TISSUE EXAMINATION Background This invention relates to tissue examination, and in particular to detecting foreign structures in, e.g., breast tissue.
One traditional way of examining breast tissue to detect foreign structures such as lumps is to palpate the breast manually. For example, the patient firmly presses on the breast with three fingers while moving the fingers in a circular palpating motion. Typically, such manual breast self-examinations can detect lumps as small as a few centimeters in diameter. Instruments for electronically examining the breast are available. One such instrument includes an array of pressure sensors which is pressed against the breast . Each pressure sensor in the array generates an electrical signal proportional to the local pressure imposed on the sensor when the array is pressed against the breast. When adjacent sensors are positioned across the boundary of a lump within the breast tissue, the sensor that lies over the lump generates an electrical signal indicating the detection of greater local pressure than the adjacent transducer element, which is located over soft tissue alone. The instrument determines whether a lump is present by analyzing the differences between the electrical signals generated by the sensors .
This invention concerns performing tissue examination with a device that includes a sensor which produces signals in response to pressure imposed on the sensor as the sensor is pressed against the tissue. In one general aspect, the tissue is vibrated, the response
of underlying tissue structures to the vibrating is detected by a processor from the signals produced by the sensor, and different types of underlying tissue structures are discriminated from each other based on the response.
Preferred embodiments may include one or more of the following features.
The processor determines the extent to which the underlying tissue structures are vibrated within the tissue based on the signals produced by the sensor, and performs the discrimination based on the extent of the vibration. For example, ribs or other structures within the tissue which are more connected to the tissue than some other structures may not vibrate as freely as other structures which are able to move more freely within the tissue. Moreover, different structures within the tissue may vibrate differently (e.g. have different vibrations and/or amplitudes) . The processor discriminates between the different types of underlying tissue structures based on these response characteristics.
More specifically, a normal tissue structure is discriminated from a potentially foreign tissue structure based on the response the structure to the vibration, and a user is notified as to whether an underlying tissue structure is a potentially foreign tissue structure. For example, different indicators are actuated to indicate whether the underlying tissue structure is a potentially foreign tissue structure (such as non-breast tissue, e.g., a carcinoma) or is a normal tissue structure. Signals corresponding to the amplitudes of the signals produced by the sensor may be displayed to allow the user to visualize the pressure signature.
The tissue is vibrated over a selected frequency range, which may be between 1 Hz. and 200 Hz. The processor determines a resonant frequency of the response
within the range based on the signals produced by the sensor, and an amplitude of the response based on the signals produced by the sensor. This amplitude may be the amplitude of the response at the resonant frequency. The processor discriminates between the different types of underlying tissue structures based on the resonant frequency or the amplitude of the response, or both. The processor may discriminate based on whether the resonant frequency is within a selected portion of the frequency range, whether the amplitude of the resonant frequency exceeds a threshold, or both. The processor may also discriminate based on whether the resonant frequency is outside of a selected portion of the frequency range or whether the amplitude of the resonant frequency is below a threshold.
Preferably, a plurality of sensors are used. The signals produced by the plurality of sensors in response to pressure imposed thereon comprise a frame. The frames of signals from the sensors are successively acquired, and the processor analyzes the frames of signals to detect the response of the underlying tissue structures to the applied vibrations. The processor determines an amplitude of the response at each frequency based on amplitudes of the signals in the frame, and designates the frequency of the response having a highest amplitude as the resonant frequency. The processor stores the frames in a memory and identifies each of the stored frames by the frequency to which the frame corresponds .
The processor determines the resonant frequency of the response by comparing the amplitudes of the signals in each frame to a threshold and determining an average of the amplitudes of the signals that exceed the threshold. The processor then determines the frame having a highest average amplitude, and designates the
frequency to which that frame corresponds as the resonant frequency of the response.
The processor employs so-called "fuzzy logic" techniques to discriminate among normal tissue and different structures. The processor evaluates the resonant frequency with respect to a selected portion of the frequency range, and evaluates the amplitude of the resonant frequency with respect to a threshold amplitude. The processor also determines a pressure profile of the underlying tissue structure based on the amplitudes of the signals in the frame having the highest average amplitude, and evaluates the pressure profile with respect to a base pressure profile. Based on any, some, or all of these evaluations, the processor develops an outcome that indicates a degree of membership of the underlying tissue structure in a class of foreign tissue structures .
In a preliminary mode of operation, the tissue is not vibrated, and preliminary frames of the signals are acquired from the sensors as the sensors are moved over the tissue. The processor analyzes the preliminary frames of signals to determine if the underlying tissue structures are candidates for further analysis. If that is the case, the processor activates a vibrator to apply the vibrations to the tissue and causes additional frames of signals to be successively acquired from the sensors. The processor analyzes these additional frames of signals to detect the response of the underlying tissue structures to the applied vibrations. The processor determines the amplitude of the resonant frequency of the response by reducing the average amplitude of the frame having the highest average amplitude by an amount that corresponds to the response of the underlying tissue detected by the plurality of sensors during the preliminary mode.
Preferably, the vibrator includes an oscillator for generating a driving signal at a selected frequency, and a mechanism for converting the driving signal to mechanical vibrations for vibrating the tissue. The oscillator produces the driving signal over the selected frequency range to cause the mechanism to produce the vibrations over this range. The mechanism includes a pair of plates each of which is configured to be applied against a surface of the tissue. The plates are spaced from each other and disposed at a selected angle with respect to each other to direct the vibrations toward a common point within the tissue.
The plates are mounted to a housing. The sensor or the plurality of the sensors are disposed on a sensor surface of the housing between the plates. The sensor surface is convex. During operation of the vibrator, plates move with respect to the sensor surface between an extended position in which the plates are applied to the surface of the tissue and a retracted position in which the plates are spaced from the surface of the tissue. Advantages of the invention may include one or more of the following advantages. The device is non- intrusive because the device examines the tissue from the surface of the breast and does not require manipulating the breast in an invasive or uncomfortable manner. The device distinguishes possible carcinomas from other structures. The device may also distinguish among different types of foreign structures in the tissue based on their response to vibration. The device requires minimal training for proper operation by the user. The device is inexpensive. The device is more accurate than manual examination of the tissue by a person unskilled in manual tissue examination.
Other advantages and features will become apparent from the following description and from the claims.
Brief Description of the Drawing Fig. 1 is a block diagram of a tissue examination device which includes an oscillating mechanism for vibrating the tissue. Fig. 2 is a diagram of a first embodiment of the tissue examination device.
Fig. 3A is a cross-sectional view of the tissue examination device showing the oscillating mechanism when not in use . Fig. 3B is a cross-sectional view of the tissue examination device showing the oscillating mechanism in use .
Fig. 3C is a cross-sectional view of the tissue examination device showing the periodic force displacements travelling through the breast tissue in response to the oscillating mechanism.
Fig. 4 is a diagram of the polymer chains in breast tissue.
Fig. 5 is a stress/strain graph of breast tissue in comparison to a more viscous material and a more elastic material.
Fig. 6 is a mechanical model representing breast tissue .
Fig. 7 is a representation of a foreign structure in breast tissue.
Fig. 8A is a graph of amplitude/frequency responses of two dampened springs having different spring coefficients .
Fig. 8B is a graph of amplitude/frequency responses of two dampened springs having different dampening factors .
Figs. 9A and 9B are mechanical models representing of a foreign structure in breast tissue.
Fig. 10 is a representation of a rib underlying breast tissue which is close to the skin.
Fig. 11A is a mechanical model of a bony prominence in breast tissue.
Fig. 11B is a mechanical model of a mass near a bony prominence . Fig. 12 is a graph representing the expected profile of harmonic frequency and harmonic amplitude response of foreign structures to the vibrations produced by the oscillating mechanisms of the tissue examination devise of Fig. 1. Fig. 13 is a graph representing the expected thresholds and the expected area of frequency and harmonic amplitude responses of carcinomas.
Fig. 14 is a flow chart of the first stage of the analysis performed by the tissue examination device of Fig. 1.
Figs. 15A and 15B represent the relation between the response wave and the sampling instances.
Fig. 16 is a flow chart of the second stage of the analysis performed by the tissue examination device of Fig. 1.
Fig. 17 is a representation of an example of the location of sensors on the sensor array indicating a potentially suspicious structure.
Fig. 18 is a flow chart of the third stage of the analysis performed by the tissue examination device of Fig. 1.
Figs. 19A and 19B are graphical representations of the static pressure response and the oscillating response of a foreign structure in breast tissue to external vibration.
Description Referring to Fig. 1, a tissue examination device 10 includes an array 12 of pressure sensors 14 carried on a thin, flexible membrane 16. Array 12 is, for example,
a contact sensor such as that described in U.S. Patent No. 4,856,993, entitled "Pressure and Contact Sensor System for Measuring Dental Occlusion" (the '993 patent), incorporated herein by reference, the individual pressure sensors 14 of which are resistive elements. Pressure sensors 14 are arranged in an orthogonal grid of rows and columns in array 12. Pressure sensors 14 are relatively small and are closely spaced to provide high resolution capable of distinguishing between areas of underlying tissue separated by 1 mm or less. Array 12 is commercially available from Tekscan, Inc. (the assignee of the ' 993 patent) .
Referring also to Fig. 2, array 12 is mounted on a sensor head 55 made from a rigid polymer such as polycarbonate. Sensor head 55 is attached to a handle 60 which is grasped by a user to place array 12 against the tissue to be examined (such as the user's breast) . The face of sensor head 55 on which array 12 is mounted is convex, with a radius of curvature of approximately 1.5 inches to enhance the mechanical coupling between sensors 14 and the underlying tissue. The optimum range of the array curvature for mechanical coupling between sensors 14 and the underlying tissue is a radius of curvature between 1" - 2.5", although a radius as low as .5" or as high as 3" may also be used.
The resistance of each pressure sensor 14 changes in accordance with the amount of pressure applied to sensor 14. The resistance change is inversely proportional to the pressure imposed on sensor 14. Thus, the resistance of each sensor 14 decreases as applied pressure increases.
The individual resistances of pressure sensors 14 are read by preprocessing circuitry 20, the output 22 of which is applied to a digital signal processor (DSP) 24. Briefly, preprocessing circuitry 20 sequentially measures
the resistance of pressure sensors 14 in response to row and column address signals 23 provided by DSP 24 to provide an indication of pressure applied to the location in array 12 that corresponds to that sensor 14. During each resistance measurement, preprocessing circuitry 20 applies a reference potential (not shown) to the addressed sensor 14, measures the voltage drop induced across that sensor 14, and generates an output 22 corresponding to the voltage drop. Thus, each pressure sensor 14 produces a signal (in this example, resistance- induced voltage) in response to the applied pressure. The operation of preprocessing circuitry 20 is more fully described in the '993 patent.
The preprocessor output signals 22 are digitized (by A/D converters, not shown) and applied to DSP 24
(alternatively, an input stage of DSP 24 may perform the A/D conversion) . The set of sequentially produced output signals 22 for all pressure sensors 14 in array 12 is termed a "frame." DSP 24 stores frames obtained for sensor array 12 in areas 26a-26n and 28a-28n of memory 30. Each memory area 26a-26n and 28a-28n contains storage locations 27 which respectively correspond to the locations of pressure sensors 14 in a frame.
A green LED 45 is illuminated when device 10 is powered on. Green LED 45 remains illuminated throughout the tissue examination procedure as a system self check. A red LED 40 and an audio circuit 50 are driven by DSP 24 at various stages of the operation of device 10 to indicate to the user whether the user is using device 10 properly, how the user should operate device 10 at different stages of the examination, and whether suspicious structures have been detected. Visual display 54 may show the user two-dimensional and three- dimensional representations of the frames in real time during the examination.
A vibrator 38 vibrates the underlying tissue by applying a periodic force displacement to the tissue. Vibrator 38 includes an oscillator 36 which drives an oscillation mechanism 41 which in turn vibrates oscillation plates 45 and 46 against the tissue.
Oscillation plates 45 and 46 apply the periodic force displacement to the tissue to vibrate underlying tissue structures. Referring also to Fig. 3A, in response to signal 42 from DSP 24, oscillator 36 begins the rotation of motor and motor gear 43 at a frequency selected by DSP 24. Motor gear 43 is cooperatively coupled to actuator 44 which is pivotally connected to sliders 47 and 48. Each one of sliders 47 and 48 are pivotally connected to a corresponding one of oscillation plates 45 and 46. Oscillation plates 45 and 46 are also pivotally connected to a housing 49 which also houses sensor array 12. The rotation of motor gear 43 imparts an oscillating motion to actuator 44 which is transferred to sliders 47 and 48 and finally to the oscillation plates 45 and 46. Fig. 3B shows plates 45 and 46 in their lowest position in the oscillation cycle.
In use, device 10 operates and analyses frames of signals 22 in two modes. Briefly, in the first mode of operation, device 10 operates in accordance with the method disclosed in applicant's copending U.S. patent application Serial no. 08/757,466, filed on November 27, 1996 (the "'466 application"), incorporated by reference herein in its entirety. In a second mode of operation, described more fully below, the underlying tissue is excited by vibrating it with vibrator 38, and the vibration response of the tissue and the structures within the tissue are examined by DSP 24.
In the first mode of operation, head 55 is manually translated across the skin by the user applying
pressure with her hand placed on handle 60 (Fig. 2) . The translation technique is essentially a series of stationary palpations which allow the user to increase breast area coverage with less exam time. Generally, the pressure imposed on the sensors 14 increases when the sensors 14 are pressed against localized areas of stiffer tissue on, within, or below the softer breast tissue. Examples of such stiffer tissue include normal breast tissue structures -- such as the nipple, the inframammary ligament, and underlying ribs -- and foreign bodies such as cysts and solid masses (whether or not pathogenic) . Consequently, as array 12 is pressed and moved against the breast, the pressure imposed on sensors 14 and, thus the resistance of sensors 14, varies in accordance with the properties of the underlying tissue structures.
In the first mode of operation, DSP 24 addresses preprocessing circuitry 20 at a rate sufficient to read 20 frames or more of output signals 22 per second. DSP 24 stores each frame of signals 22 in areas 26a-26n of memory 30. Thus, each memory area 26a-26n contains a "map" of the pressures detected by pressure sensors 14 in a frame. This map can be viewed as a "pressure signature" of the tissue structures beneath array 12. Accordingly, memory areas 26a-26n contain a time sequence of pressure signatures of the underlying tissue as array 12 is palpated across the breast. When the user is applying the correct amount of pressure, DSP 24 performs various processing tests defined by an operating program 35 stored in memory 30 on the pressure signatures stored in memory areas 26a-26n. The tests enable DSP 24 to discriminate normal underlying tissue structures from potentially foreign structures. These tests are described in detail in the '466 application. In one of these tests, DSP 24 analyzes the amplitude of each signal
22 to determine whether it is above or below a pressure threshold that is dynamically determined for that frame. Those signals 22 in the frame that exceed the pressure threshold are termed "suspicious signals" 22 or "red signals" 22. Signals in the frame that do not exceed the pressure threshold are called "blue signals."
When breast examination device 10 operates in the first mode of operation, it identifies certain suspicious foreign structures within the breast . These suspicious foreign structures may include soft lumps, fluid filled cysts, hard lumps, carcinomas, or bony prominences which have curved tops and are located close to the skin (these bony prominences may appear as lumps to device 10) . Device 10 generally filters out small structures, nipples, inframammary ligaments, flat bony prominences, and possibly other non-suspicious elements based on the pressure signatures of these structures and the processing tests performed by DSP 24.
If DSP 24 determines a potentially foreign structure is present, DSP 24 notifies the user by illuminating a red LED 40. In addition, audio circuit 50, such as a buzzer, a tone generator, or both may be actuated by DSP 24 in conjunction with red LED 40, as discussed below. Handle 60 also includes a communication port 52 for coupling the maps of signals 22 to a visual display 54, thereby allowing the user to observe the pressure signatures directly.
Following detection of a foreign structure in the first mode of analysis, device 10 then begins to operate in the second mode. Generally, in the second mode, the user holds device 10 stationary over the underlying tissue where device 10 identified a suspicious foreign structure. In contrast with the first mode in which the underlying tissue is static during examination, the tissue is actively excited by vibrating it through a
range of frequencies using vibrator 38. DSP 24 then examines frames of signals 22 corresponding to those frequencies and analyzes the response of the foreign structure detected in the first mode of operation. Vibrating the underlying tissue allows DSP 24 to better discriminate between carcinomas and benign structures. For example, whether a foreign structure moves in response to the applied vibrations may enable DSP 24 to determine whether a structure is a rib, which does not vibrate. Device 10 examines the response of the structure through range of frequencies to determine its harmonic or resonant frequency and amplitude. It is expected that different structures will have different harmonic responses because of a variety of factors including degree of viscoelasticity of breast tissue, characteristics of various foreign structures, and the degree of connectedness of the structures to the surrounding tissue. If DSP 24 determines that the foreign structure may be a carcinoma, it notifies the user through red LED 40 and audio circuit 50.
In the second mode of operation as the tissue and the structures within it vibrate, they press with varying pressure on sensors 14 in sensor array 12, just as in the first mode of operation of device 10. Pressure sensors 14 on sensor array 12 detect the force of the vibration of the breast tissue and any foreign structure in response to the applied vibration. This data is acquired and then analyzed by DSP 24 in three stages.
In the first stage, device 10 vibrates the underlying tissue at a frequency that is initially set at 1 Hz and gradually increased to 200 Hz. The periodic force displacements applied to the breast by oscillation plates 45, 46 travel in breast tissue and vibrate the tissue and any foreign structure underneath the sensor array 12. DSP 24 addresses preprocessing circuit 20 to
obtain frames of signals 22 at a sampling rate (e.g. 20 frames per oscillation cycle) sufficient to determine the peak amplitude of the harmonic frequency of structures within the tissue. Prior to storing the frame, DSP 24 performs a preliminary test on each frame of signals 22 to determine that the user is applying the proper amount of pressure on device 10. If so, DSP 24 stores the acquired frames in memory areas 28a-28n in memory 30. Memory areas 28a-28n are identical to memory areas 26a- 26n in the manner the signals 22 from prepossessing circuit 20 are stored.
In contrast to the frames stored while performing the first mode of analysis, the frames stored in areas 28a-28n are indexed by frequency of the periodic force displacement frequency of oscillator 36 and the order in which they were obtained. They are also linked to the last frame 26a-26n (the "base frame") obtained in the first mode of analysis. The linkage may be for example, a pointer in each frame identifying the memory address of the next frame. The base frame contains the pressure profile for the area being currently examined prior to the area being vibrated. In order to determine the absolute amplitude response of this area, the value of the static pressure must be subtracted from the data obtained from the sensor array 12.
At the start of vibrating the tissue at the new frequency, a transitional period exists when breast tissue and the underlying structures which adjust from their previous condition to vibrating at a new frequency. During this time, DSP 24 does not request any frames of signals 22 from pre-processing circuit 20 because the frames during the transitional period can not be accurately used for determining the harmonic frequency. In the second stage, DSP 24 determines the peak amplitude of the response obtained. DSP 24 iterates
through all the stored frames 28a-28n for this area, and calculates the average pressure value for the area of suspicion on the frame. This area consists of sensors 14 which have pressure value above a threshold which is dynamically determined for that frame. DSP 24 then determines which frame has the highest average pressure value among all the frames acquired for all examined frequencies. This value will be the raw harmonic amplitude and the frequency index of the frame will be the harmonic frequency. However, the raw harmonic amplitude includes the pressure on sensors 14 resulting from the pressure the user applies to device 10 while holding it. Therefore, to obtain the absolute harmonic amplitude, a base value is subtracted from the raw harmonic amplitude to determine the absolute harmonic amplitude. This base value is calculated by averaging the pressure signature of the suspicious area in the base frame .
In the third stage, DSP 24 uses fuzzy logic analysis to determine from the absolute harmonic amplitude and the harmonic frequency the nature of the structure being examined.
In order to better understand the invention, the nature of breast tissue and its properties will be discussed. Breast tissue is a mixture of elastin, collagen, and fat. Referring to Fig. 4, elastin and collagen are long spiral polymer chains 85. In between layers of a polymer chain, cis bonds 80 are formed. As a polymer chain uncoils in response to an external force F, these cis bonds, together with other chemical bonds in the chain, stretch 82, 86 and apply an opposite force to urge the bonds to return to their original form. The spiral shape of these chains and the resistance of cis bonds and other chemical bonds to being stretched (or compressed) result in the chains returning to their
original shape. In other words, these polymer chains have a spring-like response with a specific spring constant to any force displacement applied along their length. The elastin and collagen polymer chains also form weak interchain links 81 with one another. As these chains uncoil or slide past one another in response to an external force F, the interchain links resist being stretched or broken. Also, as the chains slide past one another, some links are broken 83 and new links 84 are formed. New links 84 resist the chains returning to their original positions. Therefore, these links dampen the spring response of the polymer chains both when they uncoil and when they recoil. In essence, the polymer chains act as dampened springs giving breast tissue a viscoelastic characteristic.
Fat also affects viscoelasticity of breast tissue. Fat in breast tissue generally exists as globules of polymer chains. Generally, increased fat content of breast tissue results in increased viscosity or dampening factor of breast tissue and a lower spring constant.
Fig. 5 shows a strain/stress graph 90 for a viscoelastic material in comparison to a more elastic material and a more viscous material . The lower graph 92a shows the strain response, shown by the y-axis, of a very viscous material in response to external stress, shown by the x-axis. The strain response of a viscous material rises at a low slope and levels off at a low level when there is a viscosity breakdown. The upper graph 92b shows the strain response of a very elastic material. The strain response rises steeply, as the material continues to stretch, and levels off at a very high level. The strain response of a viscoelastic material 91 falls somewhere in between the more elastic and viscous materials.
Before the strain response levels off, i.e. before the elasticity of the material reaches its limit, strain is responsive to the applied stress. In this area 90, viscoelastic materials exhibit a different behavior than that of either elastic or viscous materials in response to applied periodic force displacement (i.e. vibration) . In response to externally applied periodic force displacement, the response of elastic materials will be directly proportional to the applied periodic force displacement. Viscous materials, in contrast, absorb most of the applied force and do not displace in response to a periodic force displacement. In viscoelastic materials, the viscous characteristics dampen the frequency response of the elastic characteristics. This results generally in a response with a lower amplitude to an externally applied periodic force displacement than an elastic material with the same spring constant. However, at a certain frequency, a viscoelastic material vibrates with an amplitude disproportionally high when compared to the input amplitude. This is the harmonic frequency of the material, and the amplitude of the response is its harmonic amplitude.
The harmonic response of a viscoelastic material may be demonstrated by the response of a dampened spring. At the harmonic frequency, the rate of energy absorption and release of the dampener and the spring are equal . Therefore, they reflect off one another, multiplying the amplitude of the response at that frequency.
Referring to Fig. 6, the mechanical response of breast tissue to external vibration can be demonstrated by a mechanical model of several dampened springs 100. Each of the dampened springs may have different spring and dampening coefficients, because the mixture of fat, elastin and collagen is not consistent throughout the breast. There may be areas of higher or lower viscosity
or elasticity throughout the breast. However, these values will be generally within a narrow range of values.
The system of several dampened springs will respond harmonically to vibrations at a certain frequency. For an externally applied periodic displacement force P (t) , the displacement of the system X (t) will reach a maximum when P (t) is applied at the harmonic frequency, where t is the period of the periodic force. Referring to Fig. 7, when a foreign structure 112 is present in breast tissue, it changes the harmonic response of the tissue affecting either the harmonic frequency or the amplitude of that response, or both. Based on the harmonic frequency response of breast tissue and possible structures embedded within it, it is possible to detect the type of these structures by applying an external periodic force (e.g., vibrations by oscillation plates 45, 46) and measuring the resultant amplitude at the harmonic frequency. The change in harmonic response of breast tissue that contains a foreign structure is a function of a number of factors relating to the foreign structure, including the density of the structure, the extent to which it is connected to breast tissue or anchored to ribs 114, and the depth in which it is embedded 116. We will discuss the expected amplitude and frequency of various foreign structures in detail below, based on analysis of these factors.
Referring to Figs. 8A and 8B, generally, the frequency and the amplitude of the harmonic response of a viscoelastic system change with variations in spring coefficient and dampening factor. In Fig. 8A, the effect of varying spring coefficient on the frequency of the harmonic system is shown. The harmonic frequency of a
dampened spring is directly proportional to the square root of the spring coefficient over the mass of the system, i.e.,
/« A m
Therefore, for the same mass, an increase in the spring coefficient (i.e. stiffness of the spring) results mainly in an increase in the harmonic frequency. It will also result in a decrease in the harmonic amplitude. Graph 65 shows the response of dampened spring having a lower stiffness than the system represented in graph 66. In contrast, variations in dampening factor mainly affect the amplitude of the response. As the dampening factor increases, the amplitude of the response decreases. Graph 67 shows the response of a system having a dampening factor lower than the one represented by graph 68.
A discrete and unanchored stiff structure suspended within breast tissue replaces a mass of tissue equivalent in volume (Fig. 7) . This replacement is equivalent to replacement of one of the dampened springs in the mechanical model in Fig. 6 with a mass. We do not expect the density of a lump to significantly affect the harmonic response of a lump, because the density of various types of lumps and breast tissue are approximately the same or at least within a narrow range of values. Therefore, for a given volume, all these structures have approximately the same mass as the breast tissue they displace. Therefore, the effect of varying densities of lumps on harmonic frequency and amplitude and response should not be significant. An- important factor in determining the harmonic response of a foreign structure in breast tissue is how
connected it is in the tissue, for example, to the polymer chains. Generally, the more connected a structure is, the higher the spring coefficient of the system is. Referring to Fig. 9A, in a mechanical model of such a system, the structure is suspended by two dampened springs 96 and is laterally connected to two other dampened springs 95. When vibrating vertically, the dampened springs 95 and 96 provide additional stiffness and dampening to the overall system. The more connected the tissue is, the higher is the added stiffness of dampening. If the model in Fig. 9A is collapsed into the model in Fig. 9B, the effect of lateral springs 95 is adding to the spring coefficient and dampening factor of dampening springs 97 and 98. Consequently, the amplitude of harmonic response is reduced while the harmonic frequency is increased. Therefore, the degree of connectedness affects both the harmonic frequency and the harmonic amplitude.
The internal spring coefficient and dampening factor of a structure also affects the harmonic response of a lump. The mechanical model 93 of a lump may be shown as a mass connected to four dampened springs 94 (Fig. 9A) . The spring coefficients of dampened springs 94 is higher for hard lumps than for soft lumps and hard shelled cysts whose spring coefficients are low. The dampening factors of dampened springs 94, however, are higher in soft lumps than in hard lumps. Hard shelled cysts, because of their hard outer shell, would also have a lower dampening factor than soft lumps (and soft shelled cysts) . Therefore, the internal structure of a lump may significantly affect the harmonic response of the lump. We also expect the internal structure to influence the pressure profile of the lump during oscillation.
We will now discuss the expected harmonic frequency and amplitude response of various types of structures which may be present in the breast . In the case of hard shelled fluid filled cysts, the outside shell of the cysts is hard and forms few connections to the polymer chains in breast tissue. Therefore, the cysts can move more freely than other structures. Also, because there are few connections between cysts and breast tissue, the resulting spring coefficient and the dampening factor are low. Therefore, we expect that the harmonic amplitude of cysts will be higher than all other structures because of this lack of connectedness to breast tissue. However, because hard shelled cysts are hard compared to soft lumps, they will likely have a somewhat higher harmonic frequency than soft lumps. The fluid inside cysts vibrates almost independently of the cyst itself, creating a second harmonic response in addition to the primary harmonic response of the cyst. The pressure profile of cysts may also be influenced by the vibration of the fluid inside cysts. DSP 24 may be able to detect the second order of vibration of cysts and use the data to distinguish fluid filled cysts from other structures.
Hard lumps have a higher harmonic frequency than soft lumps because hard lumps are not only connected to surrounding tissue but are also internally stiff. They, therefore, create a very stiff system with a high harmonic frequency. Soft lumps are well connected to the tissue, but their softness introduces a low spring coefficient and a higher dampening factor than other structures. Soft lumps absorb more of the vibration energy than harder lumps. Soft lumps therefore will have a low harmonic frequency and an amplitude response similar to that of hard lumps. Therefore, breast tissue having a hard lump appears as more stiff (i.e., with a
lower spring coefficient) to a periodic force displacement than breast tissue that contains a soft l mp. Therefore, hard lumps vibrate at a higher harmonic frequency than soft lumps . Referring to Fig. 10, another structure which may appear as a suspicious area when device 10 operates in the first mode of operation is a rib 118 having a curved profile and being located close to the skin. In order to discuss the response of ribs, it is important to discuss the effect of depth of structures within the breast on their examination by vibrating them. Tissue examination device 10 applies a periodic displacement force at the surface of the breast and obtains the response of breast tissue to the displacement force also at the surface of the breast. Therefore, the response that device 10 obtains from the breast is affected by the depth of the structure within the breast tissue. Generally, the applied force attenuates as it travels further into the breast tissue. The viscoelasticity of breast tissue results in the tissue absorbing and dispersing the energy of the force displacement applied at the surface as the displacement wave travels through the tissue. For the same reason, the harmonic response of a structure attenuates as the response vibrations travel to the surface. Therefore, the expected amplitude of the response changes depending on the depth of the structure being investigated. This attenuation has the advantage of attenuating the response of ribs underlying the breast, thereby eliminating effect of ribs on the examination of the breast unless a rib is close to the skin. Referring to Fig. 11A, the mechanical model of a rib close to the surface of the skin is represented by a short dampened spring 120 connected to an almost immovable mass (i.e., the rib) . A rib is anchored and has a high density. In response to a periodic force
displacement, a rib vibrates at a higher harmonic frequency than all other structures because it is hard, has a higher density compared to all other possible structures in the breast, and is very well anchored. However, it also has a low harmonic amplitude because it is heavily anchored and therefore can not move as freely in response to the vibrations.
Referring to Fig. 11B, a lump 124 that is so close to the rib that it essentially rests on top of the rib acts in a similar manner to the rib. Because of its proximity to the rib, it cannot vibrate freely, and therefore has a low harmonic amplitude response.
Carcinomas generally spread to the surrounding tissue and anchor themselves within the breast. They usually have a number of tentacle-like protrusions into the surrounding tissue. The surface of these protrusions and carcinoma form connections to many polymer chains in the tissue. Hence, carcinomas may be considered to be better connected to breast tissue than all other structures. They cannot move as freely as other lumps in the breast because of these connections; thus a system containing carcinomas will appear as very stiff compared to systems containing other structures. Carcinomas are also very dense, often denser than hard lumps, and as a result carcinomas introduce a high spring coefficient. Therefore, a carcinoma will likely have a lower harmonic amplitude than all other structures except ribs. Moreover, a carcinoma will also likely have a higher harmonic frequency than other lumps. Referring to Fig. 12, based on the above discussion, it may be concluded that the harmonic responses of various structures within breast tissue, which are identified as suspicious by breast examination device 10 when operating in the first mode of operation, are different from one another. Fig. 12 shows the
expected regions of response of soft lumps 101, fluid filled cysts 102, hard lumps 103, carcinomas 104, and ribs 105, in comparison to one another. Based on the above analysis, we expect the responses of these various types of structure to fall within the ranges shown in Fig. 12. Ribs 105 should have the lowest harmonic amplitude, while fluid filled cysts 102 should have the highest harmonic amplitude. The amplitude responses of soft lumps 101, hard lumps 103, and carcinomas 104 are expected to fall in between fluid filled cysts 102 and ribs 105. Soft lumps 101 should have a lower frequency than either hard lumps 103 or carcinomas 104. We expect hard lumps 103 to have a higher amplitude response than carcinomas 104. The exact values of the range of responses may be determined in clinical studies. The range of responses may then be reduced to a number of amplitude and frequency thresholds which define a number of windows corresponding to the various types of structures which may exist within breast tissue. Tissue examination device 10 differentiates between carcinomas and all other structures. Therefore, the frequency and amplitude responses are examined to determine whether they are within a window that corresponds to carcinomas . Referring to Fig. 13, carcinoma window 106 may be derived from analysis of clinical studies. Carcinoma window 106 may have different borders than the range of carcinoma responses derived from the clinical studies (104 in Fig. 12) because, for example, it may be desirable for device 10 to have more false positive detections rather than fewer false negative detections. The thresholds of carcinoma window 106 are not precisely defined, but are instead statistically based. That is, as the response moves away from the core area of clinical responses, the likelihood that a response is that of a
carcinoma decreases. Tissue examination device 10 uses so-called "fuzzy logic" to determine the degree of membership of a response and to decide whether the structure is a carcinoma. The method of operation of tissue examination device 10 will now be described in detail.
Preliminarily, note that the pressure signatures are a function of the amount of average pressure applied to sensors 14 when the user presses array 12 against the body. The pressure applied by the user should be within a selected range in order for the pressure signatures to accurately correspond to the various tissue structure types. The limits of the pressure range are a function of size and sensitivity of array 12. For array 12 discussed above, the acceptable pressure range is 0.2 psi to 2 psi.
Because the proper amount of user-applied pressure is important, in the first mode of operation, a preliminary pressure test is performed on the frame to determine whether the average amount of pressure applied to all sensors 14 is within the acceptable range. This preliminary pressure test also determines if a minimum number of sensors 14 are obtaining a reading across width of array 12 such that DSP 24 recognizes that entire array 12 is in contact with the skin. In the first mode of operation, if the frame fails the "preliminary pressure test (e.g., if the average applied pressure is below or above the acceptable range) , the frame is considered invalid and is not tested any further in the initial test procedure, and DSP 24 proceeds to the next frame stored in memory 30.
In use, in the first mode of operation the user translates the sensor head 55 over the skin. Sensor array 12 can be moved across a section of the breast vertically or horizontally while the user listens to the
low pitched humming tone 50 which indicates to the user that she is applying proper downward pressure to device 10. If red LED 40 is illuminated and audio circuit 50 generates an alarm tone during any portion of the scan, the user should scan that area of the breast again
(either in the same direction or in another direction, e.g., horizontally).
If red LED 40 and audio circuit 50 generates the alarm tone again, the user stops translating breast examination device 10. DSP 24 enters the second mode and performs a three stage analysis at this point. In the first stage, it vibrates the tissue through a range of frequencies, samples the tissue response, and stores the data. In the second stage, DSP 24 examines the stored data to determine the harmonic frequency and amplitude. In the third stage, DSP 24 examines the values obtained in the second stage to determine if the structure being examined is a carcinoma. These three stages will now be described in detail in reference to Figs. 14-19B. Referring to Fig. 14, DSP 24 signals 200 oscillator 36 to oscillate oscillation plates 45 and 46. Referring also to Fig. 3A, oscillation plates 45 and 46 remain in their raised position while device 10 operates in the first mode, upon notifying the user of detection of a suspicious mass, DSP 24 provides signals 42 to oscillator 36 to vibrate oscillation plates 45 and 46.
Referring also to Fig. 3C, two periodic force displacements 70 and 73 generated by vibrating plates 45, 46, travel towards each other at an angle. Horizontal vectors 71, 74 of two applied forces 70, 73 are directly opposed to each other. Any horizontal movement of a mass 76 directly underneath sensor array 12 due to one applied force is cancelled out by an opposite but equal horizontal displacement due to the other force. Therefore, the tissue and any structure within it vibrate
perpendicularly 77a, 77b to the surface of sensor array 12 in response to the vertical vectors 72, 75.
DSP 24 begins oscillating the underlying tissue starting at 1 Hz and increases the oscillation frequency gradually to a maximum of 200 Hz. The frequency is increased in increments of 5 Hz. DSP 24 at all times during the examination keeps track of the frequency at which oscillation plates 45 and 46 are oscillating. Oscillation plates 45 and 46 in turn vibrate the area of the breast directly underneath sensor array 12. The input amplitude of the vibration (i.e., the periodic force displacement) is constant throughout the procedure because the amplitude of the response is proportional to the input amplitude. If the input amplitude is significantly varied, comparison of amplitude responses at different frequencies may not yield accurate results. Moreover, the amplitude must be high enough to mechanically oscillate the tissue and any foreign structure within the tissue. Low amplitude displacement waves may merely reflect from the structure or attenuate to an extent that the response wave could not be detected by sensor array 12.
In step 205, DSP 24 supplies oscillator 36 with the frequency at which the tissue should be vibrated. DSP 24 stores this frequency as the current vibration frequency. DSP 24 allows one to two seconds for the tissue to adjust to the new frequency (step 210) . Because prior to application of the new frequency the tissue was either static or vibrating at a different frequency, it will take one to two seconds before the tissue and any structure within it are fully excited at the new frequency and are responding at their peak amplitude to the new frequency.
In step 215, DSP 24 calculates the sampling rate for sampling the frequency response (i.e., the response
wave) of the area being examined. Referring also to Fig. 15A, in order to detect the amplitude of the response, the response from the tissue must be sampled at a rate fast enough for an amplitude at or close to the peak amplitude to be sampled. In the preferred embodiment, this sampling rate is 20 samples per each cycle of the wave. For example, for a vibration frequency of 200 Hz, the required sampling rate would be 4000 Hz.
However, sensor array 12 can only provide data at a frequency of 20 Hz or less. Referring to Fig. 15A, the response wave will be similar or identical in each cycle because the applied periodic displacement force has a constant frequency and amplitude at each frequency step. Therefore, the frequency of the response (which is the same as the oscillation frequency) and the amplitude of the response will be constant. Therefore, for waves requiring a sampling rate above the maximum data output rate of sensor array 12, the 20 required samples may be taken from different cycles of the response wave. During each cycle when sensor array 12 can provide data (that is, a cycle at least l/20th of a second after the previous cycle at which a sample was taken) , the sampling point is moved by l/20th of the oscillation period from the position of the previous sampling instance relative to the wave cycle. The time period between each sampling point (Tacq) is therefore determined according to the following formula:
Tacq = Tacq_max + 1/20 (Tosc),
where Tosc is the period of the response wave and Tacq_max the upper limit of the rate at which data may be obtained from sensor array 12 (i.e. 1/20 sec.) . Hence, referring to Figure 15B, for response waves with
frequencies above the l/20th of 1/Tacq_max, samples 150 are taken from different cycles of the response wave 151. From the twenty samples taken for each cycle, ten of them will be from the trough of the response wave. However, pressure sensors 14 can sense only positive pressure. Therefore, the frames corresponding to the wave trough do not contain any useful information. They will likely sense zero pressure. These frames are automatically filtered out when DSP 24 examines all the frames stored in memory 30 to determine the harmonic amplitude, because their values will be less than the frames from positive amplitude of the response wave.
In step 220, DSP 24 signals preprocessing circuit 20 to obtain a frame at the appropriate sampling period. A preliminary pressure test similar to the preliminary test in the first mode of operation is first performed on the frame. The values from pressure sensors 14 are averaged and compared to a minimum and a maximum threshold pressure value. This preliminary pressure test also determines if a minimum number of sensors 14 are obtaining a reading across width of array 12 such that DSP 24 recognizes that entire array 12 is in contact with the skin (step 225) . If the frame fails the pressure threshold test, DSP 24 stops the humming tone, to indicate that the user should adjust the pressure on device 10 (step 230) . A new frame at the same sampling period relative to the response wave as the sampling point of the previous frame is then obtained by not moving the sampling point by 1/20 Tosc (step 220) . If the pressure exerted by the user is within the appropriate range, DSP 24 reactivates the humming tone in step 240 (if it was previously deactivated in step 230) . The frame is then stored in memory 30, indexed with the frequency of the applied periodic force displacement and the order in which it was obtained (step 245) . The frame
is also linked to the base frame by, for example, having a word in the base frame pointing to the memory address of the linked frame.
DSP 24 continues to obtain frames at the appropriate sampling points until the required number of frames for this frequency is obtained (step 250) . In the preferred embodiment, 20 frames for each vibrating frequency are obtained. When the 20 frames for the current frequency are obtained, DSP 24 then begins vibrating the tissue at the next frequency (step 256) .
Referring to Fig. 16, in the second stage, DSP 24 examines stored frames 28a-28n to determine the harmonic frequency and amplitude of the underlying tissue. DSP 24 retrieves each of the frames and its frequency (step 305) . DSP 24 next calculates the average of the values from the "red" sensors for the area of suspicion (step 310) . A threshold test is used to determine which sensors are "red" . DSP 24 derives the threshold dynamically for each frame by determining the average pressure detected by all sensors 14 in array 12, and multiplying the average by an empirical value (the "red/blue factor") . (The average pressure for a frame is obtained by adding the pressure values detected by sensors 14 and dividing the result by the number of sensors 14 in array 12. )
DSP 24 compares the pressure value of each location in the frame (i.e., the amplitude of signal 22 produced by each sensor 14 in array 12) with the dynamic threshold. If the pressure value produced by a sensor 14 is above the dynamic threshold, the location of the sensor 14 is marked "red". If the pressure value is below the dynamic threshold, the location of the sensor 14 is marked "blue". The area of suspicion is the area of red sensors which was identified in the threshold test .
Referring also to Fig. 17, for example, pressure sensors 170 which are located in d4-5, e3-6, f2-7, g3-6, and h4 are the red sensors in this frame. When DSP 24 examines frequency indexed frames 28a-28n linked to the base frame in memory 30, only the values corresponding to red sensors 170 will be examined to determine the harmonic frequency and amplitude, because these sensors contain the pressure values corresponding to the response of the structure under examination. The location of red sensors relative to the frame of course may change because of lateral movements of the lump or human error in keeping device 10 in one location over the breast. Therefore, red sensors must be identified in each frame. Referring to Fig. 16, in step 310, DSP 24 calculates an average of the values from the red sensors. If this value is greater than the value stored as the greatest average up to this point (step 315) , the average, the frame, and the indexed frequency are stored as those of the frame with the greatest average (step 320) . (Of course, this value is reset prior to examining the first frame.) Steps 305-320 are repeated until all frames 28a-28n corresponding to the range of examined frequencies for this suspicious area are examined (step 325) . At the end of this stage, DSP 24 has determined the frame having the greatest average. This average will be the closest sampled average pressure value proportional to the harmonic amplitude. The frequency of this frame will be the harmonic frequency or the closest examined frequency to the harmonic frequency of the structure being examined.
Referring to Fig. 18, in the third stage, DSP 24 determines whether the structure being examined is a carcinoma or a benign structure. In step 400, DSP 24 subtracts from the average of the red sensors in the harmonic frequency frame, the average of the values from
the red sensors in the base frame. This subtraction yields the absolute value of the harmonic amplitude. Referring to Fig. 19A, during the first mode of operation of device 10, a static pressure profile 183 was obtained for the area under examination. Referring to Fig. 19B, when the area is vibrated at the harmonic frequency, the pressure imposed on the sensors is increased above that of the static pressure. The data from sensor array 12 is raw amplitude 180 which is the sum of absolute amplitude 181 and static pressure 182. In order to obtain absolute amplitude 181, static pressure 182 obtained from the base frame must be subtracted from raw amplitude 180. This is done in step 400 by subtracting the average value of red sensors in the base from the average value of red sensors in the harmonic frequency frame.
Referring again to Fig. 18, following adjusting for the static pressure, the harmonic frequency and the harmonic amplitude are then compared to the threshold values of carcinoma window (106 in Fig. 13) determined from clinical studies (steps 405 and 410) . The result of these comparisons are stored so that together with the result of the pressure profile test (step 415) , described below, can be analyzed by fuzzy logic techniques (step 420) , described below. In step 415, a pressure profile test, similar to the one used in the first mode of operation and described in the '466 application, is performed on the frame for the harmonic frequency. The pressure profile test in step 415 is a 3-D (three-dimensional) test in which DSP 24 analyzes the amplitudes of values from the red sensors to determine whether the pressure signature of the tissue structure is approximately lump- like in three dimensions. For example, the pressure profile test enables DSP 24 to determine whether the central region of the pressure signature is relatively large (like that of a soft cyst) ,
or is small (like that of a solid mass) . Due to their somewhat spherical shapes, foreign structures (such as cysts, benign solid masses, or carcinomas) induce pressure signals 22 with amplitudes that increase progressively as pressure is sampled from the periphery of the structures to their center. Accordingly, in the pressure profile test DSP 24 determines the edge profile, the relative stiffness, and the relative curvature of the underlying structure based on how the amplitudes of suspicious signals 22 change from the periphery of the structure toward the center of the structure. DSP 24 evaluates the edge profile to determine whether it is extremely sharp (which indicates that the structure may be a rib, rather than a lump) or is more moderate. DSP 24 also determines whether the structure's stiffness and curvature are more indicative of a lump than of a normal tissue structure.
This test is performed during the first mode of analysis. However, we expect that because of the increased amplitudes at harmonic frequencies, some of the characteristics of the pressure profiles for various structures will be further exaggerated and therefore more easily identifiable. Because during harmonic oscillation lumps exert more pressure on sensor array 12 than in the first mode of analysis, we expect the curvature of the pressure profile to be steeper than the curvature of the pressure profiles in the first mode analysis. The result of clinical studies will further enhance the pressure profile test for oscillating lumps. In step 420, DSP 24 applies so-called "fuzzy logic" techniques (also known as "soft thresholding") to weigh the results of steps 405, 410, 415. This technique is a neural network concept that develops parameters of imprecise measurements. DSP 24 weighs the differences between the harmonic frequency and amplitude, and
thresholds determined in steps 405 and 410 and the edge profile, stiffness, and curvature determined in step 415. It then develops a "degree of membership" outcome ranging from 0 to 1 of the characteristics of the suspicious region as a carcinoma (i.e., the degree to which, based on the weighed results of steps 405, 410, 415 the suspicious region resembles a carcinoma) . DSP 24 weighs the results of steps 405, 410, and 415 equally, but alternatively could assign different weights to these results. In determining the degree of membership of a structure, DSP 24 may use results from only one, or a combination, of the results from the tests in steps 405, 410, and 415. The results of the clinical studies will provide better indication of which tests in combination with the fuzzy logic technique yield the best results. If in step 420 DSP 24 determines that the structure may be a carcinoma, then DSP 24 causes audio circuit 50 to generate a high pitch alarm tone to notify the user (step 425) . The user then should consult a physician at the end of examination. The user should now continue with the examination in accordance with the first mode of analysis. If DSP 24 determines that the structure is not a carcinoma, it provides a mid-pitch hum to indicate to the user to continue with the examination according to the first mode of operation of device 10. Other embodiments are within the scope of the claims .
Factors such as age, fitness level, percent body fat, life style, child bearing and breast feeding history, smoking habit, and so on may affect the ratio of the mixture of fat, collagen and elastin in breast tissue. The ratio of the mixture in turn determines the viscoelastic characteristics of the breast tissue. Therefore, these factors may in effect change the spring coefficient and damping factor for the system, affecting
harmonic response of breast tissue and structures within it. Clinical studies for determining the threshold, therefore, may also provide data to create a variety of thresholds for different population groups based on these factors. The preferred embodiment may then be calibrated to take into account these factors based on each user's characteristics .
In one embodiment, device 10 would be configured for different populations of women. For example, one embodiment may be configured for women between 35-45 years of age who are non-smokers and are moderately active. Another embodiment may be configured for women between 25-35 years of age and who smoke and are moderately active. Various embodiments may be provided for the complete range of population. The devices may then be prescribed by physicians, or bought off the shelf or over the counter. In another embodiment, a clinician may enter a woman's characteristics into an expert system which then calibrates the device (using an EPROM for example) for that woman.
In the first embodiment described herein, the average of the red sensors are examined in the second and the third stages of the second mode. In alternative embodiments, the median value or the sensor having the highest pressure value may be examined. In yet another alternative embodiment, the response wave may be reconstructed fully in 3D using the sample data and then most relevant values for analysis chosen on a frame by frame basis. In an alternative embodiment, a Fourier transform may be used to determine the peak amplitude response. The Fourier transform may be performed on the pressure value of one of pressure sensors 14 having the highest output for a given area, the average the pressure values of the red sensors, or all of the red sensors.
In alternative embodiments, other oscillating mechanisms may be used, including piezo-electric or magnetic devices.
In the first embodiment, all samples are stored and then examined for determining the harmonic amplitude and frequency. In an alternative embodiment, the frames can be examined as they are received from preprocessing circuit 20.
In an alternative embodiment, a single oscillation plate may be used to vibrate the tissue. Alternatively, more than two plates may be used. These plates may be configured such that the response would be perpendicular or parallel to the sensor array.
In an alternative embodiment, oscillation plates 45, 46 vibrate the breast tissue out of synch.
Therefore, a lump within the breast, would appear to array 12 to be moving laterally or in a pendulum swing as well as possibly vertically. DSP 24 then would analyze the lateral movement of the lump to determine the nature of the lump .
In an alternative embodiment, rather than determining the harmonic frequency and amplitude, only the fact that a structure is able to move or oscillate within the breast is examined. The tissue is vibrated through a range of frequencies and DSP 24 determines whether the structure vibrates or "moves in response to the applied force. Based on whether the structure moves. DSP 24 then can determine the structure type. For example, a bony prominence, a very soft mass, or an area of thickening in breast tissue, may not oscillate or move at all in response to an applied force or it may not oscillate or move in response to a particular range of frequencies. Therefore, they can be distinguished from other structures based on the fact they do not move in response to an applied force.
In an alternative embodiment, the phase difference between the response wave and the vibrating wave may also be analyzed to determine the characteristics of the foreign structure. In an alternative embodiment, multiple suspicious areas under the array may be examined. In one embodiment, the frames may be divided into halves or quarters corresponding to a virtual division of the array surface into halves or quarters. The red areas in each section may then be examined separately by limiting the analysis in the second and third stage only to those sensors within the particular section being examined.
Alternatively, a mapping technique may be used to examine each red area in the base frame. The mapping technique comprises matching red areas in each of the obtained frames by size, characteristic, and location to the red area in the base frame under examination. Generally, the structure under investigation may move with respect to the array because of its oscillation. The user may also move the array during the second mode of analysis. Therefore, the red sensors in the base frame will not necessarily be the red sensors of the frames obtained in the second mode operation. Therefore, the red sensors of a frame must be mapped onto the red sensors of the base frame, i.e. DSP 24 must determine whether red sensors in a frame correspond to the same structure as the red sensors in the base frame. DSP 24 implements this on the basis of the assumption that generally the red sensors in a frame will have the same approximate area and will be in the same vicinity as the red sensors in the base frame.
DSP 24 first finds the center of red areas in both frames by drawing a hypothetical rectangle around the red areas. The borders of the rectangles will coincide with the outer most sensors of each red area. DSP 24 then
calculates the intersection of the two diagonals of the rectangle which is taken to be center of the red area. Alternatively, the center of red areas may be obtained using other standard techniques, such as center of mass or Centroid Weighted Technique.
If the distance between the centers of two red areas in two frames is within a radius of tolerance (e.g. 1 cm) and the area of red sensors in the two frames are similar in size, shape, and/or pressure profile, the two areas are considered to be the same. Each frame for each frequency is mapped in this manner, in order to determine which signals correspond to the same structure in various frames. For each red area in the base frame, the second and third stages are repeated while using this mapping technique to ensure examination of the same area.
Still other embodiments are within the scope of the claims.