Ultrasonic mammography

Wiesław Bicz
  

The goal of the project is to build a device, that can work in a similar way as classical x-ray mammography, but using ultrasonic waves.

Following assumptions was made:

A. It is very good known, that ultrasonic techniques allow to detect cancer, but used with devices, that are already on the market, requires experienced specialists for interpretation of results. Today measurements are made, using classical ultrasonography, that is based on measurement of scattering, caused by impedance differences.
B. There are some published measurement results, showing, that cancerous tissue has different sound velocity than normal one. If local velocity changes could be detected, cancerous tissue can be surely detected too.
C. It is also possible to measure local differences in attenuation and dispersion (frequency dependent sound velocity changes). It can happen, that mapping such differences can show cancerous tissue too.
D. Using the experience, we have collected with holographic techniques for finger recognition, and ideas developed during projects, we have made, it is possible to assume, that it can be possible to achieve much higher resolution, that is possible with classical ultrasonography. And this will allow to detect objects of the size of about 0.1mm.
E. Breast measurement with ultrasound can be made often and pictures obtained allow to make geometrically correct 3D reconstruction. Results obtained in different periods can be compared - this allows to detect changes, that can indicate cancer growth.

There are three general possibilities (terms, that I have used here are not widely used - I am using my own definitions, because there is no common terminology for this problem):

1. Classical ultrasonography:

A narrow beam of ultrasound is used, many kinds of scans are possible (circular, linear, conical, spiral etc.). Combination of different scans can give higher resolution (averaging occur). Classical ultrasonography method is easy to implement, but cannot give very high resolution (there is no possibility to produce very narrow beam of ultrasound). There is no problem with energy and the visualization of the results is relatively easy. The method is time consuming. First scans we have made are of this kind was circular - as shown on the following picture:

Ultrasonography can be also made with the method, that is very popular today - using synthetic aperture and transducer arrays. This method is similar to the point 2, but not as consequent as in this case, because the beam of each transducer has a relatively small angle.

2. Tomography with fan shaped beam:

A beam of sound is used, that is narrow in one surface, but wide in the perpendicular one. This can produce slices of the object. The resolution of the method can be higher than of the first one, but the implementation will be more difficult. The visualization is not very difficult. Measurement must be slower than with the first method (more computations are necessary). There should be no energy problem.

3. Holography:

Sound beam is wide in all directions (spherical wave). This method allows to achieve the highest resolution but the implementation is more difficult. The reasons are:

a) Natural method of visualization of this kind of data is hologram, but no kind of holographic display exist today. Another form of visualization should be found or holographic display should be made.
b) This method can produce measurements even in the real time, but if classic method of visualization should be used, there could be a problem with quickly visualization (large computing power is necessary).
c) The problem with energy can be larger than in the methods 1 and 2. This method needs sender with higher energy and better receiver, but it seems, that the energetic problems won't be too large and can overcome.

4. Holographic scanning:

It is also possible, using classical scanning (1) to use not a narrow, but a wider beam (but not as wide as with the holographic method) and to calculate the picture using the assumption, that the beam has some width and the scattering comes not from a line, but from a conical region. This is a method called often in the literature holographic method. The advantage is: The resolution can be much better than with the first method, but computing power and time needed for picture construction is large.

5. Transmission tomography:

This technique uses transmitters and receivers positioned on opposite sides of the object. It is also possible to use a reflector instead of the receiver. If this technique will be combined with the idea called here as "tomographic scanning" very high resolution can be obtained. It allows to measure local sound velocity, attenuation and dispersion distribution. 

Cancer detection in the classical mammography (using X-rays) is based on the detection of small calcium concentrations (their size can be even lower than 10um). Almost no soft structures can be detected. Today praxis is based not on automatic detection, but on experience of people, interpreting the results.

Classical mammography has some important disadvantages:

- it is possible to make measurements only about one time in a half year;
- cancer detection is based only on the detection of one feature.
- The devices, that are necessary are expensive, requires specially equipped rooms, etc.

The typical opinion about existing ultrasonic methods is:

a) They cannot detect small calcium grains as possible with x-rays. The size, that can be detected is about 0,1mm with the best devices.
b) Ultrasonic methods are very strong dependent on the quality of the devices used and knowledge and experience of people using them.
c) The advantage of the ultrasonic method is: They can detect soft structures and allow to evaluate their position in 3D space.
d) It is very good known, that ultrasound is not dangerous and measurements can be made very often.

Using the device, we propose to develop, it will be possible to eliminate all problems with device quality and knowledge of personal. This device should be able to measure automatically following object parameters (as 3D object map):

A. Sound velocity;
B. Attenuation;
C. Dispersion (sound velocity frequency dependent);
D. Acoustical impedance 
E. From parameters A and D local density can be calculated.

Cancerous structures can be detected similar as with x-ray mammography, if small calcinations can be detected. This can be made, using measurement of parameter D, if the resolution is high enough. And we assume, that it will be possible for us to achieve the resolution of about 0.1mm and to detect even smaller objects due to the scattering, they are causing.

Changes of parameter A and C can probably directly show the cancerous tissues too.

Because all parameters can be observed as 3D matrix and measured often (even every day), it will be possible to compare the results automatically and try not only to detect cancer, using the information about the structure, but also about the changes, that can be detected. This requires some knowledge about the growth of cancerous tissue, that will be available after some results are collected and comparison is possible.

 
The following pictures shows the basic idea of the device:

The breast is smeared with a contact gel and squeezed between two plates (similar as in the classical mammography machine). Behind this plates an ultrasound array, moving transducers or a combination of both of them is placed.

We assume, that the best method of measurement will be the use of a idea, that is shown in the following picture and described in the attached paper about ultrasonic holography:

The device should probably have one measuring head having one (eventually more) sending transducer and two or more circular arrays with receivers: One as shown on the picture, and the second one on the opposite side of the breast. The transducers situated on the same side as the sender will receive information about scattered waves and this on the opposite side about transmitted waves. 

Because the resolution, that we want to achieve, requires a relatively large amount of transducers in different positions, it is better to make a combination: use a limited amount of transducers in two or three circular arrays and additionally move the whole head around the breast and make measurements in different positions. Schematically it is shown on the following picture:

The plates, between them the breast is squeezed must not be plane - they can also be concave, but it is difficult to tell, what shape of this plates would be the best solution.

The device must not be very large, the size of about 50x50x70cm can be sufficient, but it can happen, that motors and other elements, that are necessary for the head movement will need larger space. Computer will need additional space, but it is realistic, that we will need only a standard computer for this device. I assume, that the data collection can take some minutes (may be 3-5), but the evaluation can be longer, and even made with additional computer.
   

 
1. The largest amount of the work requires the software, that will be necessary for reconstruction of pictures from measured signals.
2. It is also necessary to develop the software for evaluation of data with the goal of detection of cancerous tissue.
3. The development of the measuring head will be the second significant development step, that will be necessary here.
4. Additionally the mechanical parts must be developed.
5. It will be surely necessary to make some prototypes, until a properly working device can be presented.
6. Relatively large work will be also necessary for development of electronics for the purpose of this device.


If you are interested in this project, please contact me.


Wieslaw Bicz 

2004/03