Method and apparatus for sensing and processing biopotentials

[0002] This application is a continuation-in-part application of provisional applications Serial No. 60/076,353 filed Feb. 27, 1998, 60/080,775 filed Apr. 6, 1998, and 60/094,643 filed Jul. 30, 1998.

[0003] A breast cancer investigation is usually triggered by the discovery of a breast mass, and although in many instances the mass is the result of a benign lesion, this is not discovered until a patient has often undergone a battery of diagnostic tests which are sometimes invasive. A need exists for a noninvasive, adjunctive method and device which will aid a physician in making a clinical decision as to whether or not additional diagnostic tests for breast cancer are warranted.

[0004] Breast cancer is thought to originate from epithelial cells in the terminal ductal lobular units of mammary tissue. The development of breast cancer results in regions of epithelial electrical depolarization within the breast parenchyma, which led to the theory that the measurement of skin surface electropotentials would provide data to indicate the presence of underlying abnormal proliferation indicative of cancer. Many methods and devices have been developed in an attempt to implement this theory.

[0005] For example, U.S. Pat. No. 4,328,809 to B. H. Hirschowitz et al. deals with a device and method for detecting the potential level of the electromagnetic field present between a reference point and a test point on a living organism. Here, a reference electrode provides a first signal indicative of the potential level of the electromagnetic field at the reference point, while a test electrode provides a second signal indicative of the potential level of the electromagnetic field at the test point. These signals are provided to an analog-to-digital converter which generates a digital signal as a function of the potential difference between the two, and a processor provides an output signal indicative of a parameter or parameters of the living organism as a function of this digital signal. For breast cancer detection, Hirschowitz et al. shows that a test electrode can be placed in each quadrant of a human female breast and that multiple measurements can be taken during a test period with each test electrode and a reference electrode. These multiple measurements are digitized, normalized, and summed to provide an average or mean output signal indicative of a parameter of the living organism under test.

[0006] Similar biopotential measuring devices are shown by U.S. Pat. No. 4,407,300 to Davis, and U.S. Pat. No. 4,557,271 and 4,557,273 to Stoller et al. Davis in particular discloses the diagnosis of cancer by measuring the electromotive forces generated between two electrodes applied to a subject.

[0007] Often, the measurement of biopotentials has been accomplished using an electrode array, with some type of multiplexing system to switch between electrodes in the array. The aforementioned Hirschowitz et al. patent contemplates the use of a plurality of test electrodes, while U.S. Pat. No. 4,416,288 to Freeman and U.S. Pat. No. 4,486,835 to Bai disclose the use of measuring electrode arrays.

[0008] Unfortunately, these previous methods for employing biopotentials measured at the surface of a living organism as an adjunctive aid to diagnosis, while basically valid, are predicated upon an overly simplistic hypothesis which is not effective for many disease states. The prior methods and devices which implement them operate on the basis that a disease state is indicated by a negative polarity which occurs relative to a reference voltage obtained from another site on the body of a patient, while normal or non-malignant states, in the case of cancer, are indicated by a positive polarity. Based upon this hypothesis, it follows that the detection of disease states can be accomplished by using one measuring electrode situated externally on or near the disease site to provide a measurement of the polarity of the signal received from the site relative to that from the reference site. Where multiple measuring electrodes have been used, their outputs have merely been summed and averaged to obtain one average signal from which a polarity determination is made. This approach can be subject to major deficiencies which lead to inaccuracy, particularly where only surface measurements are taken.

[0009] U.S. Pat. Nos. 4,955,383 and 5,099,844 to M. L. Faupel disclose a method and apparatus using electropotential differentials between averaged values provided by a plurality of different sensors. This method operates on the basis that maximum differentials between areas of diseased tissue and apparently normal tissue in other areas of a breast for a breast cancer investigation provide informative parameters not subject to the inaccuracy of previous methods.

[0010] Still, the accurate measurement of DC biopotentials for sensing disease, such as breast cancer, is very difficult to accomplish, for the DC potentials to be sensed are of a very low amplitude. Due to factors such as the low DC potentials involved and the innate complexity of biological systems, the collected data signals tend to include a substantial amount of noise which makes accurate analysis difficult. Also, biological systems are notorious for their complexity, nonlinearity and nonpredictability, and wide variations from the norm are not uncommon. Thus it is necessary to develop a method and apparatus for obtaining the necessary data from the measurement of biopotentials and then to extract and analyze pertinent information which is relevant to a condition under study.

[0011] In an attempt to accomplish this, the method and apparatus of the previous Faupel patents was combined with one or more preprogrammed neural networks as illustrated by U.S. Pat. No. 5,697,369 to D. M. Long Jr et al. and U.S. Pat. No. 5,715,821 to M. L. Faupel. With a neural network, data can be processed by several layers of interacting decision points or neurons. The network must be taught to recognize patterns from input data to produce a predictive output, and this can prove to be a complex and often time consuming process involving many variables. Therefore, a need has arisen for a simple method and apparatus for use in the analysis of female breast electropotentials to minimize the effects of noise on the measurement data and to compensate for the biologic variability which affects the measurement data. Past methods and devices have concentrated on developing accurate data from sensed biopotentials which will provide an indication of the probability that a malignancy exists. These methods have ignored the effects of various biologic variables such as menstrual status or timing of the menstrual cycle. (Electrical Potential Measurements in Human Breast Cancer and Benign Lesions; Tumor Biology, 1994, pages 147-152). However, in younger women (ages 18-56) with a palpable breast mass, the prevalence of cancer becomes lower as age decreases. The sensitivity of mammography is limited in this less than 56 age group due to the density of breast tissue. Because of this, a considerable degree of diagnostic uncertainty remains after physical examination and mammography, and as a result, the open biopsy procedure performed on women in this 56 year or under population yields a low percentage of malignancy diagnoses. Physicians dealing with breast cancer detection have a need for a noninvasive technique which will provide data to effectively aid them in reaching a clinical decision that a suspicious lesion is benign and does not require a breast biopsy while still providing an indication that malignancy is probable when such is the case so that a decision can be made to conduct a biopsy.

[0012] It is a primary object of the present invention to provide a novel and improved method and apparatus for sensing and processing electropotentials from a female breast which minimizes the effects of noise on the measurement data.

[0013] Another object of the present invention is to provide a novel and improved method and apparatus for sensing and processing electropotentials from a human subject which involves weighting the measured electropotentials to compensate for biologic variables.

[0014] A further object of the present invention is to provide a novel and improved method and apparatus for sensing and processing electropotentials from a human subject which involves weighting the measured electropotentials in accordance with the age of the subject.

[0015] Yet another object of the present invention is to provide a novel and improved method and apparatus for sensing and processing electropotentials from a female breast which involves weighting and/or discarding the electropotential measurement in response to the dates of menses and/or hormonal levels.

[0016] A further object of the present invention is to provide a novel and improved method and apparatus for sensing and processing electropotentials from a female breast where a primary electropotential measurement is obtained from an area of a symptomatic breast which is directly over a lesion and this primary electropotential measurement is processed with an electropotential measurement obtained from one or more areas of the symptomatic breast which are spaced from the lesion to provide noise reduction.

[0017] Yet a further object of the present invention is to provide a novel and improved method and apparatus for sensing and processing electropotentials from a female breast where a primary electropotential measurement is obtained from an area of a symptomatic breast which is directly over a lesion and this primary electropotential measurement is processed with electropotential measurements obtained from one or more areas of the symptomatic breast which are spaced from the lesion and from areas of the asymptomatic breast which correspond with the areas of the symptomatic breast where measurements are taken to reduce noise.

[0018] Another object of the present invention is to provide a novel and improved method and apparatus for sensing and processing electropotentials from a female breast which involves weighting the measured electropotentials to compensate for biologic variables which include one or more of patient age, hormonal levels, time in the patient menstrual cycle, and time of day during which the measurement is taken.

[0019] These and other objects of the present invention are accomplished by the apparatus and method of the present invention. In the most basic mode of operation, a biopotential sensor is placed directly over a lesion in a symptomatic breast, a reference sensor is located either away from the symptomatic breast, or directly over the nipple of the symptomatic breast, and a plurality of electropotential test measurements are taken during a test period and averaged to obtain a primary test potential. This primary test potential can be weighted to compensate for one or more biologic variables such as patient age, hormonal levels, time in the patient menstrual cycle, and the time of day during which the measurement is taken. These biologic variables are provided to a processor for the apparatus by one or more data input devices which can include a keyboard and possibly sensors for some biologic conditions. The processor stores weighting factors for each biologic variable and various levels of the biologic variable, and applies one or more of these weighting factors to the primary test potential measurement which is then displayed or compared to a reference value and the result displayed. For some period of a female menstrual cycle, the processor will either prevent the display and will instead provide an indication that no accurate measurement can be taken, or will weight the measurement taken to compensate for the period during the menstrual cycle when the measurement is obtained.

[0020] To provide noise enhancement of the primary test potential average measurement, the symptomatic breast is divided into four equal quadrants with one quadrant containing the lesion. At least one of the remaining three non-lesion quadrants is provided with a biopotential sensor, and during the test period a number of electropotential measurements equal to those taken by the sensor over the lesion are taken and averaged. This average value from the non-lesion quadrant is subtracted from the primary test potential average to provide a differential value which then can be compared to a predetermined reference value and the result displayed. Often both the average from the non-lesion quadrant and the primary test potential average are weighted before the subtraction step by the processor to obtain the differential. Also, the processor will often provide additional weighting for biologic variables.

[0021] Further enhanced noise reduction of the primary test potential average measurement is obtained by placing at least one biopotential sensor in each non-lesion quadrant of the breast and taking the same number of electropotential test measurements during a test period from each non-lesion quadrant that are taken with the sensor over the lesion. At the end of the test period, the processor averages the measurements taken by each sensor to obtain an average, and the average for the sensor over the lesion constitutes the primary test potential. The processor may then obtain the median value of the averages for the non-lesion quadrants or alternatively the mean value, and it will then subtract this median or mean value from the primary test potential to provide the differential value. Again the mean or median value and the primary test potential may be weighted before the subtraction and additional weighting may be provided for biologic variables. The resultant differential value is then compared to a predetermined reference value.

[0022] Maximum noise reduction of the primary test potential average is obtained as described above using averages obtained from the biopotential sensors over the non-lesion quadrants of the symptomatic breast. Also mirror image sensors are arranged on the asymptomatic breast to duplicate the positions of the sensors on the symptomatic breast. During a test period, the processor causes all sensors to take the same number of plural electropotential measurements, and at the end of the test period the average value of the measurements for each individual channel is computed.

[0023] For the symptomatic breast, the processor computes either the median or the mean value for the non-lesion quadrants. For the asymptomatic breast, the processor computes a maximum voltage differential value (MVD) which is the difference value obtained by subtracting the lowest average value obtained for the asymptomatic breast from the highest average value. Then the processor subtracts the median or mean value for the non-lesion quadrants of the symptomatic breast from the primary test potential average and adds the MVD from the asymptomatic breast to obtain a differential value which is compared to a predetermined reference.

[0024] Alternatively, instead of the MVD from the asymptomatic breast, the processor may obtain a median value or a mean value from the averages for the asymptomatic breast and add these to the result of the subtraction of values for the symptomatic breast. Again, all values from the symptomatic and asymptomatic breast may be weighted and additional weighting may be provided for biologic variables.

[0025] For screening purposes, at least one sensor may be placed in each breast quadrant and multiple measurements taken with each sensor and averaged to obtain a quadrant average value for each quadrant. The average values are compared, and if a quadrant average value varies from the others by more than a predetermined amount, this quadrant is designated as a potential lesion quadrant.

[0026] FIG. 1 is a block diagram of the apparatus of the present invention;

[0027] FIG. 2 is a flow diagram illustrating the basic operation of the apparatus of FIG. 1;

[0028] FIG. 3 is a diagram showing a first sensor placement method of the present invention;

[0029] FIG. 4 is a flow diagram illustrating the operation of the apparatus of FIG. 1 with the sensor placement of FIG. 3;

[0030] FIG. 5 is a diagram showing a second sensor placement method of the present invention;

[0031] FIG. 6 is a flow diagram illustrating the operation of the apparatus of FIG. 1 with the sensor placement of FIG. 5;

[0032] FIG. 7 is a flow diagram modification of the flow diagram of FIG. 6;

[0033] FIG. 8 is a diagram showing a third sensor placement method of the present invention;

[0034] FIG. 9 is a flow diagram illustrating the operation of the apparatus of FIG. 1 with the sensor placement of FIG. 8;

[0035] FIG. 10 is a flow diagram illustrating a second mode of operation of the apparatus of FIG. 1 with the sensor placement of FIG. 8;

[0036] FIG. 11 is a flow diagram illustrating a third mode of operation of the apparatus of FIG. 1 with the sensor placement of FIG. 8;

[0037] FIG. 12 is a chart showing average electropotential breast measurements taken from a plurality of women during the menstrual cycle;

[0038] FIG. 13 is a flow diagram illustrating the operation of the apparatus of FIG. 1 to weight measurement results for menses;

[0039] FIG. 14 is a sectional diagram of two biopotential sensors and a unique sensor mount;

[0040] FIG. 15 is a diagrammatic view of sensor placement for gradient approximation;

[0041] FIG. 16 is a diagram showing a sensor placement for breast measurements using a nipple sensor;

[0042] FIG. 17 is a diagram showing sensor placement for breast screening; and

[0043] FIG. 18 is a flow diagram illustrating a screening mode of operation for the apparatus of FIG. 1 with the sensor placement of FIG. 17.

[0044] FIG. 1 discloses a basic block diagram of the apparatus of the present invention indicated generally at 10 for evaluating skin surface electropotentials associated with suspicious breast lesions prior to a complete diagnostic workup. To accomplish this, a plurality of DC biopotential test sensors for sensing DC electropotentials, such as sensing electrodes 12 and at least two reference sensors such as electrodes 14 and 14aare used to provide analog outputs indicative of DC electropotentials measured on the breasts of a human female subject. When sensing electrodes form the sensors 12 and 14, 14a, they may include a layer of silver 15 forming an electrode, a layer of silver chloride 17, and an ion conducting cream or gel 19 as shown in FIG. 14 (see also U.S. Pat. Nos. 5,660,177 and 5,823,957).

[0045] The method of this invention contemplates the uses of a variety of different sensor arrays depending upon the intended application for which the device 10 is used. The aim is to measure the areas of electrical activity which occur as a function of the underlying biological activity of the organ system. The number of sensors, such as sensing electrodes 12 used in the measurement will also be a function of the specific application.

[0046] In FIG. 1 for purposes of illustration, two sensor arrays 16 and 18 are shown with each array consisting of four sensors 12 providing four separate output channels for each array. In actual practice, for some applications, each array will contain more sensors as will be described.

[0047] The sensors 12 of the sensor arrays 16 and 18 should be mounted in a manner which permits the sensors to be accurately positioned against the curved surface of the skin of a subject in the area of a test site while still maintaining spacing and the position of the sensors in a predetermined pattern to be described. The sensors 12 and reference sensors 14 and 14amust all be of a type suitable for detecting DC electropotentials indicative of the potential level present in the breasts of a female subject. These sensors should be of a type which do not cause a substantial battery effect between the breasts under test and the sensors and must have a very low DC offset potential. (See U.S. Pat. Nos. 5,217,014, 5,560,357, 5,660,177 and 5,823,957).

[0048] The device 10 is a multi-channel device having leads 20 extending separately from the sensors 12 in each array and leads 22 and 22aextending from the reference sensors 14 and 14arespectively. Each sensor 12 in combination with a reference sensor forms a separate data channel which transmits one or more analog signals indicative of the DC electropotential at a specific site on the breast. The leads 20 from the array 16 are connected to a solid state multiplexor 24, while the leads from the sensor array 18 are connected to a second solid state multiplexor 26. Each sensor array connected to the device 10 provides a plurality of outputs to a multiplexor connected to the array, and this multiplexor switches between the leads 20 during a test period to connect the analog signals on each lead sequentially to a multiplexor output such as the output lines 28 and 30 to create a time division multiplexed output. By providing a high speed solid state multiplexor for each array, it is possible to repeatedly sample electropotentials from each of the plurality of sensors in an array during a test period of minimal duration.

[0049] In the device 10 of the present invention, the analog signals on the outputs from each multiplexor are passed through separate relatively high frequency low pass filter amplifiers, such as the filter amplifiers 32 and 34. These filter amplifiers have a relatively high cutoff frequency of 40 Hertz or more, and thus require a stabilization period with analog signals of the amplitude provided on the output lines 28 and 30 to the filters.

[0050] The analog output signals from the filter amplifier 32 connected to the multiplexor for the sensor array 16 are directed to an analog to digital converter 36, while the analog output signals from the filter amplifier 34 for the sensor array 18 are connected to an analog to digital converter 38. The analog to digital converters operate to convert the input analog signals to output digital signals which are a function of the analog inputs.

[0051] The analog to digital converters 36 and 38 operate in response to timing signals provided on a timing line 40 which synchronize the conversions with the line frequency of the AC power line 42 for the device 10. The AC line frequency is a large source of noise which adversely affects the biopotential signals sensed by the device, and this line frequency noise is minimized by synchronizing the analog to digital conversions with the line frequency. To accomplish this, an AC timer section 44 in a central processor unit 46 senses the AC power line frequency and provides four timing pulses on the timing line 40 at equal positions A, B, C and D on the sine wave for the AC line cycle. The timing pulses occur equal distances from the peak or 90° point of each half cycle and on opposite sides thereof. Ideally, these timing pulses occur at points on the half cycle which are 90° from the peak point. Thus, a timing pulse is provided at an equal position on the rise and fall curve of each half cycle, causing a conversion to occur in response to each timing pulse. Noise generated during the rise portion of the half cycle tends to be cancelled by noise generated during the fall portion.

[0052] The multiplexors 24 and 26, the filter amplifiers 32 and 34 and the analog to digital converters 36 and 38 form an isolation section 48 which is electrically connected to a subject by means of the sensor arrays 16 and 18. This isolation section is provided with a lower power dedicated power supply 50 which does not provide power sufficient to cause injury to a subject. The power supply 50 receives AC power from the AC powerline 42 and includes a dual isolation circuit including two transformers between the AC powerline and the isolation section which provide a dual barrier to the AC powerline. The power supply 50 converts the input AC to a low voltage DC which powers the isolation section 48. The isolation section is electrically isolated from the central processor unit 46 which is connected to the AC powerline 42. To achieve this electrical isolation, all signals between the isolation section and the central processor unit may be conducted over optical cables 52 as optical signals. Thus, the timing signals from the AC timer section 44 are converted to light pulses by a conversion unit 54, such as a light emitting diode, transmitted over an optical cable 52 and reconverted to electrical pulses by a reconversion unit 56. Similarly, the electrical digital outputs from the analog to digital converters 36 and 38 are converted to light pulses and transmitted to the central processor 46 where they are reconverted into electrical digital signals. Alternatively, an optoisolator chip shown in broken lines at 58 may replace the optical cables 52, conversion units 54 and reconversion units 56 to convert the electrical signals to optical signals and to accomplish the reconversion. The electrical digital signals from either the reconversion unit 56 or the optoisolator chip 58 are directed to a synchronous switching or de-multiplexor 60.

[0053] The de-multiplexor 60 is synchronized with the multiplexors 24 and 26 and provides timing signals on a line 62 which are transmitted as optical signals to the isolation section 48 where they are reconverted to electrical timing signals which are sent over a line 64 to the multiplexors. Digital filter arrays 66 and 68 in the software for the central processing unit include a dedicated digital filter such as two-pole, Infinite Impulse Response (IIR) filter, with a Butterworth response, for each electrode channel in the sensor arrays 16 and 18 respectively. Thus, as the multiplexors 24 and 26 are simultaneously transmitting analog signals from a selected sensor channel in the sensor arrays 16 and 18, the digital signals indicative of these analog signals are being directed by the de-multiplexor to the digital filters in the arrays 66 and 68 which are dedicated to those channels. When the multiplexors switch channels, the de-multiplexor switches to corresponding digital filters.

[0054] Filtered digital data from the digital filter arrays 66 and 68 are analyzed by a DC range sensing section 69 of the central processing unit (that is in fact formed by a software program) which is programmed to sense the magnitude of the DC electropotential signals represented by the filtered digital signals. Digital signals indicative of DC signals within a predetermined range of millivolts (for example −200 to +200 millivolts) are accepted while signals outside this millivolt range are rejected as spurious. The accepted signals are directed to processing section 70 of the central processor unit 46 having a RAM memory 72 and a ROM memory 74. This data is stored in memory and is processed by the processing section in accordance with a stored program to perform the condition sensing functions of the present invention. The processing section combines sensed electropotential data with data provided by a data input section 75, which may constitute a keyboard or a keyboard in combination with other data sensing units, all of which provide inputs to the processing section. The output from the processing section is connected to control the display on an indicator unit 76.

[0055] It should be understood that for clarity of description, sections of the central processor unit 46 have been illustrated as operative blocks, but these sections may constitute software controlled functions.

[0056] With reference to FIGS. 2-13, the operation of the apparatus 10 will be clearly understood from a brief consideration of the steps of the invention which the device is programmed to perform. During a test period initiated by the processing section 70, the electropotential between selected reference sensors 14 and 14aand selected test sensors, such as sensors 12 is measured, converted to a digital signal and stored for processing by the processing section 70. Sequential measurements between the reference sensor and the selected test sensors are taken. In accordance with the method of the present invention, the measurement indications on each individual channel are not averaged with those from other channels, but are instead kept separate and averaged by channel within the processing section 70 at the end of the test period. For the duration of a single test period, for example, from eight measurement channels, the processing section will obtain eight average signals indicative of the average electropotential for the test period between the reference sensors 14 and 14aand each of the test sensors in the sensor arrays 16 and 18. Of course, more reference sensors can be used, although only one reference sensor per array has been shown for purposes of illustration.

[0057] Having once obtained an average signal level indication for each channel, the results of the measurements taken at multiple sites are analyzed mathematically to determine the relationships between the average signal values obtained, and these signal values are often weighted for noise reduction and to compensate for biologic variables.

[0058] The basic measuring operation controlled by the central processing unit 46 will best be understood with reference to the flow diagrams of FIGS. 2 and 3. The operation of the unit 10 is started by a suitable start switch as indicated at 78 to energize the central processing unit 46, and this triggers an initiate state 80. In the initiate state, the various components of the device 10 are automatically brought to an operating mode, with for example, the indicator 76 being activated while various control registers for the central processing unit are reset to a desired state.

[0059] Subsequently, a test period is initiated at 82 wherein the various components of the system are tested for proper operability. During this test period, the electrode arrays 16 and 18 may also be tested to make certain that electrodes are being used which accurately measure DC biopotentials.

[0060] If all system components test out properly during the system test period, then timing of the analog to digital converters in accordance with the AC line frequency begins at 84 and the timing of the multiplexors and de-multiplexors begins at 86. With the analog to digital converters, multiplexors, de-multiplexors and digital filters in operation, it is now possible to monitor the biopotential signals from one or more test area during a monitoring period begun at 88. During this monitoring period, conditions in the test area contacted by the electrode arrays 16 and 18 are stabilized so that subsequent reliable measurements of DC biopotentials can be obtained. This monitoring period can be an extended time period, for example ten minutes, which is used for all patients and is sufficient to insure stabilization.

[0061] During the test period the digitized signals received from the various sequenced channels are monitored at 92 to determine whether or not each biopotential represented by the signals is within a predetermined range of millivolts. Digitized values indicative of DC signals outside this range are discarded at 94 and the remaining signals are used to provide an average or normalized value for each channel. Then, at 96, the central processor unit determines whether the test period has expired and the desired number of measurements have been taken, and if not, the collection of measurement samples or values continues. At the end of the test period, the average value for each channel is obtained at 98 by summing the values obtained for that channel during the test period and dividing the sum by the number of measurements taken. These average values are then processed in the various manners hereinafter described.

[0062] As shown in FIGS. 3, 5, 8, 16 and 17, in accordance with the present invention, the symptomatic breast 100 of a patient containing the suspicious lesion 102 is divided into four, equal, imaginary quadrants, one of which contains the lesion. The quadrants are defined by an imaginary horizontal line 104 which is perpendicular to an imaginary vertical line 106, both of which pass through and intersect at the nipple 108 of the symptomatic breast. Sensors from a sensor array, for example the array 16, are intended to be positioned on the symptomatic breast 100, and a lesion test sensor SC of that array provides a primary electropotential value used in the measurement of the present invention. Preferably, the lesion 102 is a palpable lesion so that the sensor SC can be centered directly over the lesion. Although only one sensor per breast quadrant has been shown for illustrative purposes in FIGS. 3, 5, 8, 16 and 17, multiple sensors can be used in each quadrant to provide an electropotential value for that quadrant.

[0063] In FIGS. 3 and 4, in accordance with the simplest form of the method and operation of the apparatus of the present invention, only the lesion test sensor SC and one other non-lesion test sensor in the sensor array 16 are used. This second sensor 12ais positioned in a quadrant of the symptomatic breast which is outside of the quadrant 110 containing the suspicious lesion. This could be a horizontally positioned quadrant 112, a vertically positioned quadrant 114, or a diagonally positioned quadrant 116. The reference sensor 14 may be positioned on the skin of the subject remote from the symptomatic breast, such as on the hand or the sub-xyphoid area of the subject, or directly over the nipple of the breast. During a test period, the processor 70 causes the same number of plural measurements to be taken by each of the sensors SC and 12aand an average is obtained at 98 for each sensor. Then, at 118, the processor subtracts the average value for the sensor 12afrom the average value for the sensor SC to obtain a differential value. The values used to obtain the differential value may be weighted at 117 in a manner to be described. Then this differential value is compared at 120 with a predetermined reference stored in one of the memories 72 or 74. Preferably the reference is a predictive index developed by stepwise regression using differentials from women with lesions the status of which has been identified to provide a multivariate model. Comparison of the differential value with this predictive index provides predictive information regarding the presence or absence of malignancy. If the differential falls within the malignancy ranges, malignancy is predicted at 122. Conversely, if the differential does not fall within the malignancy range, no malignancy is predicted at 124. Thus the predetermined reference provides the median value in a population predictive of the probability of cancer.

[0064] In testing the 55 year and under female population, the device 10 is used to obtain measurements from women with no breast lesions, women with a breast lesion who have been determined not to need biopsy, women with a breast lesion which has been shown by biopsy to be benign, and women who biopsy shows have a malignant breast lesion. The measured electropotentials have been shown to be consistently the highest for women with malignant lesions and consistently lowest for woman with no lesions. Measured electropotentials are higher for woman with lesions who have been determined not to need biopsy than for women with no breast lesions, and still higher for women with a breast lesion which biopsy has shown to be benign.

[0065] Referring to FIGS. 5 and 6, an enhanced noise rejection arrangement and operation for the apparatus 10 is illustrated. Here, in addition to the lesion test sensor SC in quadrant 110 of the symptomatic breast 100, non-lesion test sensors 12a, 12b, and 12care located in each of the non-lesion quadrants 112, 114 and 116 of the symptomatic breast. The processor 70 during a test period causes the same number of plural measurements to be taken by the sensors SC, 12a, 12band 12cand an average is obtained at 98 for each sensor as previously described. Then, as illustrated by FIG. 6, the processor 70 calculates a median value of the average values for the sensors 12a, 12band 12cat 126 and at 128 subtracts this median value from the average value from the sensor SC to obtain a differential value. Again, the values used to obtain the differential value may be weighted at 127 in a manner to be described. This differential value is then compared at 120 to the predetermined reference value previously described to obtain a prediction of malignancy at 122 or of non-malignancy at 124.

[0066] Alternatively, as illustrated by FIG. 7, rather than calculate the median value of the average values for the sensors 12a, 12band 12c, the processor 70 may be programmed to calculate the mean value of these averages at 130. This mean value is then subtracted at 132 from the average value from the sensor SC to obtain a differential value which is then compared to a reference at 120 as previously described. Weighting may be applied at 131 for the calculation of the differential value.

[0067] Turning now to FIGS. 8 and 9, a greatly enhanced noise rejection arrangement and operation for the apparatus 10 is shown. Here, as in FIG. 5, the non-lesion test sensors 12a, 12band 12care positioned in the non-lesion quadrants of the symptomatic breast 100. In this case, however, the asymptomatic breast 134 is divided into four quadrants 136, 138, 140 and 142 in the same manner that was done for the symptomatic breast, so that the quadrants in the asymptomatic breast match those in the symptomatic breast. Then sensors 12d, 12e, 12fand 12gare positioned in each of the quadrants 136, 138, 140 and 142 respectively to match the positions of the sensors SC, 12a, 12band 12crespectively on the symptomatic breast. During a test period, the processor 70, in addition to causing a plurality of measurements equal in number to be taken by the sensors SC, 12a, 12band 12c, also causes an equal number of measurements using the reference sensor 14ato be taken by the sensors 12d, 12e, 12fand 12g. Then, as previously described in connection with FIG. 2, the processor obtains an average for each sensor in the symptomatic and the asymptomatic breast.

[0068] Referring now to FIG. 9, at the end of the test period, the processor 70 samples at 144 the highest average value and the lowest average value from the sensors on the asymptomatic breast and then at 146 subtracts the lowest average from the highest average to obtain a maximum voltage differential value (MVD) for the asymptomatic breast. Also, as described in connection with FIG. 6, the processor computes the median value of the averages from the sensors 12a, 12band 12con the symptomatic breast at 148 and at 150 subtracts this median value from the average value for the sensor SC to obtain a basic differential value, and then, the processor adds the maximum differential voltage value from the asymptomatic breast to the basic differential value to obtain a final differential value (FDIFF). The values used to obtain the final differential may be weighted at 152 in a manner to be described.

[0069] At 154, the final differential value is, as previously described relative to FIGS. 4, 6 and 7, compared with a predetermined reference and if it doesn't fall within a determined range, an indication of no malignancy is given at 156. If on the other hand the final differential value falls within the determined range, malignancy is indicated at 158.

[0070] Referring to FIG. 10, a second form of the final differential can be calculated by the processor 70 by computing the median value (MED 2) of the averages for each sensor channel for the sensors 12d, 12e, 12fand 12gat 160. This median value for the asymptomatic breast is then added to the differential value at 162 which has been obtained by subtracting from the SC value the median value of the averages for the sensors 12a, 12band 12ccomputed at 148. These values may be weighted at 161 to obtain a final differential value. Now the processor operates in the same manner described with respect to FIG. 9 to obtain an indication of malignancy or non-malignancy.

[0071] As illustrated in FIG. 11, the processor 70 may operate at 130 in the manner described in connection with FIG. 7 to compute a mean value from the averages for sensors 12a, 12band 12cand then at 166 to obtain a differential value for the symptomatic breast by subtracting this mean value from the average value from sensor SC. Here, however, the processor obtains a mean value (MEAN 2) for the asymptomatic breast at 164 which is the mean of the average values from the sensors 12d, 12e, 12fand 12g. Then this asymptomatic breast mean value is added to the differential value at 166 to obtain the final differential value which is then processed at 154 in the manner described with respect to FIG. 9 to obtain an indication at 156 or 158. The values used to obtain the final differential value may first be weighted at 165.

[0072] In FIG. 8, only four sensors have been shown on each breast, but fewer or more sensors could be used so long as the sensors on the asymptomatic breast are placed at sites which mirror image the sensor placement sites on the symptomatic breast. Also, preferably the same number of sensors are used in each non-lesion quadrant of the symptomatic breast.

[0073] Turning now to the weighting factors which may be used with the values arithmetically combined to determine the differential value in FIGS. 4, 6 and 7 and the final differential value in FIGS. 9, 10, and 11, these weighting factors may be stored in the memories for the processor 70, may be computed by the processor in accordance with data provided at the data input 75 relative to patient physiological conditions, or preferably are a combination of both. Considering first the non-physiological weighting factors, these are determined on the basis of training data obtained using the apparatus 10 on a test population of women within the desired age range with palpable lesions of a known type to establish the predictive value of the variables used to determine the differential or final differential value. For example, to obtain the differential values in FIGS. 4, 6 and 7 and the final differential values in FIGS. 9, 10 and 11, the average value of the electropotentials from SC is weighted by a value which is higher than the value used to weight the average of the electropotentials for sensor 12ain FIG. 4, the median of the electropotentials from sensors 12a, band cin FIGS. 5, 9 and 10, and the mean of the electropotentials from sensors 12a, band cin FIGS. 7 and 11. As an example, in FIGS. 5, 9 and 10, the weighting of the average SC and the median at 128, 150, and 162 could be in accordance with the following specific equation:

[0074] 0.0623SC−0.0380MED=DIFF

[0075] When values from the asymptomatic breast are added to the equation, these values are weighted by a value which is less than the weighted value applied to SC, MED or MEAN. For example, in FIG. 9 at 150, the equation could be:

[0076] 0.0623SC−0.0380MED=DIFF+0.0214MVD=FDIFF

[0077] Finally, a constant can be added to each of the calculations for the differential values in FIGS. 4, 6 and 7 and the final differential values in FIG. 9, 10 and 11. This constant has two components; a first due to the mean of the variables in the training data and a second due to the prevalence in the data which affects the output scale. By applying this constant, the equation for FIGS. 5, 9 and 10 at 128, 150 and 162 could be:

[0078] −4.6689+0.0623SC−0.0380MED=DIFF

[0079] Thus in FIG. 9 at 150, the complete equation could be:

[0080] −4.6689+0.0623SC−0.0380MED=DIFF+0.0214MVD=FDIFF

[0081] In addition to weighting on the basis of training data, each of the arithmetic calculations in FIGS. 4, 6, 7, 9, 10 and 11 can be weighted in accordance with physiological factors present in the patient which affect the outcome of the calculation. These physiological factors include patient age, menses, i.e. the time in the patient's menstrual cycle, the effects of hormonal agents and the time of day when the test is taken. Weighting factors for one, a plurality or all of these physiological factors can be applied to the arithmetic calculation which results is the differential or final differential value which is evaluated.

[0082] Because of the increasing prevalence of cancer with age, a statistically significant improvement in discrimination between malignancy and non-malignancy is found when the calculation is weighted for age. Thus a weighting factor which varies with age can be added to the value of SC in FIGS. 4, 6, 7, 9, 10 and 11. The age of the patient is entered at the data input 75, and the processor 70 provides a weighting factor from a stored factor index based upon age. This age weighting factor is added to SC as follows:

[0083] FIG. 4: Age factor+SC−AV 12a=DIFF

[0084] FIGS. 6, 9, 10: Age factor+SC−MED=DIFF

[0085] FIGS. 7, 11: Age factor+SC−MEAN=DIFF

[0086] Obviously the age weighting factor or the other physiological weighting factors can be combined with the weighting factors previously described. Thus, when age is taken into account, the complete equation for FIG. 9 could be:

[0087] −4.6689+Age factor+0.0623SC−0.0380MED=DIFF+0.0214MED=FDIFF

[0088] It is a well known fact that various changes occur in the human female body during the phases of the menstrual cycle. These changes result from changes in levels of certain hormones, and a spectrum of changes occur in the female breasts which affect the measurements made by the device 10. Interestingly, the measurements of malignant lesions have been found to remain relatively stable throughout the menstrual cycle, but benign lesions measurements vary significantly as illustrated by the chart in FIG. 12 which shows the average measurements for a plurality of women taken at various times in the menstrual cycle. This directly affects the separation of potentials corresponding to benign and malignant tumors. Consequently, a test result which indicates malignancy may be accurate regardless of menses, but when the result is indicative of a benign tumor the result may not be accurate due to the effect of menses. When the processor 70 senses that the results have occurred during a period in the menstrual cycle where benign results have an unacceptably high probability of being false, the processor can cause the display to show a recommendation to delay the test until a specified date along with a range of dates during which a retest may be performed. Alternatively, a retest could be performed immediately by first providing an input to the processor to have the processor add a weight factor to SC or another electropotential variable in the equation. Also it is possible to have the processor add a menstrual variable to the entire equation which, like the weight factor, is determined by the period in the menstrual cycle to compensate for the variation in benign measurement indications. The addition of this factor during a remeasurement would determine whether or not the differential was truly in the benign range.

[0089] Also during menses, when the results of previously taken adjunctive cancer tests are input at 75 to the processor 70, these tests relative to the real time test taken by the device 10 can be weighted in accordance with menses. With reference to FIG. 13, relevant data input at 75 can include patient age, family and personal medical history, medication or hormone usage, results and dates of recent adjunctive cancer tests (such as mammography or ultrasound), results and dates of recent hormone measurement tests (such as sputum or urine sample analysis), and date of onset of last menstrual period and time of day. Using all related menstrual data, including hormone usage and test results, the system processor 70 performs an analysis 168 and then decides at 170 if a cancer test should be performed by the system 10 at this time. If the analysis indicates that the results may have an unacceptably high probability of being false, the indicator 76 is instructed to display a recommendation 172, such as delaying the test until a specified date, along with a range of dates during which the test may be performed. If the analysis indicates that the results may have a reliable outcome, the indicator is instructed to display a recommendation to proceed with the test 180. The real time test is then performed at 182, collecting and processing the potentials from sensors 12. The processor 70, according to its programmed algorithm, combines at 184 the results of the analysis 168 with that of the real time test 182. Finally, the results are displayed at 186 via the indicator 76 in the form of the real time test results of the real time test 182 and the patient data analysis 168.

[0090] The combined results 186 may take several forms. As an example, it may be a weighted combination of the results of the real time test and the recent adjunctive tests. The weights are determined by the dates of menses and/or hormonal levels. The rationale for this is that the real time test is more reliable under certain conditions relating to the hormonal levels and the adjunctive tests are more reliable under certain other conditions. As a simple example, during the second week of the menstrual cycle the real time test and the adjunctive test are equally reliable. Thus, they are weighted equally in the final results. On the other hand, during the fourth week of the menstrual cycle, the adjunctive test is only half as reliable as the real time test and it is given a 25% weighting vs. a 75% weighting for the adjunctive test. This may be reflected in a statistically derived sensitivity and specificity index.

[0091] In addition to the menstrual cycle, another factor which appears to affect DC biopotential breast measurements is the time of day during which a measurement is taken. Physiological changes which apparently occur in females as a day progresses cause significant changes in the data obtained from biopotential breast measurements taken with the device 10, particularly when a palpable lesion is present. For women measured at the same time in the menstrual cycle or for measurements taken in post menopausal women, there have been found to be substantial differences in the results of measurements taken from the same women early in a day and later in the day. Using 2:00 pm as a hypothetical boundary time, tests have shown that DC biopotential breast measurements taken in the afternoon from an electrode placed over a palpable breast lesion may be 8 to 10 mV higher than the same measurements taken in the morning. Consequently, the time of day entered at data input 75 for the apparatus 10 can be a significant factor for measurements taken by one or more of the sensors 12. The time of day can be used by the processor 70 to apply a weighting factor to the measurements taken in the manner previously described.

[0092] The effects of estrogen increase the electropotential difference between depolarized and polarized tissue more when malignant tissue is present than when a lesion is benign. This in part is responsible for the effect menses has on DC biopotential breast measurements. The effects of estrogen can be turned to practical use by applying estrogen to a symptomatic breast 100 before measurements are taken. This can be accomplished by adding estrogen to the ion conducting gel or cream 19 in the electrodes applied to the symptomatic breast, or estrogen can be applied directly to the symptomatic breast before application of the electrodes.

[0093] In applying electrodes to the symptomatic breast, it has been found to be beneficial to maintain a pressure on electrode SC which forces this electrode downwardly against the breast over the lesion. The application of this downward pressure to only the electrode SC can be achieved by the design of a harness or brassiere 188 which mounts the electrodes for the symptomatic breast. The electrode SC can be mounted to project outwardly further from the mounting unit 188 than the remaining test electrodes as shown by the broken lines in FIG. 14. Resilient cushioning 190 between the electrode SC and the mounting unit 188 will force the electrode SC against the symptomatic breast with a greater force than that applied to adjacent test electrodes on the symptomatic breast.

[0094] It has been observed that the gradient around a tested lesion is relatively large among malignant cases. As shown in FIG. 15, this gradient may be approximated by arranging electrodes in one or more concentric circles on the symptomatic breast with the electrode SC at the center of the circles over the lesion. At least four electrodes SU, SI, SO and SL are spaced on a concentric circle 192 around the central electrode SC. Using measurements from these five electrodes SC, SU, SI, SO and SL relative to a reference sensor 14, the gradient may be approximated.

[0095] Curve_=SC−¼(SU+SL+SI+SO)

[0096] If SC through error has not been placed sufficiently near the site of the maximum depolarization, there could be diminished separation between malignant and benign measurement results. When this occurs, the above equation fitted to the electropotential response surface of the symptomatic breast can be used to estimate the location of maximum response for SC. The five sensors SC, SU, SL, SI and SO in the quadrant of the lesion can be seen in the language of factual design as a 2²with a center point. With this design, the curvature, main effects and interaction between two orthogonal, directional axes A and B can be checked. If SC is properly placed at the location of maximum response, the curvature contrast derived from the above equation should be statistically greater or equal to zero. If this curvature contrast is statistically less than zero, the sensor SC may have been incorrectly placed.

[0097] The raw electrical measurements provided by SC, SU, SL, SI and SO may be in a wide range, and to normalize the gradient, curve_s is divided by the standard deviation (std) of SU, SL, SI and SO. The variable to be evaluated is:

[0098] Curve_sd=Curve_s/std(SU+SL+SI+SO)

[0099] To provide more accurate measurements based upon those from SC and surrounding sensors, the number of sensors concentrically spaced around the lesion can be increased beyond the four sensors described as indicated at 194, 196, 198 and 200 in FIG. 15.

[0100] In some cases, a lesion is not readily discemable and the sensor SC cannot accurately be located over the lesion. Since all conduits in the breast lead to the nipple, the sensor SC can be located over the nipple 108 on a suspected symptomatic breast 100 as shown in FIG. 16. Then, preferably at least one non-lesion test sensor 12a, 12b, 12cand 12dis placed in each of the four quadrants of the symptomatic breast as illustrated in FIG. 16. However, if non-lesion test sensors are used in only one quadrant, the process described in connection with FIGS. 3 and 4 will be performed.

[0101] When non-lesion test sensors are located in all four quadrants of the symptomatic breast, then a process similar to that described in connection with FIGS. 5 and 6 or 5 and 7 is used except now average values for four sensors 12a, 12b, 12cand 12dare used to calculate either a median or mean value to be subtracted from the average value for SC rather than average values for only three sensors as previously described.

[0102] Also, when a nipple sensor SC and non-lesion test sensors are used on the symptomatic breast, enhanced noise rejection can be achieved by using mirror image sensors as shown in FIG. 16 on the asymptomatic breast 134. Again, these sensors 2e, 12f, 12g, 12hand 12iprovide five averaged outputs instead of the four described in connection with FIGS. 8-11, but these five average outputs are used to provide a maximum voltage differential value (MCV), a median value, or a mean value for the asymptomatic breast as previously described. These asymptomatic breast values are combined with the symptomatic breast values as described relative to FIGS. 8-11.

[0103] The apparatus 10 may be used to screen a breast 100 as illustrated in FIGS. 17 and 18. Here, at least one sensor 12a, 12b, 12c, 12dis placed in each quadrant of a breast to be screened and during a screening period, the apparatus operates in accordance with the process described in connection with FIG. 2 to obtain an average potential value for each breast quadrant. The processor 70 then compares the average potential values for each breast quadrant, and if one breast quadrant value varies from the others by more than a predetermined amount, the processor identifies that breast quadrant as a potential lesion containing quadrant. As shown in FIG. 18, the processor may operate to average the breast quadrant values to obtain a breast average value at 202 and to then compare each individual breast quadrant value with the breast average value at 204. If a breast quadrant value varies from a breast average value by more than a predetermined amount x at 206, the processor 70 identifies that breast quadrant as a potential lesion containing quadrant at 208. Once this occurs, the processor may then institute a test period using the sensor in the lesion quadrant as SC and the remaining one or more quadrant sensors as described in connection with FIGS. 3-11, or a nipple sensor SC may be added as described in connection with FIG. 16. If the quadrant average value does not vary by more than X at 206, no lesion is indicated at 210.