screen by a volume-rendering method e.g. image ordering (also called ray casting), object ordering (also called compositing or splatting), or domain rendering (by e.g. Fourier transformation). Also methods like maximum intensity projection (MIP), minimum intensity projection (MIN-IP), iso-surface projection, and transparency projections can be applied.

In one embodiment, the three-dimensional (3D) screen may also be capable of providing a two-dimensional image (2D) if needed. Appropriate types of 3D screens include, but are not limited to, stereoscopic display devices, auto- stereoscopic display devices, holographic display devices, volumetric display devices. Three-dimensional imaging (3D) may particularly advantageously be applied within the visualising of substantially solid tissue, because the present invention provides an improved perception of depth and/or lowering of the amount of over-projection in the image.

A 3D screen provides direct appreciation of signal positions along the depth direction in the voxel dataset. Hence there is less need for the visual clues are needed as compared to the case where 2D screens are used for 3D dataset visualisation. A specific feature of a 3D display, as compared to 2D displays, is the ability to provide meaningful visualisation of selected small volumes from anatomical structures that are more continuous and solid, such as tissues and organs. This feature may also make it possible to navigate within the voxel dataset obtained from solid tissues.

In one embodiment, the three-dimensional image (3D) screen may be an autostereoscopic display device and comprises a switchable aperture with horizontal slits positioned in front of a two-dimensional (2D) screen.

In another embodiment, the three-dimensional image (3D) screen is an autostereoscopic display device and comprises at least a first and a second switchable aperture positioned in front of a two-dimensional (2D) screen. For more details on this type of 3D screens the reader is referred to WO 2007/119064.

In general, it may be advantageous that there is further generated an additional image on the three-dimensional (3D) screen along the viewing direction from the set of voxel data within the first volume (IV). This may provide a kind of background image for the image generated from the second volume (2V), which is particularly advantageous for navigation within a large and/or complex image data set, especially with solid or dense tissue.

The set of voxel data may be provided by an imaging technique such as magnetic resonance imaging (MRI), computed tomography (CT), positron electron tomography (PET), single photon emission computed tomography (SPECT), ultrasound scanning, rotational angiography, positron electron tomography computed tomography (PET-CT), positron electron tomography magnetic resonance (PET-MR), and tomo-synthesis. Also other combinations of these imaging modalities may be applied within the context of the present invention.

Though the present invention is aimed at medical imaging, it is contemplated that the present invention may additionally be exploited within other technical fields where computer-aided reconstruction and design are applied. Such fields include, but are not limited to, architecture, geology, mechanical designing, and so forth. Thus, it is contemplated that the part of the human body represented by the set of voxel data in the first volume could instead by a building part, a geological body or object, a mechanical body or object, and so forth.

In a second aspect, the invention relates to a visualisation system for visualising a set of voxel data on a screen, the screen being capable of providing a three- dimensional image (3D), the set of voxel data representing a three dimensional image of at least a part of a human body in a first volume (IV), the system comprising:

- recording means for storing a set of voxel data comprising spatial information for each voxel point, and, for each voxel point, a corresponding set of voxel values,

- selecting means for selecting a second volume (2V) from the first volume (IV), the second volume being a sub-volume of the first volume,

- processing means for applying, on the sub-set of voxel data within the second volume (2V), a substantially continuous and spatially non-uniform modulation function (MF) on one or more type(s) of voxel values along a viewing direction, the applied modulation function (MF) being a direct function of the spatial coordinate along the viewing direction, and - image generating means capable of generating an image on the three- dimensional (3D) screen along the viewing direction from the sub-set of modulated voxel data within the second volume (2V).

In a third aspect, the invention relates to a computer program product being adapted to enable a computer system comprising at least one computer having data storage means associated therewith to control an visualisation system according to the second aspect of the invention.

This aspect of the invention is particularly, but not exclusively, advantageous in that the present invention may be implemented by a computer program product enabling a computer system or similar processing means to perform the operations of the first aspect of the invention. Thus, it is contemplated that some known visualisation system may be changed to operate according to the present invention by installing a computer program product on a computer system controlling the said visualisation system. Such a computer program product may be provided on any kind of computer readable medium, e.g. magnetically or optically based medium, or through a computer based network, e.g. the Internet.

The first, second and third aspect of the present invention may each be combined with any of the other aspects. These and other aspects of the invention will be apparent from and elucidated with reference to the embodiments described hereinafter.

Brief description of the Figures

The present invention will now be explained, by way of example only, with reference to the accompanying Figures, where

Figure 1 is a schematic drawing of a patient being inserted into a scanning device according to the present invention,

Figure 2 is a schematic illustration of a scanning device for obtaining computed tomography (CT) images by attenuation of X-rays from an X-ray source that rotates around the patient, JL O

Figure 3 is a schematic illustration of how a sequence of so-called slices is combined into an image on a screen,

Figure 4 is a 3D reconstruction of cross-sectional CT images of the lungs in a patient made by a maximum intensity projection (MIP),

Figure 5 is a schematic drawing of a method for visualising a set of voxel data on a screen according to the present invention,

Figure 6 is an example of a modulation function for performing a substantially continuous and spatially non-uniform modulation according to the present invention, and

Figure 7 is also a cross-sectional CT MIP image, similar to Figure 4, of the lungs in a patient but the modulation function of Figure 6 is applied according to the present invention.

Detailed description of an embodiment

The following definitions will be used:

A "voxel" is a unit cube with a unit vector along the X-axis, a unit vector along the

Y-axis, and a unit vector along the Z-axis of a three-dimensional image.

A "part of the human body" is an anatomical region of interest. The part of the human body is represented as a plurality of voxels having voxel signal intensity (e.g. CT values) in the range of specified signal intensity included in a spatial region specified for the said part of the human body.

Figure 1 is a schematic drawing of a patient 50 being inserted into or positioned in a scanning device 100 according to the present invention. A part 51, e.g. the head as shown, of the patient 50 is inserted into the scanning device 100 for medical imaging, though the entire patient could also be inserted in a scanning device if designed accordingly. The scanning device 100 could be any type of medical imaging devices such as magnetic resonance imaging (MRI), computed tomography (CT), positron electron tomography (PET), single photon emission computed tomography (SPECT), ultrasound scanning, and rotational angiography, but in the following the teaching of the present invention will be demonstrated with respect to X-ray attenuated computed tomography (CT) for illustrative purposes.

Figure 2 is a schematic illustration of a scanning device 100 for obtaining computed tomography (CT) from a centrally positioned patient 50. The CT scanning device 100 comprises an X-ray source 110 and a two-dimensional X-ray detector 120 for measuring the attenuation of the X-ray, indicated by X, through the patient 50 at the shown peripheral position of the source 110 and the detector 120. The X-ray source 110 and a two-dimensional X-ray detector 120 are rotationally displaceable around the patient 50 as indicated by the two curved arrows to a desirable peripheral position V. It is then possible to do a reconstruction of a stack of transversal 2D images based on recordings of X-ray attenuation from the detector 120 from a continuous range of angles V around the patient 50. Essentially, the X-ray absorption coefficients are processed into a spatial distribution of CT values as it is well known in the art. Subsequently, reformatting slices into other planes can also be performed.

Figure 3 is a schematic illustration of how a sequence of so-called slices Sl, S2, S3, and S4 is combined into an image 300 on a screen 200 for CT imaging of lungs.

The slices Sl, S2, S3, and S4 are typically two-dimensional images acquired by a scanning device 100 as shown in Figures 1 and 2. Each slice Sl, S2, S3, or S4 thereby represents a cross-sectional set of voxel data i.e. spatial coordinates with corresponding voxel (CT) values. The thickness of the slices may be from 0.1 mm to 2 mm. For illustrative purposes only four slices are shown in Figure 3, but any number of slices, which the scanning device 100 can record, may be used in the context of the present invention.

The set of slices Sl, S2, S3, and S4 constitutes a set of voxel data, the set of voxel data represents a three dimensional image of at least a part of a human body, i.e. the lungs, in a first volume (IV). By volume rendering, a projection of the set of voxel data along the viewing direction 310 can be generated and shown as an image 300 on the three-dimensional (3D) screen 200.

An autostereoscopic 3D display device can be implemented by synchronising a high frame rate screen for displaying a two dimensional image with a fast switching shutter. If each frame on the screen is synchronised with a corresponding slit, and the images and slits are run at sufficient speeds to avoid flicker, typically 50 Hz or above, then a 3D image can be created. When the high frame rate screen is viewed through one open slit of the shutter, each eye sees a different part of the screen, and hence each eye sees a different part of an image displayed on the screen. One image is displayed on the screen whilst the corresponding slit is open. Similarly, when a second slit is open another corresponding frame is displayed. By repeating the process sufficiently fast, such that each slit is perceived as flicker- free, the entire shutter will represent a window into a 3D scene. It is assumed that the individual images displayed represent the correct perspectives through each slit. For reference on an implementation of such a 3D display device, the reader is referred to WO 2007/119064 and WO 2005/112474, which are both hereby incorporated by reference in their entirety.

The arrow 310 indicates a viewing direction of the image 300. The volume rendering is performed along the arrow 310 by e.g. image ordering, object ordering or domain rendering or any other suitable volume rendering. For performing a volume rendering, a transfer function can be applied on the set of voxel data so as to define colour and opacity and other relevant parameters for the imaging process. In the following, the maximum intensity projection (MIP) will be applied for illustrative purposes.

Figure 4 is a 3D reconstruction of cross-sectional CT images of the lungs in a patient made by a maximum intensity projection (MIP) as seen on a two- dimensional screen because three-dimensional view cannot easily be represented on paper. The reconstruction is made from 20 axial CT images with a distance of 1 mm. For MIP images, the voxel value with the maximum intensity along a direction parallel to the viewing direction 310 is set as the sample value of the image 300 for that direction. The MIP technique is simple and requires no depth cues. MIP is typically used for visualising long, thin objects such as blood vessels or for visualising a sub-volume of a lung tissue.

Figure 5 is a schematic drawing of a method for visualising a set of voxel data on a screen 300, the set of voxel data representing a three dimensional image of at least a part of a human body 50 in a first volume IV. Initially, a set of voxel data comprising spatial information for each voxel point is provided, and, for each voxel point, a corresponding set of voxel values. Secondly, a second volume 2V is selected from the first volume IV, the second volume thereby being a sub-volume of the first volume as shown in Figure 5. On the sub-set of voxel data within the second volume 2V, there is applied a substantially continuous and spatially nonuniform modulation function MF on one or more type(s) of voxel values, e.g. voxel intensity or transparency, along a viewing direction 310. The applied modulation function MF is a direct function of the spatial coordinate along the viewing direction 310. Finally, there is generated an image 300 on the three-dimensional (3D) screen 200 along the viewing direction 310 from the sub-set of modulated voxel data within the second volume 2V, which can be seen and examined by the viewer 400.

As indicated by the two oppositely pointing, dashed arrows next to the second volume 2V, the second volume 2V can be displaced by a viewer 400 along the viewing direction 310. The viewer may use a computer interaction device such as a mouse or keypad for displacing the second volume 2V. The second volume can thereby be displaced by a viewer 400 in a substantially continuous manner along the viewing direction 310. During the displacement, the size of the second volume 2V is substantially unchanged. It should be mentioned that the viewing direction can also be changed by a viewer 400.

Figure 6 is an example of a modulation function MF for performing a substantially continuous and spatially non-uniform modulation according to the present invention. The horizontal axis is a spatial coordinate designated X which is parallel to the viewing direction 310 in question. The vertical axis indicates the relative reduction of the voxel value being modulated, thus 100% relative reduction corresponds to complete reduction, whereas 0% relative reduction corresponds to no reduction of the voxel value. Various mathematical functions are applicable within the teaching of the present invention. However, tests performed by the inventor indicate that bell-shaped functions, either inverted or normal, work well for modulation of the voxel values within the second volume 2V. To provide the viewer 400 with an improved overview of the body part 50, it is also important that the modulation function MF is smooth and gradually changing within the second volume 2V. While the set of voxel data is inherently discrete, and therefore discontinuous, due to the limited resolution and/or sampling of the imaging technique applied, the modulation function can be substantially continuous and modify the voxel values at the discrete spatial positions where the modulation function MF is applied.

Figure 7 is also a cross-sectional CT MIP image similar to Figure 4 (thus also displayed on a two-dimensional screen for illustrative purposes) of the lungs of a patient but additionally the modulation function MF of Figure 6 is applied according to the present invention. Upon comparison of Figures 4 and 7, it is apparent that the present invention in particular provides a relative enhancement of the centre of the second volume 2V while preserving the visualising of depth in the image 300. The improved image quality enables better identification of nodules in the lungs. In Figure 7, two nodules indicated by arrows Al and A2 are clearly visible, whereas these nodules are hardly visible in Figure 4 which is a conventional CT MIP image.

In order to quantify the advantages obtained by the present invention, a preliminary test was performed by diagnostic reading of two-dimensional images generated with three different rendering techniques from chest CT slices. 52 simulated round lesions ranging between 2 and 8 millimetres in size were randomly placed in the lung fields of ten different slice positions with 30 mm intervals. Original CT slices were reconstructed to 0.625 millimetres thickness, 1 millimetre distance. Table 1:

As it is indicated from Table 1, the present invention provides 1) more true positive findings of lesions, 2) a comparably number of false positive finding of lesions, and 3) a substantially lower number of false negative finding of lesions in the lungs, as compared to the conventional MIP methods. Table 1 only represents a preliminary test, but more substantiated tests have supported the results of Table 1; Andreas Abildgaard et al., Improved visualisation of artificial pulmonary nodules with a new subvolume rendering technique, 57^th Nordic Conference of Radiology and 18^th Nordic Conference of Radiography, Malmø, Sweden, 9-12 May, 2007.

There are not yet completed full-scale tests of the present invention on a three- dimensional screen, but preliminary experiments and pilot tests indicate that the findings will be comparable or even better than for two-dimensions, cf. Table 1 above, Preliminary tests indicate that the invention will be particularly advantageous when solid tissues are visualised on 3D screens. In this embodiment anatomical structures that are completely invisible on 2D screens have been made distinctly visible on a prototype 3D screen using the present invention.

A practical example of an application of medical visualisation with a 3D display device is the study of the image data from the liver, biliary tract and pancreas in patients with focal pathology in this area. For radiological or surgical anatomical ^ 7 000413

analysis of this region, the appreciation of the anatomical relationship between the structures is mandatory.

With the use of 2D screens, this can be achieved by either looking at separate, thin images generated from the voxel data that depict the anatomically relevant structures in the slice, or by looking at 3D models of the vessels or bile ducts.

With a 3D screen however, it is the experience of the present inventor that it is possible also to scroll through the anatomic data using a second volume 2V that shows all the relevant anatomical structures simultaneously. By rotation of the second volume 2V, the viewer can then scroll in another direction, and thereby obtain an improved appreciation of the actual anatomy.

The invention can be implemented in any suitable form including hardware, software, firmware or any combination of these. The invention or some features of the invention can be implemented as computer software running on one or more data processors and/or digital signal processors. The elements and components of an embodiment of the invention may be physically, functionally and logically implemented in any suitable way. Indeed, the functionality may be implemented in a single unit, in a plurality of units or as part of other functional units. As such, the invention may be implemented in a single unit, or may be physically and functionally distributed between different units and processors.

Although the present invention has been described in connection with the specified embodiments, it is not intended to be limited to the specific form set forth herein. Rather, the scope of the present invention is limited only by the accompanying claims. In the claims, the term "comprising" does not exclude the presence of other elements or steps. Additionally, although individual features may be included in different claims, these may possibly be advantageously combined, and the inclusion in different claims does not imply that a combination of features is not feasible and/or advantageous. In addition, singular references do not exclude a plurality. Thus, references to "a", "an", "first", "second" etc. do not preclude a plurality. Furthermore, reference signs in the claims shall not be construed as limiting the scope.

Claims

1. A method for visualising a set of voxel data on a screen, the screen being capable of providing a three-dimensional image (3D), the set of voxel data representing a three dimensional image of at least a part of a human body in a first volume (IV), the method comprising:

- generating an image on the three-dimensional (3D) screen along the viewing direction from the sub-set of modulated voxel data within the second volume (2V).

2. A method according to claim 1, wherein the second volume (2V) is displaceable by a viewer along the viewing direction.

3. A method according to claim 2, wherein the second volume is rotatable by a viewer around a rotational axis.

4. A method according to claim 2 or claim 3, wherein the size of the second volume (2V) is substantially unchanged during displacement and/or rotation by the viewer.

5. A method according to claim 1, wherein the type of voxel value being modulated is selected from: signal intensity, transparency, and colour value.

6. A method according to claim 1, wherein the non-uniform modulation function has a maximum point or a minimum point along the viewing direction. ^

7. A method according to claim 6, wherein the maximum point or the minimum point is not positioned at an end or a front portion of the second volume, preferably the maximum point or the minimum point being positioned within a

5 central portion constituting 20%, 30%, 40% or 50% of the second volume (2V).

8. A method according to claim 6, wherein the non-uniform modulation function has a substantially symmetric shape around the maximum point or the minimum point.

10

9. A method according to claim 6, wherein the non-uniform modulation function (MF) has a maximum point along the viewing direction, and the modulation function is constantly decaying away from the maximum point

15 10. A method according to claim 9, wherein the ratio of the derivative of the modulation function (MF) to the modulation function (MF) is maximum 5%, maximum 10%, maximum 20%, maximum 30%, maximum 40%, or maximum 50 %.

20 11. A method according to any of claims 6-10, wherein the non-uniform modulation function has a shape chosen from: a bell shape, a Gaussian shape, or a parabolic shape.

12. A method according to any of the preceding claims, wherein the set of voxel 25 data comprises a plurality of slices, each slice representing a cross-sectional portion of the part of the human body.

13. A method according to claim 12, wherein the second volume comprises at least 10 slices, preferably at least 20 slices, or even more preferably at least 30

30 slices.

14 A method according to claim 12, wherein the second volume comprises a maximum of 50 slices, preferably a maximum of 80 slices, or even more preferably a maximum of 100 slices. 35

15. A method according to any of claims 12-14, wherein each slice represents a cross-sectional portion with a thickness of approximately 2 mm, preferably a thickness of approximately 1 mm, or even more preferably a thickness of approximately 0.5 mm.

16. A method according to any of the preceding claims, wherein the part of the human body contains substantial amounts of solid tissue.

17. A method according to claim 16, wherein the part of the human body is an organ parenchyma selected from: liver, pancreas, spleen, brain, and kidney.

18. A method according to claim 16, wherein the part of the human body is selected from the musculoskeletal system, the central nervous system, cardiovascular system, urinary tract, lymphatic system, and bile ducts.

19. A method according to any of the preceding claims, wherein the image is generated on the screen by a volume-rendering method.

20. A method according to any of the preceding claims, wherein the three- dimensional image (3D) screen is a stereoscopic display device, an autostereoscopic display device, a holographic display device, or a volumetric display device.

21. A method according to any of the preceding claims, wherein the three- dimensional image (3D) screen is an autostereoscopic display device and comprises a switchable aperture with horizontal slits positioned in front of a two- dimensional (2D) screen.

22. A method according to any of the preceding claims, wherein the three- dimensional image (3D) screen is an autostereoscopic display device and comprises at least a first and a second switchable aperture positioned in front of a two-dimensional (2D) screen.

23. A method according to claim 1, wherein there is further generated an additional image on the three-dimensional (3D) screen along the viewing direction from the set of voxel data within the first volume (IV).

24. A method according to claim 1, wherein the set of voxel data is provided from a technique selected from: magnetic resonance imaging (MRI), computed tomography (CT), positron electron tomography (PET), single photon emission computed tomography (SPECT), ultrasound scanning, rotational angiography, positron electron tomography computed tomography (PET-CT), positron electron tomography magnetic resonance (PET-MR), and tomo-synthesis.

25. A visualisation system for visualising a set of voxel data on a screen, the screen being capable of providing a three-dimensional image (3D), the set of voxel data representing a three dimensional image of at least a part of a human body in a first volume (IV), the system comprising:

- recording means for storing a set of voxel data comprising spatial information for each voxel point, and, for each voxel point, a corresponding set of voxel values, - selecting means for selecting a second volume (2V) from the first volume (IV), the second volume being a sub-volume of the first volume,

- processing means for applying, on the sub-set of voxel data within the second volume (2V), a substantially continuous and spatially non-uniform modulation function (MF) on one or more type(s) of voxel values along a viewing direction, the applied modulation function (MF) being a direct function of the spatial coordinate along the viewing direction, and

- image generating means capable of generating an image on the three- dimensional image (3D) screen along the viewing direction from the subset of modulated voxel data within the second volume (2V).

26. A computer program product being adapted to enable a computer system comprising at least one computer having data storage means associated therewith to control a visualisation system according to claim 25.