ULTRASOUND MACHINE CALIBRATION
The present invention relates to a method and apparatus for calibration of a conventional diagnostic ultrasound machine.
A conventional diagnostic ultrasound machine has a probe which is moved across the surface of a body by a clinician. Ultrasound pulses are sent into the body from points along the surface of the probe and, by measuring the reflections which come back, the machine is able to build up a two-dimensional picture of the tissue in a particular plane within the body. This is called a B-scan because the strength of the ultrasound reflections is used to determine the "brightness" of each point in the image. In order to build up a three-dimensional reconstruction of the inside of a body the position and orientation of each two-dimensional scan (or slice) has to be determined as the scanning takes place. This is conventionally achieved by the use of a magnetic sensing device such as the Fastrak magnetic tracker produced by Polhemus Inc. or a Bird sensor produced by Ascension Technology Corporation. Other types of sensor may be employed, such as optical sensors in which a fixed camera detects the position of markers on the probe (accurate, but expensive and immobile) , acoustic sensors, or else mechanical linkages may be used (accurate, but immobile, cumbersome and obstructive) . Three-dimensional ultrasound imaging is a significantly cheaper procedure than other 3D diagnostic imaging procedures and it is also safer than X- ray computed tomography (CT) as it does not involve exposing the patient to radiation.
However, ultrasound data contains more noise than equivalent images from CT scanning or MRI (magnetic resonance imaging) data. Furthermore, the process of setting up the ultrasound machine with its position and orientation sensor is an involved procedure since, in order to be able to reconstruct the 3D ultrasound image it is
necessary not only to utilise information from the sensor to provide data which defines the relative position and orientation of a mobile part of the position sensor with respect to the fixed datum of the sensor, but also the offset and orientation of the scan plane with respect to the mobile part of the sensor (this is usually expressed in terms of three translations x, y, z) and three Euler angles (azimuth, elevation and roll) and also the x and y scaling of the two-dimensional B-scan image. These further pieces of information are referred to herein as "calibration information" .
Conventionally, the basic way of finding out the offset and orientation of the ultrasound scan plane with respect to the magnetic sensor receiver is to use a simple ruler and a protractor. However, this technique requires considerable skill on the part of the user and a very good knowledge of Euler angles. Clinicians are generally not comfortable doing this. There is also an inaccuracy involved in that it is not possible to tell precisely where the scanning beam comes out of the body of the ultrasound probe .
In order to improve on this, essentially, manual, technique, various groups around the world have developed a technique which involves scanning an array of wires and using the positions of the wires in the scans together with an optimisation algorithm, to calculate the most likely offset and orientation values. Again, however, this is a skilled procedure and meticulous labelling of the positions of the wires in the scan imaging is required. Since there is frequently noise in the two-dimensional images, it is sometimes extremely difficult to do this and it may prove very difficult to get the optimisation algorithm to converge to provide the required calibration data.
The present invention is aimed at overcoming these problems.
According to the present invention therefore a method of providing calibration information for a position and
orientation sensor of an ultrasound scanning machine which comprises a probe producing a scanning beam in a plane, comprises disposing the probe within a positioning device; locating the positioning device on a planar surface; the positioning device having a first component including a spacer and defining a linear feature, the linear feature being fixed in position relative to the spacer and the spacer being shaped such that the linear feature is constrained to move in a plane parallel to the planar surface on which the spacer is positioned for calibration purposes, and a second component with which the probe is engaged, the second component being engagable with the first component and movable relative thereto only in a plane containing the linear feature and including means for fixing the position of the probe within the second component such that the linear feature of the first component is disposed centrally within the scanning beam of the probe; and moving the first and second components over the planar surface and relative to one another while recording position and orientation data produced by the sensor.
Thus, the invention provides a linear feature in the B-scan image which may be detected automatically with standard image processing algorithms. The invention also includes a positioning device having a first component including a spacer and defining a linear feature, the linear feature being fixed in position relative to the spacer and the spacer being shaped such that the linear feature is constrained to move in a plane parallel to the planar surface on which the spacer is positioned for calibration purposes, and a second component with which the probe is engaged, the second component being engagable with the first component and movable relative thereto only in a plane containing the linear feature and including means for fixing the position of the probe within the second component such that the linear feature of the
first component is disposed centrally within the scanning beam of the probe.
Preferably, the linear feature comprises a wire or wires or a beam and the spacer comprises a pair of circular disks or wheels, the wire, one of the wires or the edge of the beam being disposed on the axis of the wheels or disks so that when the wheels move over a planar surface, the wire, the one wire or the edge of the beam is constrained to move parallel to that planar surface. Preferably, the second component comprises a clamp which is arranged to grip "the sides of the ultrasound probe and a pair of side portions containing parallel slots which are a close sliding fit over the wire, the wires or the beam of the first component so that the clamp is constrained to move in a diametral plane of the wheels or disks.
Operating the probe while moving the positioning device over a planar surface enables the position of the probe to be sensed relative to the position of the fixed datum of the sensing device with the linear feature constrained always to lie centrally within the plane of the ultrasound beam so that the position of the linear feature detected within each ultrasound scan can be used (since the linear feature is constrained to move in a plane parallel to the planar surface) to provide accurate position information which can be used to calibrate the position and orientation of the sensor of the ultrasound probe prior to use.
The use of appropriate optimisation algorithms (which themselves do not form a part of the present invention) enables the offset and orientation of the ultrasound scan plane to be determined accurately in relation to the receiver of the position sensor and the computation of the x and y scale factors to be determined. One example of a device according to the present invention will now be described with reference to the accompanying drawings in which:
Figure 1 is a schematic illustration of an ultrasound device;
Figure 2 is a diagrammatic representation of a second component of the position determining device; Figure 3 is a corresponding diagrammatic representation of a first component;
Figures 4(a) to 4(f) show the movements of the probe of an ultrasound device relative to the plane of the linear feature (the phantom plane) that is assumed to be horizontal;
Figure 5 shows an exploded view of an ultrasound device with the positioning device of the invention; and
Figure 6 shows an alternative component of a positioning device of the invention. Figure 1 illustrates an ultrasound device 3 which includes a probe 4 from which is generated a scanning beam of ultrasound, images from the ultrasound machine being fed in use to a computer 5. In order to sense the position and orientation of the probe 4 a position and orientation sensing device such as the Polhemus Faεtrak device 6 is used, the device having a transmitter 7 and a receiver 8, from which signals are sent back to the device 6 to provide position and orientation information to the computer 5.
In use the probe 4 is clamped into the second component 2 of the position determining device illustrated in Figures 2, 3 and 5.
The second component 2 shown in Figures 2 and 5 , comprises a pair of side plates 20 which are arranged to be parallel to one another and which are adjustable towards and away from one another on four adjusting screws 21.
This enables an ultrasound probe to be positioned between the plates and clamped therein by appropriate rotation of the adjusting screws 21. Each of the side plates 20 has an elongate slot 22, the slots in the two plates being parallel to one another and of the same length and width.
The first component 1 comprises a pair of circular disks or wheels 10 which are parallel to one another and
coaxial. Fixed in position between the disks or wheels 10 is a straight beam 11, one edge 12 of which lies on the axis of the wheels 10. The width of the beam 11 is such that it is a close sliding fit within the slots 22 in the end plates 20 of the second component or clamp 2. This means that the clamp is constrained to move in a diametral plane of the disks or wheels 10 and the location of the beam 11 on the wheels 10 means that rotational movement of the first component on a planar surface maintains the edge 12 a fixed distance (the radius of the disks 10) from the planar surface.
In use the ultrasound probe is clamped into the clamp and is positioned so that the beam 11 lies centrally within the scanning plane of the ultrasound beam. For calibration purposes the probe is activated and the positioning device is moved over the planar surface by rolling the first component on the surface and by tipping and translationally moving the clamp relative to the beam 11 in order to provide a series of calibration scans in which the position of the beam 11 is readily determined thus relative to the receiver of the position sensing device.
The data retrieved from the probe during the calibration scans can be used, by means of suitable optimisation algorithms, to determine all the required calibration parameters, including the x and y scale factors.
An example is now given to show how the optimisation algorithm is able to provide the 8 calibration parameters required. These are:
1. The x-direction translation parameter of the transformation from the position sensor to the scan plane of the probe.
2. The y-direction translation parameter of the transformation from the position sensor to the scan plane of the probe.
3. The z-direction translation parameter of the transformation from the position sensor to the scan plane of the probe.
4. The azimuth rotation parameter of the transformation from the position sensor to the scan plane of the probe.
5. The elevation rotation parameter of the transformation from the position sensor to the scan plane of the probe. 6. The roll rotation parameter of the transformation from the position sensor to the scan plane of the probe.
7. The x-direction scale in the ultrasound image.
8. The y-direction scale in the ultrasound image.
Three further parameters have to be calculated by the algorithm internally, although they are not needed as part of the calibration information.
9. The z-direction translation parameter of the transformation from the phantom plane to the datum of the position sensor.
10. The elevation rotation parameter of the transformation from the phantom plane to the datum of the position sensor.
11. The roll rotation parameter of the transformation from the phantom plane to the datum of the position sensor.
The optimisation algorithm needs 11 independent equations to solve for these 11 unknowns. These 11 constraints are provided by automatically detecting the position and orientation of the linear feature in the scan image as the probe is moved through a sequence of calibration scans.
A minimal calibration scan sequence is now described. It is illustrated in Figures 4(a) to 4(f) which show the
movements of the probe relative the plane of the linear feature (the phantom plane) that is assumed to be horizontal.
(a) The initial position of the probe provides one constraint for the optimisation process.
(b) The probe is moved up and down. This provides one constraint.
(c) The probe is rotated to the left and to the right. This provides three constraints. (d) The probe is rotated forwards and backwards. This provides three constraints.
(e) The probe is moved horizontally across the phantom plane in both directions. This provides two constraints. (f) The probe is rotated about a vertical axis. This provides one constraint.
This is a minimal sequence of calibration scans. In practice, as all the processing is automatic, it is easy to use several hundred diverse scan positions and this greatly increases the robustness of the system.
The above example gives one indication of how the algorithms are able to calculate the scale values as well as the transformation from the scan plane to the position sensor. Others are, of course, possible.
To improve the transmission of the ultrasound beam at the interface of the probe 4 and the external media, the calibration process may be carried out in a water bath. Also, in order to improve the accuracy of the calibration, a solution in which sound travels at the same speed as in human tissue can be used instead of water.
To improve image quality, the planar surface, eg. the bottom of the fluid bath, can be roughened to disperse reflected ultrasound waves. It is also important to position the probe 4 in the clamp 2 such that the plane defined by the two slots 22 lies in the centre of the ultrasound beam. This can
normally be achieved with sufficient accuracy by eye. However, an alternative straight beam 11 may be used having wedges 30, as shown in Figure 6.
One pair of wedges 30 is mounted toward one end of the beam 11, and another pair is mounted toward the other end. When the probe 4 is mounted correctly, each pair of wedges 30 will be imaged as a pair of vertical bars having the same height. Should the probe be moved off-centre, then one wedge 30 of the pair will produce a longer vertical bar than the other. Symmetry must be checked for both pairs of wedges 30 to ensure proper alignment.