DE102006021042A1 - Mammogram analyzing device, has unit for executing threshold value procedure, which uses threshold value, where device is designed and provided to determine contour line of mammogram based on image characteristics of mammogram - Google Patents

Abstract

The device has a threshold value processing unit (12) for executing a threshold value process, which uses a threshold value, and a scan line processing unit (13) for executing a scan line procedure, which determines gradients of a mammogram along different scan lines. The device is designed and provided to determine a contour line of the mammogram enclosing an object area (20) based on image characteristics of the mammogram. An independent claim is also included for a method for analyzing a mammogram.

Description

The The invention relates to a device and a method for computer-aided analysis differ in their image characteristics of distinctive mammograms.

such Devices are designed, a mammogram for computer-aided analysis to consult a doctor in the diagnosis and diagnosis of the Support mammogram. The Mammogram to be processed is the device in digital Form, wherein the mammogram, depending on the used X-ray device, analog recorded and subsequently digitized or digitally recorded. Such devices can used in this context in particular for mammography screening in the context of checkups the Mum a patient, for example, taken for early breast cancer detection and to be examined.

Basically exists in the computer-aided Analysis of a mammogram the problem that the contour lines, the is called an object area (also known as a region of interest (ROI)) mammogram limiting lines in the mammogram to be detected have to, to detect the mamma shown in the object area of the mammogram and segment. Are the contour lines of the mammogram recorded, Thus, the object area by means of a binary masking image, at in the form of a mask, the object area by the pixel value 1 and a background area is represented by the pixel value 0, or based on the contour lines, the one through a list of contour points coded coherent, Closed curve corresponds, displayed and output.

For computer-aided analysis from mammograms, in particular for computer-assisted microcalcification detection and for the detection of compaction herds in a mammogram pictured mom, and for the processing of mammograms at diagnostic facilities A doctor is required by the mammogram Mum pictured as accurately as possible and capture from the image background to separate. About that In addition, it may also be necessary to have the mom of others in the mammogram contained objects, such as the muscle pectoralis or external, in the form of markers contained in the image objects, or dense glandular tissue To distinguish the mamma from the surrounding adipose tissue.

in the Mammograms to be examined in mammography examinations differ fundamentally in their image characteristics, for example, depending on the examination modality, from the manufacturer of the X-ray machine and of the type of recording, significantly different from each other, so the mammograms not reliably examined by using simple detection methods can be. Under image characteristics here the bit depth, the dynamic range, the gray value distribution in the object area and in the background area and understood the image noise that characterizes the nature of a mammogram and for different mammograms can be very different. Especially here have the background areas and also the object areas different mammograms frequently - depending on the Type of recording and the manufacturer of the X-ray machine - different gray value distributions and Structures on, for example, conditioned by that in the background areas contained noise, so that different mammograms on the one hand no or only slightly noisy background areas (in particular in directly-digitally recorded mammograms) or on the other hand very noisy Background areas may have in extreme cases, even brighter than the object area of the image to be imaged Can be mom. Additional difficulties for the Processing prepare about it out in the mammogram contained markers or bright edges along the picture edge, which is a reliable Detection of the contour line of the object area complicate or even impossible do.

A method for computer-aided automatic processing of medical images, for example, from EP 0 731 959 B1 in which a contour line of an object area is determined, in which a global histogram of the mammogram to be examined is evaluated. The EP 0 731 959 B1 In this case, a method is provided in which an image representing an anatomical feature is segmented, a gradient image in the region of the anatomical feature is derived from the segmented image, local extremes are determined, and an inner and outer skin line is determined in this way. Basically, this poses the problem that such a method is not applicable to all types of mammograms to allow a reliable detection of a contour line, regardless of the type of mammogram.

Of the present invention is based on the object, a device and a procedure available to provide a computer-aided analysis of themselves their image characteristic distinctive mammograms for reliable detection allow the contour line of an object region of the mammogram.

This task is performed by an object solved with the features of claim 1.

The Invention is based on the knowledge that in terms of the picture characteristic strongly distinguishing mammograms a uniform procedure is not sufficient for the detection of an object area, but dependent different methods used by the image characteristic Need to become, to be reliable a contour line enclosing an object area of a mammogram detect. The device according to the invention provides therefore funds available by means of which different methods alternatively or in combination A mammogram can be applied to the reliable and robust way determine the object area of the mammogram enclosing contour line.

through The invention provides a uniform interface, by means of which different mammograms, as direct-digital or later digitized image and in their image characteristics, So for example, in terms of bit depth, recording, manufacturer and image noise, can be different, can be processed for all Mammograms processed the object area of the mammogram descriptive contour line is output. The device can provide a user with query options in this context, by means of which the object area and thus the object to be detected can be specified by the user, so depending on the User presets a contour line that the mom with or without skin flaps, the pectoralis muscle or the dense glandular tissue becomes.

advantageously, For example, the device may provide a means for performing edge treatment for determining the lateral boundary lines of the mammogram exhibit. In this way it is possible before processing the mammogram to pre-process the mammogram to detect the contour line around the detection of the contour line disturbing edge strips and objects to remove in the mammogram. By means of the edge treatment can hereby in particular objects in the form of markers intended for marking in Image are included, can be eliminated.

Farther the device preferably has a means for performing a Contour processing for smoothing and correction of the contour line. In the detection of the contour line arises frequently the problem that sections of the contour line is faulty or even can not be determined. By means of contour processing, such sections then become smoothed and added, in this way, a closed, the object area exactly mapping contour line to obtain.

Furthermore For example, the device may be a means of scaling and positioning of the mammogram, by means of which the mammogram in his Size and position can be adjusted so that, on the one hand, the object area of the mammogram the mammogram optimally fills and on the other hand, the mammogram is comparable to other mammograms is. This is based on the idea that, for example during screening a patient, taken in principle both momas of a patient and have to be compared to the diagnosis. By providing funds by which Mammograms mammograms are automatically entered their size and position be matched, the diagnosis for a Doctor then relieved considerably.

In In a further embodiment, the device additionally has a Findings on, by means of which a doctor treating enter a finding in an interactive way directly into the mammogram can, with the device processing the input, with the Mammogramrn relates and saves. In this way, the findings input for one Doctor's diagnosis of a mammogram made easier and more reliable since the findings are entered directly into the mammogram and transmission errors, For example, when creating a report using a separate input device can occur, be avoided.

The Task is about it beyond solved by a method having the features of claim 6.

The inventive method offers the advantage that differing in their image characteristics Mammograms, in particular with regard to the bit depth, the dynamic range, Gray value distribution and noise, for example, depending on from the manufacturer of the X-ray device used and the recording modality can vary, processed can be being independent of the type of mammogram a uniform output in the form of a describing the object area contour line or one based on the Contour line created masking image is done. Essential advantage The method is thus the versatility, the one computerized Allows analysis of a wide variety of mammograms and in this way apparently completely makes different mammograms comparable.

In principle, either a threshold value method or a scan line method is used depending on the type of mammogram. It is also conceivable in this connection to provide different variants of the methods for different types of mammograms, for example by processing direct-digital mammograms and CR mammograms (CR) and mammograms with a very noisy background using different threshold methods, while using mammograms with a pronounced, distributed gray value gradient in the background a scan line method is used, which manages without a global threshold.

Especially In this case, it is conceivable and advantageous for a product to be processed on the basis of one Mammogram automatically selected which method is used to detect the contour line of the mammogram is used. As part of the procedure will then automatically examined the image with regard to its image characteristics and, due to the image characteristic, the method to be used selected.

After this the contour line of the object area, so for example the one Mammal bordering contour line, has been detected, the object area in the form of a masking image, which is determined on the basis of the contour line is displayed, and / or in the form of the contour line and output become. The main advantage here is that for different, in terms of their image characteristic of distinctive mammograms a uniform Output generated in the form of a masking image or a contour line becomes, thus independent of the type of mammogram a uniform edition provided becomes.

In An advantageous embodiment of the method is prior to detection the contour line an edge treatment to determine the lateral Boundary lines of the mammogram performed. In the border areas of a Mammograms can be artifacts in the form of bright stripes or mammogram markers to mark the mammogram, the detection of the Disturb the contour line and can lead to false results in the detection of the contour line. As part of The edge treatment may therefore be disturbing objects in marginal areas of the mammogram, and cut off to possible sources of error while to eliminate the detection of the contour line from the outset.

advantageously, In the edge treatment, scan lines become perpendicular to one outer edge of the mammogram, along these scan lines the mammogram in each case the position of the steepest gradient of the mammogram determined along the scan lines and from the positions of this steepest Gradients the lateral boundary line of the mammogram determined. are for all four outer edges of the mammogram the lateral boundary lines have been detected so will the outside cut off the boundary lines arranged areas of the mammogram, Thus, the mammogram is reduced in size and in the peripheral areas of the mammogram Mammogram arranged disturbing Removed objects.

In A further embodiment of the method can after the detection the contour line a contour post-processing for smoothing and Correction of the contour line to be performed. This may optionally be required if the detected contour line is incomplete or is faulty, for example, if the mamma to be imaged does not Completely is shown in the mammogram. Furthermore, by means of contour postprocessing, objects, for example, the pectoralis muscle, of the object area to be imaged, For example, the mom, be separated, so that the contour line exclusively the Includes object area, but not others, outside the object area arranged objects.

advantageously, the contour processing takes place by the detected contour line is decomposed into contour segments, the contour segments are smoothed missing contour segments added and faulty contour segments are removed. The addition missing Contour segments occurs in that the contour line between two detected contour segments is interpolated to the contour line close. The removal of defective contour segments is done by not to the desired Object area belonging Objects are cropped by extrapolating the contour line of the object area becomes.

As explained above the basic idea of the invention is to provide different methods for Detection of the contour of the mammogram describing the object area to use for different Mammograms a consistent interface to reliable and To provide robust detection of the contour line. In principle, be in this context both threshold method and scan line method used, depending from the image characteristic of the respective mammogram to be processed the appropriate method selected becomes. Fundamentally, a distinction is made in this context between Threshold method that has at least one global threshold for detecting the contour line of the mammogram, and scan line method, that can do without a global threshold.

The basic procedure in the thresholding method is to calculate a histogram of the mammogram from the histo at least one threshold value is determined, the mammogram is binarized on the basis of the at least one threshold value and a contour line is determined from the binarized mammogram. In this context, binarization is understood to mean that the pixels of a mammogram whose gray values are above the threshold value have the value 1 and the pixels whose gray values are below the threshold value are assigned the value 0 in order to place the mammogram in the object area in which all pixels are grouped with the value 1, and to separate the background area with all other pixels. The contour line is then determined by tracking the edge of the object area.

Depending on The mammograms to be processed can have different variants Such a thresholding be used. In a The first variant of the threshold value method is used to determine the at least one threshold value starting from the histogram go through from the lowest gray value, a local maximum of Histogram, the steepest gradient of the falling edge of the histogram searched for the local maximum, based on the steepest Gradients a threshold and determines the threshold for binarization of the mammogram used. Such a threshold method, the histogram of the mammogram from below, starting with the lowest gray value of the mammogram, analyzed upwards, is especially suitable for processing mammograms with strong noisy background, for example in retrospectively digitized Mammograms (film scan modalities). For such mammograms the exact position of the contour line is not exactly defined and is visually difficult to estimate. For the Separation of background area and object area becomes the threshold then in the descending flank of the first histogram mountain, the represents the noise, arranged, and binarized by means of this threshold, the mammogram. It is conceivable in this context, the resulting contour line using morphological algorithms (closing, opening) later smooth.

In an alternative variant of the thresholding method becomes the histogram descending starting from the highest Traversing the gray value, a local maximum of the histogram is determined, searched for a minimum of the histogram following the local maximum and determines a threshold based on the minimum. These steps are repeated until the histogram is completed is and therefore for each local maximum having a subsequent minimum Threshold has been determined. In this way, a number different thresholds, each for binarization of the mammogram and to determine different contour lines be used. From the thus determined different contour lines then becomes the smoothest contour line selected and output as a contour line of the object area. Such a thing Threshold method is particularly useful for detecting the Contour line for direct digital mammograms or CR mammograms.

For mammograms however, with a pronounced Gray level gradient in the background area of the mammogram is a threshold value method not readily usable. Therefore, it is preferred for such mammograms used the scan line method without a global threshold manages to detect the contour line. In the scan line method are preferred starting from a pinning point of the mammogram in a central region of the mammogram, for example in the center of gravity of the mammogram, is arranged, spanned radial scan lines and read the mammogram along the scan lines. Based on Gray values of the mammogram along the scan lines will then be used for each scan line each determined a contour point, wherein the contour point the place corresponds to the maximum gradient along the considered scan line. By Connection of the thus determined contour points of the individual scan lines then the contour line is established and output.

advantageously, can be used both in the different variants of the threshold method or be provided in the scan line method, the mammogram by means of a locally acting on a group of pixels of the mammogram Mean value filter before the actual detection of the contour line to filter. Such average filters are used in one area of the mammogram, for example for a group of 3 × 3 or 7 × 7 pixels, locally to form an average and thus the gray value distribution of the mammogram to smudge and smooth, so that high-frequency Noise is suppressed and removed from the mammogram.

In an advantageous embodiment of the method, the determined contour line is used to scale and position the object described by the contour line and represented by the mammogram. The basic idea here is that a doctor who wants to find different mammograms representing a patient's breasts in different views makes it possible, by means of a uniform positioning and scaling of the mammograms, to present the mammograms in a way that makes the diagnosis and comparison of the mammograms simplest Way is possible. For this purpose, for each of the mammograms to be examined, one by the Determining the image focus point line and a rectangle enclosing the contour line of the object area and optimally scaling and positioning each mammogram on the basis of the contour line and the rectangle, taking into account the scaling of the other mammograms. By means of scaling and positioning, the mammograms to be compared are then displayed in such a way that they are comparable on the one hand and the object to be represented, ie the mamma, optimally fills the mammogram, whereby background areas which contain no information are cut off.

In a further variant of the method can be determined by the determined Contour line of the object represented by the mammogram an interactive entry of a finding by a doctor in the Mammogram determines the coordinates of the input and in a desired Output format to be transformed. To be fundamental in this way allows a doctor to enter a finding directly in the displayed mammogram the analysis of the mammogram the information contained in the findings automatically process, display and save. Especially it is advantageous in this context if the coordinates the doctor's input can be transformed into a time model, by means of which the position of the finding in the mum on distinct and clear Way can be displayed and displayed.

Of the The idea underlying the invention is based on the following the embodiments illustrated in the figures will be explained in more detail. Show it:

1 a schematic representation of a device for computer-aided analysis of differing in terms of image characteristics mammograms;

2 a schematic representation of a first threshold method for detecting a contour line of the mammogram;

3a . 3b a mammogram and an associated, indicating the gray scale distribution of the mammogram and in accordance with the Schwellwertverfahrens according to 2 generated histogram;

4 a schematic representation of a second threshold method for detecting a contour line of the mammogram;

5a - 5d a mammogram and from the mammogram in accordance with the Schwellwertverfahrens according to 4 derived histograms;

6a - 6c schematic representation of a scan line method for detecting a contour line of the mammogram;

7a Mammogram with scan lines spanned in the mammogram for detecting the contour line;

7b Representation of the mask for determining the threshold value in step 305 according to 6a ;

8th schematic representation of the method for edge treatment;

9 Mammogram with mammogram scanning lines for mammogram edge treatment;

10 schematic representation of the method for contour finishing;

11 Mammogram with a determined contour line;

12a . 12b Sketches for explaining the method for contour processing;

13a Mammograms of a right and left mamma in craniocaudal and mediolateralobliquer view;

13b optimally positioned mammograms according to 13a ;

13c optimally positioned and scaled mammograms according to 13a ;

14a Detail of a mammogram with a registered in the mammogram input of a finding and

14b . 14c Standardized findings masks for displaying an entered finding.

In 1 is a device 1 for the computer-aided analysis of mammograms, which serves to process mammograms differing in image characteristics. Image characteristic here is understood to be the characteristic peculiarity of the mammogram, which can be defined for example by the bit depth, the dynamic range, the gray value distribution of the mammogram and the noise contained in the mammogram and can vary greatly depending on the type of recording modality used and the manufacturer of the X-ray apparatus , Fundamental problem with such mammograms is that in terms of image characteristics under In general, outgoing mammograms are not readily processed to reliably detect and segment a mammogram presented in a mammogram for different types of mammograms. In particular, in this case, the background areas of the mammograms with respect to their gray value distribution and structure - especially depending on the manufacturer of the X-ray device - vary greatly and additionally included in the image markers or bright edges along the outer edges of the mammogram interfere with the detection of the contour line.

Exemplary mammograms, which differ greatly in terms of their gray value distribution, their dynamic range and their noise, are in 3a . 5a and 7a shown. The mammograms 2 form here in an object area 20 one each from a background area 21 mum, with the representation of the mum in terms of contrast between object area 20 and background area 21 , the recognizability of the contour line of the mamma and the structure in the mamma of mammogram 2 to mammogram 2 varies greatly. So is in 3a presented a mammogram that has been recorded analog and subsequently digitized and therefore a weak contrast between object area 20 and background area 21 and has a relatively high noise. In 5a is a direct digitally recorded mammogram is characterized by a high contrast and low noise. In 5a Again, an analog recorded and subsequently digitized mammogram is shown, which has a pronounced gray scale progression both in the object area 20 as well as in the background area 21 having. A problem therefore exists for all types of mammograms a uniform interface for detecting the contour line between the object area 20 and background area 21 to provide.

By means of the device 1 a means is provided to process a wide variety of mammograms 2 allowed. In this case, the knowledge is exploited that different with regard to their image characteristics Mammogramme 2 can not be processed with a uniform method for detecting a contour line enclosing an object, but rather for a reliable and robust detection of the contour line different methods are required. The device 1 is designed so that, depending on the image characteristics of each processed mammogram 2 the appropriate method is selected and used to detect the contour line. The device 1 thus provides an interface by means of which a variety of mammograms 2 be processed and for all mammograms 2 a uniform output is provided in the form of the contour line and / or the masking image determined from the contour line.

The device 1 for computer-aided analysis, according to 1 an input unit 10 on, by means of which a digital mammogram 2 into the device 1 is entered. The mammogram 2 can be generated in this context, for example by means of a direct-digital recording method or subsequently, when using analog X-ray equipment, digitized. The mammogram 2 is then sent to a processing unit 11 forwarded a means 12 to carry out a threshold process, an agent 13 for performing a scan line method, means 15 to perform a marginal treatment, an agent 16 to perform a contour postprocessing, a means 17 for scaling and positioning the mammogram 2 and a result input device 18 to enter a report. The processing unit 11 serves in this context to process the mammogram by the contour line of the object area 20 of the mammogram 2 represented object, in particular the mamma, is determined and output. The device 1 also has an output unit 14 on which a user uses the object area 20 of the mammogram 2 in the form of the contour line or a masking image determined from the contour line.

The device 1 may be designed in particular as a workstation, which is a mammogram to be processed 2 from an X-ray machine receives the mammogram 2 processed and via the output unit 14 , for example, a monitor, outputs and displays. In this context, the funds provide 12 . 13 . 15 . 16 . 17 . 18 not necessarily separate units, but rather may be software implemented in the workstation and thus be implemented by the workstation. The separate presentation of the funds 12 . 13 . 15 . 16 . 17 . 18 according to 1 serves in this context only for clarity. The findings input device 18 may for example be realized by the conventional input devices of the workstation, in particular a keyboard and a mouse, by means of which a physician creates a finding and enters into the workstation.

• 1. Für Mammogramme 2 mit verrauschtem Hintergrund, insbesondere für Mammogramme, die mit auf analoge Weise aufgenommen und nachträglich digitalisiert worden sind, wird ein erstes Schwellwertverfahren eingesetzt.
• 2. Bei Mammogrammen, die auf direkt-digitale Weise oder mittels CR-Modalitäten erfasst worden sind, wird ein zweites Schwellwertverfahren eingesetzt. Diese Mammogramme 2 zeichnet aus, dass der Hintergrundbereich in der Regel einen sehr niedrigen Grauwert und ein niedriges Rauschen aufweist.
• 3. Für Mammogramme 2 mit einem ausgeprägten Grauwertverlauf im Hintergrund wird ein Scanlinienverfahren eingesetzt, das ohne globalen Schwellwert auskommt.

Depending on the type of mammogram to be processed 2 a threshold method or a scan line method is used. In particular, three different types of mammograms can be distinguished in this context, for each of which different methods for detecting the contour line are used.

• 1. For mammograms 2 with a noisy background Reason, in particular for mammograms that have been recorded in an analogous manner and subsequently digitized, a first threshold method is used.
• 2. Mammograms acquired in a direct-digital manner or by CR modalities use a second threshold method. These mammograms 2 indicates that the background area usually has a very low gray level and low noise.
• 3. For mammograms 2 with a pronounced gray value gradient in the background, a scan line method is used which does not require a global threshold value.

Depending on the type of mammogram 2 then the respectively suitable method for detecting the contour line is used. The selection of the respective method to be used can be within the device 1 take place automatically, the image characteristic of a mammogram to be examined 2 for example, based on one from the mammogram 2 derived histogram, which is the gray scale distribution of the mammogram 2 indicates, is analyzed.

The individual methods for the detection of the contour line of a mammogram 2 will be described in detail below.

In 2 is shown schematically the course of the first threshold method for images with noisy background. 3a by way of illustration, by way of example only, with the method of FIG 2 to be processed mammogram. First, in the method according to 2 in step 101 the mammogram 2 subjected to an edge treatment, in the context of which bright edge strips and markers or the like in the edge region of the mammogram 2 are cut off (the method for performing the edge treatment will be explained later in detail). Subsequently, the mammogram 2 by means of a mean value filter in step 102 low-pass filtered. As a mean value filter, a filter can be used in this context, the gray values of the pixels of the mammogram 2 smeared over a range of, for example, 3 x 3 pixels by forming the average of the pixels over this range, the average filter on the entire mammogram 2 applied and the mammogram 2 is so low-pass filtered.

From the filtered mammogram 2 will be in step 103 a histogram is calculated, in which the distribution of gray values of the mammogram 2 is shown. Such a histogram, from the mammogram 2 according to 3a has been derived is in 3b with the lowest gray values at the left edge of the histogram and the highest gray values at the right edge of the histogram. Horizontal here is the index i of the gray values and vertically the frequency of the gray values in the mammogram 2 applied.

From the histogram is in 104 a threshold for the binarization determined. The threshold is determined by taking the histogram from the lowest gray value (far left in the histogram in 3b ) in ascending order until a local maximum Max is found which is above a frequency threshold S. The frequency threshold S in this case represents an arbitrarily selected threshold for the selection of a local maximum, which may for example be given by S = number of pixels of the mammogram / 256.

In other words, the frequency threshold indicates that a gray scale to qualify for a local maximum is at a frequency of 1/256 of the total number of pixels in the mammogram 2 must be represented. Starting from the local maximum Max, the position of the largest falling gradient G of the histogram is searched for the local maximum Max, ie the steepest point in the falling edge of the histogram after the local maximum Max. The gray value at the location of the steepest gradient G, optional is corrected upward by an offset, for example, two gray scale levels, then becomes the threshold B for binarizing the mammogram 2 used.

In step 105 becomes the mammogram 2 scaled by a factor of 1/5 and thus reduced in its resolution. The binarization of the mammogram 2 takes place in step 106 In the context of binarization, all pixels with a gray value above the threshold value receive a pixel value of 1 and all pixels below the threshold value a pixel value of 0. From the binarized mammogram 2 now results in a contour line that the object area 20 of the mammogram, which describes the mamma, but can be severely frayed and irregular. In step 107 Therefore, the contour line is morphologically smoothed by means of a closing and / or an opening operation to in step 108 trace the contour line and select the largest of the resulting contour lines. Subsequently, in step 109 the particular contour line is scaled back to the original size, in step 110 the contour smoothed and in step 111 reworked.

The method described is particularly suitable for mammograms 2 with a very noisy background, with the noise at the bottom of the histogram is filtered out by the threshold B is placed immediately above the noise representing mountain in the histogram. The resulting contour line must, as it is usually shredded and uneven, be smoothed and reworked.

In 4 schematically the course of the second threshold value method is shown, which is used in particular for the detection of the contour line in low-noise mammograms, such as direct digital mammograms or CR mammograms.

This will be in step 201 first the mammogram 2 scaled to 1/4 of the edge length and thus the resolution of the mammogram 2 reduced. Subsequently, in step 202 a list of thresholds for determining different contour lines of the mammogram 2 determined. This is from the mammogram 2 a histogram calculated in 5a is shown. Again, the lowest gray value is shown on the far left and the highest gray value on the far right. From this histogram, a second histogram is displayed in 5b Derived, for example, 256 shades of gray and the low-resolution histogram 32 May contain grayscale. To subdivide the histogram in mountains and valleys a frequency threshold S is set, for example, by S = (2 × number of pixels) / (5 × number of gray levels) may be given, wherein all gray scale areas with a histogram value below this threshold value as valleys and all gray scale areas with a histogram value above the threshold value are referred to as mountains. The histogram according to 5b is now, starting from the highest gray value (far right in 5b ), from above, until the first mountain is reached with the local maximum Max1 '. Starting from the first mountain range, the following valley is searched for and traversed to the last gray value Min1 'before the beginning of the next mountain. The gray value Min1 'immediately before the following mountain with the local maximum Max2' is then converted into the gray value Min1 in the high-resolution histogram, as in 5c and corrected for the exact transition of the valley to the next mountain. From the gray value Min1, a threshold value B1 is then derived by dividing the gray value Min1 by 1/8 of the valley width, as in 5c shown, is shifted upwards.

After this threshold B1 is determined, to find the next threshold, the next Go through mountains with the local maximum Max2, the next valley searched and determined another threshold. These steps will be repeatedly until the entire histogram H has passed through and thus all sorts of things Threshold have been detected.

In step 203 becomes the thus determined list of thresholds for binarizing the mammogram 2 used and derived from the different binarized mammograms each have a contour line. From the thus determined contour lines is in step 204 the smoothest contour line is selected, wherein the smoothest contour line in this context means the contour line with the smallest form factor. Subsequently, in step 205 the contour is scaled back to its original size and in step 206 the contour is finished.

Possibly can also in this case before the determination of the contour line a Edge treatment performed become.

The above-described threshold value method is used in particular for direct digital mammograms and CR mammograms, in which the object to be detected has gray values in the upper brightness range. In such mammograms, therefore, the substantial portion of the signal is concentrated in a relatively narrow range of gray levels, with the gray level falling away from the clearly visible, smooth breast edge relatively quickly towards the background area. Since the average brightness of the background area is unpredictable in this case, determination of the contour line on the basis of a fixed, global threshold value, corresponding to the first threshold value method, is not expedient in this case. Therefore, here are all possible thresholds by evaluating the histogram of the mammogram 2 determines and derived from these thresholds different contour lines, from which then the smoothest contour line is selected.

In 6a - 6c is shown schematically the course of the scan line method for detecting the contour line. The scan line method is used in particular when the mammograms 2 a pronounced gray scale gradient in the background of the mammogram 2 exhibit. First, this is the mammogram 2 in step 301 subjected to a border treatment, in the context of which light stripes and markers in the marginal area of the mammogram 2 be cut off.

Subsequently, in step 302 the image center of gravity M of the mammogram forming the support point 2 calculated to, as in 7a pictured, in step 303 radial scan lines R aufzuspannen, starting from the focal point M over the entire mammogram 2 erstre CKEN. In step 304 becomes the mammogram 2 now using a mean value filter (7 × 7) low-pass filtered and thus smoothed. In step 305 a threshold value for determining a start point Kp 'for the scan line method is determined, which is used to determine the contour points Kp of the scan lines R in step 306 is used. The steps necessary for determining the threshold value and the contour points Kp are explained in detail below. If the contour points Kp are determined, then the contour points Kp of all scan lines in step 307 a polygon is formed in step 308 is transformed into a closed contour line in step 309 smoothed and in step 310 is reworked.

For determining the contour points Kp in the scan line method according to 6a For each scan line R, the location of the steepest gradient in the edge region of the object region is searched. Since the mammograms to be processed by the scan line method 2 be noisy and with large gradient tainted structures also inside the object area 20 In this case, the scan lines R are not completed completely, but instead only pass outwards starting from a starting point Kp ', the starting point Kp' in each case within the object region, ie the mamma, must be arranged and by means of the in step 305 determined threshold is determined. By using the starting points Kp ', the scan line method is robust and reliable, in that it searches in particular only in the region for the steepest gradient, in which potentially the contour line of the object region 20 located.

For determining the threshold value for determining the starting points Kp 'in step 305 will be there, as in 6b shown schematically, first a noise detection filter on the mammogram 2 applied (step 321 ). Subsequently, the mammogram filtered in this way becomes 2 binarized (step 322 ), on the binarized mammogram 2 applied a morphology filter (step 323 See, for example, a double dilation operation and a triple erosion operation to standardize and smooth the binarized mammogram 2 can be performed) and thus a in 7b illustrated mask 22 generated by means of which the original mammogram 2 is masked (step 324 ). When masking the original mammogram 2 by means of the mask 22 then only the gray values remain in the in 7b dark areas are left, from which then in step 325 the mean is formed and in step 326 the threshold value is determined by adding an offset to the mean value. The thus determined threshold value thus represents the mean value of the gray values in the edge region of the breast corrected upwards by an offset and defines the starting points Kp 'which correspond to the points along the scan lines R at which the gray values of the scan lines exceed the threshold value For the first time, the focus falls below M.

By means of the method according to 6b Thus, starting points Kp 'are determined, from which the scan lines R are passed through to determine the position of the steepest gradient and thus the contour points Kp. By using the starting points Kp 'it is guaranteed that not inside the object area 20 is searched for the steepest gradient, but only in the edge region of the object area 20 , In the determination of the starting points Kp 'is exploited here that the mammogram to be examined 2 just in the edge area of the object area 20 is noisy, so that the edge area by means of a noise detection filter (see step 321 ) and thus the area in which the contour line is arranged, can be filtered out to search specifically in this area for the contour points Kp.

The determination of the steepest gradient in detail is schematically in 6c shown. To determine the contour point Kp of a scan line R, first the gray values of the mammogram are determined 2 read out along the scan line (step 341 ). Then, the position at which the gray values are determined in step 305 below the detected threshold and represents the starting point Kp ', (step 342 ) and then from this starting point Kp 'the position of the strongest sloping gradient sought (step 343 ) indicating the contour point Kp on the scan line R.

The different methods described above for detecting the contour line in different mammograms, such as in 2 indicated with a method for edge treatment of the mammogram 2 be combined before determining the actual contour line. Such edge treatment is required when in peripheral areas of the mammogram 2 bright stripes or objects in the form of markers or the like are arranged, which can be disturbing in the detection of the contour line and can lead to erroneous results.

Such a method for edge treatment is schematically shown in FIG 8th represented and in 9 illustrated. This in 9 illustrated mammogram 2 contains an object 23 in the form of a check box that can display, for example, patient data and admission information. The object 23 is, as well as the object area 20 , coded by gray values, can show large gradients at its edges and thus the analysis of the mammogram 2 to disturb. As part of the Randbe The action thus becomes peripheral areas of the mammogram 2 with disturbing objects 23 away.

As part of the edge treatment, the objects must 23 detected and then from the mammogram 2 be eliminated. First, according to the in 8th illustrated method in step 401 Scan lines P in the mammogram 2 arranged perpendicular to an outer edge of the mammogram to be examined 2 run. For example, 40 scan lines per outer edge can be used to create objects in the border area of the mammogram 2 to capture with a sufficient resolution. In step 402 For each scan line, the location Q of the maximum gradient is determined. The locations Q of the maximum gradients on the individual scan lines R are then determined in step 403 Equalizing lines A1, A2, A3 determines, to determine the equalization lines A1, A2, A3 respectively for the locations Q of eight consecutive scan lines P a line is determined which best approximates the locations Q of the eight observed scan lines P. In step 404 and 405 the thus determined regression lines A1, A2, A3 are discarded if they are not sufficiently parallel to the outer edge of the mammogram 2 run (step 404 ) or the locations Q belonging to the compensation straight lines are not close enough to the compensation straight lines A1, A2, A3. In step 404 Thus, it is ensured that only equalization lines are used, which run substantially parallel to the edge of the image. In step 405 In addition, it is ensured that only those compensation lines are used whose associated locations Q do not scatter too far around the compensation straight line. For the steps 404 . 405 In this case, a suitable error measure can be defined, which defines a Tolleranzschwelle, which are exceeded when the Tolleranzschwelle the equalization lines are discarded.

Applied to the equalizer lines A1, A2, A3 in 9 This means in more detail that the compensation straight lines A1, A3 are maintained, since they both run parallel to the edge of the image and precisely map the locations Q belonging to the compensation straight lines A1, A3. The regression line A3, however, which is neither parallel to the outer edge of the mammogram 2 The places Q, which are supposed to approximate, are sufficiently mapped, are discarded and are not considered further.

In step 406 Then, the distances of the locations Q to the outer edge in the scan line 1 and the scan line 8 of the respective compensation straight line A1, A3 are calculated for all non-discordant compensation lines A1, A3, in order to select the maximum of the two distances. Of all remaining equalizer lines A1, A3 becomes the one with the greatest distance to the outer edge of the mammogram 2 selected and used to define the boundary line (step 407 ), with the boundary line parallel to the outer edge of the mammogram 2 is defined by the maximum distance of the selected balance line. The mammogram 2 is then truncated along the boundary so that all objects 23 in the border area of the mammogram 2 Be removed beyond the boundary.

For the in 9 case shown that means the mammogram 2 along the balancing line A3 and its vertical extension is cut vertically upwards and thus the edge area beyond the boundary line defined by the straight line A3 and its vertical extension removed and in the analysis of the mammogram 2 not considered further.

Furthermore, the device 1 , as in 1 specified, additionally a means for performing a contour post-processing 16 exhibit. The contour postprocessing serves the purpose of the threshold value method according to 2 or 4 or the scan line method according to 6a - 6c to trace the contour line determined so that it exactly matches the object area of a mammogram to be detected 2 indicates.

In 10 is a mammogram 2 shown with a contour line K, which is composed of contour segments Y1, Y2, Y3, Y4 and edge segments X1, X2, X3 and the mamma to be imaged in the object area 20 partly incomplete.

In particular, in the mammogram 2 according to 10 The mamma in the lower part is partially cut off (along the marginal segment X3) and the contour segment Y4 continues to be faulty when in the mammogram 2 segmented breast shown and only the mamma bordering contour line is to be determined, which should be neglected by the contour segment Y4 skin fold, however. By means of the method for contour processing, it is then possible to adapt the contour line K such that the object area is displayed in the desired manner.

The process for contour post-treatment is schematically shown in FIG 11 shown. First, this is in step 501 the contour line K is broken down into segments, with a distinction being made between edge segments X1, X2, X3, which run along the outer edge of the mammogram 2 run (edges, corners), and contour segments Y1, Y2, Y3, Y4, which do not touch the outer edges or only at one point. In step 502 consecutive edge segments X1, X2 are combined to form a combined edge segment. Subsequently, in step 503 from all combined edge segments the longest Edge segment X1, X2 selected, and in step 504 All remaining edge segments X3 and the contour segments Y1, Y2, Y3, Y4 are combined into one segment. This results in chest sand, which is the contour of the mamma in the mammogram 2 displays. In step 505 Then the breast is smoothed and in step 506 customized. Subsequently, in step 507 the contour line K is composed of the breast edge and the edge segments X1, X2, so that a complete, the object area 20 with the mammogram 2 Mum enclosing contour line K results.

To adjust the chest sand in step 506 On the one hand, the contour line in the area between the contour segments Y2 and Y3 is supplemented so that a realistic contour line K results. Furthermore, it may be expedient to remove the contour segment Y4 from the contour line, as this does not indicate the chest wall, but a skin fold, which is not directly to the mammogram 2 Mum heard.

The basic procedure for adjusting the chest sand in step 506 according to 11 is based on 12a and 12b illustrated. In order to supplement and adjust the breast edge between the contour segment Y2 and Y3, first the point of the breast edge is selected, that in the mammogram 2 farthest to the right, corresponding to point U in 12a , Subsequently, an auxiliary point Z is determined whose coordinates in the horizontal and vertical directions correspond to xZ = xU / 3, yZ = yU, where xZ and yZ are the coordinates of the auxiliary point Z and xU and yU are the coordinates of the point U in the horizontal and vertical directions in the mammogram 2 specify. Proceeding from the auxiliary point Z, a starting point C 'and an end point Z' on the breast edge are searched for by starting at the starting point C 'a line at an angle of 45 ° to the distance spanned by the points Z, U towards the breast and for the end point C 'the solder is beaten in the direction of the breast edge, the point of intersection with the breast edge, the points C', Z 'results (see 12a ). Now another auxiliary point C is determined by looking for the point C arranged vertically above the point C 'on the path formed by the points Z, U. Subsequently, all contour points Kp between the starting point C 'and the end point Z' are determined, and a curve is adapted to these contour points, for example using a curve of the following form: R = A (φ - φ0) 2 + B, wherein in the context of the adaptation, the parameters A, B and φ0 are determined and R and φ represent the coordinates of the contour points Kp in polar coordinate notation about the point C. In order to correct the course of the contour line K, now the curve, as in 12b shown, extrapolated, so that the curve is continued beyond the end point Z 'to the left. Contour points that lie outside the curve thus determined are then discarded, and the extrapolated curve becomes the new contour segment Y4 '.

By means of the method for contour finishing both areas of the breast edge can be shot as well as contour segments that are not part of the object area to be detected are excluded, so that the result is a contour line is made available, that in the mammogram 2 the best possible picture of the object contained.

The methods described above run completely automatically and determine from an input mammogram 2 a contour line, depending on the type of mammogram 2 Different methods are used and in this way different, differing in terms of their image characteristics mammograms 2 can be processed. Regardless of the type of mammogram 2 For example, a user is provided with a contour line or a masking image determined by the contour line as output.

After the object region contained in a mammogram is detected and segmented, ie the contour line enclosing the object region has been determined, the mammogram can be further processed. On the one hand, the detection of the contour line now enables the automatic alignment, scaling and positioning of the mamma contained in the mammogram as well as in the mammogram 13a and 13b shown. Secondly, based on the detection of the contour line, an interactive input of findings directly into the mammogram by a physician is possible, as in 14a - 14c shown.

In mammography examinations, mammograms of the left and right breasts of a patient are recorded and diagnosed by a physician. Frequently, four mammograms are created within the scope of such a mammographic examination, whereby one craniocaudal (CC) and one mediolateraloblique (MLO) image are made of each patient's breast. The images obtained in this way are compared with each other in pairs, whereby the craniocaudal images of the right and left breast as well as the mediolateral images of the right and left breasts are subjected to a symmetry comparison. In the diagnosis, this symmetry comparison between left and right breasts of a patient plays an essential role, since architectural disturbances in the form of Asymmetries between left and right breast can be an indication of a tumor possibly present in a mamma.

at such mammography examinations turns out for one Doctor the problem that the pictured Mammas usually only a portion of the Take mammograms while the majority of the mammogram through the background area containing no information filled out is. About that In addition, the mammograms are usually shifted from each other and must thus, to be able to compare the mammograms meaningfully with each other, relatively be positioned to each other.

Exemplary mammograms 2a . 2 B . 2c . 2d such a mammography examination are in 13a - 13c in which the craniocaudal (CC) images of the right (R) and left (L) breasts as well as medio-lateral (MLO) images of the right (R) and left (L) breasts are clearly shifted vertically relative to each other. At the same time they fill the object area 20 engaging mamma each only a small part of the mammogram 2a . 2 B . 2c . 2d off while much of the mammogram 2a . 2 B . 2c . 2d through a background area 21 is taken.

To the mammograms 2a . 2 B . 2c . 2d In order to optimally position and scale such a mammogram, a contour line K1, K2, K3, K4 is first determined in each case, which corresponds to that in the object area 20 illustrated mom from the background area 21 of the mammogram separates.

After the contour line for each mammogram 2a . 2 B . 2c . 2d is determined, is for each mammogram 2a . 2 B . 2c . 2d , which determines center of gravity M1, M2, M3, M4 and respectively determines a height line H1, H2, H3, H4, which points through the center of gravity M1, M2, M3, M4. Based on the contour lines H1, H2, H3, H4 are the mammograms 2a . 2 B . 2c . 2d then aligned for the craniocaudal (CC) and mediolateralobliquen (MLO) images in pairs vertically aligned.

Subsequently, using the contour lines K1, K2, K3, K4 for each mammogram 2a . 2 B . 2c . 2d a surrounding rectangle R1, R2, R3, R4 determined, wherein in the determination of the surrounding rectangle R1, R2, R3, R4, the vertical alignment is taken into account on the basis of the contour lines H1, H2, H3, H4. Of the surrounding rectangles R1, R2, R3, R4, the largest rectangle is now selected and based on this rectangle - in the in 13b the R3 rectangle - an optimal scaling factor determined by which all mammograms 2a . 2 B . 2c . 2d uniformly scaled. The scaling factor is determined so that the mamma surrounded by the largest rectangle R3 the mammogram 2a . 2 B . 2c . 2d optimally filled with minimization of the background area and thus enlarged displayed for diagnosis.

A uniform scaling factor is used in all directions, so that the aspect ratio of the mammogram is not changed in each case. At the in 13b In this case, the scaling factor is chosen to be the one in the mammogram 2c shown mamma is maximally increased in the horizontal direction and thus extends over the entire display area, corresponding to the area S3.

The mammograms 2a . 2 B . 2c . 2d are then exclusively in the areas S1, S2, S3, S4 according to 13b output, so that the optimally scaled and positioned mammograms 2a ' . 2 B' . 2c ' . 2d ' according to 13c result.

Significant advantage of automatic positioning and scaling of mammograms 2a . 2 B . 2c . 2d In a mammographic exam, that is a manual fitting of the mammograms 2a . 2 B . 2c . 2d by a doctor is no longer necessary, so for the doctor to diagnose and find the mammograms 2a . 2 B . 2c . 2d thus considerably easier.

Furthermore, the apparatus and method for analyzing a mammogram may also be used to assist a physician in presenting a report. Conventionally, diagnostic devices known from the prior art, for example in the form of a workstation, have an image viewer, in particular a Picture Archiving and Communication System (PACS) viewer and a RIS (Radiological Information System) client. In the case of prior art diagnostic devices, the physician regards the mammogram 2 in the image viewer and must then switch to the findings input window, there to enter his findings, the findings input by manual text input, dictation with recording or speech recognition can be made, but in principle takes place in the separate input window.

In particular, in regular screening tests, the problem arises that a doctor must find a large number of mammograms in a short time. A disadvantage of conventional diagnostic devices is that the doctor is distracted from the image when entering findings, since the findings input is not directly in the mammogram 2 takes effect, thereby the time effek tive time, the doctor on the visual observation and diagnosis of the mammogram 2 and that the change between the finding input and the consideration leads to an increased susceptibility to diagnosis due to increased concentration requirements. Furthermore, transmission errors may occur, which is due to the fact that the doctor as a key information in the findings input the location of the findings in the mammogram 2 must document.

Mammograms by means of a computer-assisted mammography analysis 2 be automatically analyzed and examined, the findings input by a physician can be significantly supported and simplified. For this purpose, the device has a findings input device 18 on the in 14a - 14c illustrated manner processed by a physician entered findings.

In the examination of one or more mammograms created during the examination 2 gives a doctor how in 14a shown a finding E directly in the mammogram 2 whereby a selection box can be provided with predefined findings inputs (for example the possible entries "microcalcification", "condensing stove", "asymmetry" or the like) 2 will now be the contour line of the mammogram 2 Mum detected. By comparing the contour line with a report mask EM, then, as in 14b and 14c represented, the coordinates of the input findings E transferred to the findings mask EM, wherein the location of the finding E in the findings mask EM determined by the comparison of the mamma described by the contour line with the standardized findings mask EM, in this way in the findings mask EM transmitted and is displayed in the findings mask EM. For example, the standardized findings mask EM can specify the standard contour of an average breast and serves to visualize and visualize the findings.

In this context, a finding E in a mediolateralobliquen mammogram 2 automatically in a mediolateraloblique findings mask EM (see upper half in 14b ), while finding E in a craniocaudal mammogram 2 is accordingly displayed in a craniocaudal findings mask EM.

The entered finding is also automatically documented in a so-called time model, in which, similar to aviation, the location of the findings in the findings mask EM is indicated by a time, the time indicating a direction from a central point of the mom, in the entered finding E is in the mamma. The coordinates of the finding E are here converted into the time model and, as in 14c displayed, displayed. At the in 14c In the example shown, the finding E entered by the physician is thus automatically converted into the display "Microcalcification at 5:30 o'clock" and subsequently stored as a structured textual finding.

Of the The idea underlying the invention is not limited to the embodiments described above. In particular, can the described method can also be applied analogously to the Automatically detect density of breast parenchyma in a mammogram and to document. This is a segmentation of the chest area carried out, to determine the dense chest areas and, by characteristics like contrast and area the chest areas, according to predetermined (for example, by the American College of Radiology) to classify criteria and to take over in a structured findings input.

Furthermore The described method can also be applied to the quality of a Automatically capture mammogram and its positioning and to document. Also for this is a segmentation of the contour line carried out, for example, mamille, muscle and angle of the lower skin fold to investigate. On the basis of predetermined (for example by the "European Guidelines defined for mammography screening ") Criteria will then be a quality assessment of these parameters Mammogram calculated and taken into the structured findings input.

1

contraption

2, 2a, 2b, 2c, 2d

mammogram

2a ', 2b', 2c ', 2d'

Optimal positioned and scaled mammogram

10

input unit

11

processing unit

12

medium to carry out of the thresholding method

13

medium to carry out of the scan line method

14

output unit

15

medium to carry out the edge treatment

16

medium to carry out the contour postprocessing

17

medium for scaling and positioning the mammogram

18

Evaluation input device

20

Property area

21

Background area

22

mask

23

object

101-111

steps

201-206

steps

301-310

steps

321-326

steps

341-343

steps

401-407

steps

501-507

steps

A1, A2, A3

fit line

B B1

threshold

e

findings

EM

findings mask

G

maximum gradient

H1, H2, H3, H4

contour

I

index the gray values

K

contour line

kp

contour point

kp '

starting point

M, M1, M2, M3, M4

Image focus

Max, Max1, Max2

local maximum

Max1 ', Max2'

local maximum

min1, min2

gray value

Min1 ', Min2'

gray value

P

scan line

Q

place of the maximum gradient

R

scan line

S

frequency threshold

S1, S2, S3, S4

Area to output the optimally positioned and scaled mammograms

X1, X2, X3

edge segment

Y1, Y2, Y3, Y4

contour segment

Claims (26)

Computer-aided analysis device with regard to its image characteristics of distinctive mammograms ( 2 . 2a . 2 B . 2c . 2d ), the device ( 1 ) - means for carrying out a threshold value method ( 12 ), which has at least one threshold (B, B1) for binarizing the mammogram ( 2 . 2a . 2 B . 2c . 2d ), and - means for performing a scan line method ( 13 ), which determines the gradient of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) along different scan lines (R), wherein the device ( 1 ) is formed and provided, depending on the image characteristics of the mammogram to be processed ( 2 . 2a . 2 B . 2c . 2d ) by means of the threshold value method and / or the scan line method, an object area ( 20 ) encompassing contour line (K) of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) to investigate. Device according to claim 1 or 2, characterized in that the device ( 1 ) means for performing edge treatment ( 15 ) for determining lateral boundary lines of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) having. Device according to one of the preceding claims, characterized in that the device ( 1 ) means for performing a contour postprocessing ( 16 ) for smoothing and correcting the determined contour line (K). Device according to one of the preceding claims, characterized in that the device ( 1 ) a means for scaling and positioning ( 17 ) of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) having. Device according to one of the preceding claims, characterized in that the device ( 1 ) a findings input device ( 18 ) for the interactive input of a finding (E) into the mammogram ( 2 . 2a . 2 B . 2c . 2d ) having. Method for computer-aided analysis relating to the image characteristics of distinctive mammograms, for use in a device ( 1 ) according to one of claims 1 to 5, wherein, depending on the image characteristic of the mammogram to be processed ( 2 . 2a . 2 B . 2c . 2d ) - a threshold value method in which at least one threshold value (B, B1) for binarizing the mammogram ( 2 . 2a . 2 B . 2c . 2d ) and / or - a scan line method in which the gradient of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) is determined along different scan lines (R), for the detection of an object area ( 20 ) Contouring contour line (K) of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) is used. Method according to claim 6, characterized in that on the basis of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) automatically selects whether the threshold method and / or the scan line method is used to detect the contour line. Method according to claim 6 or 7, characterized in that the object area ( 20 ) in the form of a masking image, which is determined based on the contour line (K), and / or the contour line (K) is displayed and output. A method according to claim 6 to 8, characterized in that prior to the detection of the contour line (K) an edge treatment for determining the lateral boundary lines of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) is carried out. A method according to claim 9, characterized in that in the edge treatment for determining a lateral boundary line of the mammogram ( 2 . 2a . 2 B . 2c . 2d a) along perpendicular to an outer edge of the Mam mogram ( 2 . 2a . 2 B . 2c . 2d ) directed scan lines (P) the gray values of the mammogram ( 2 . 2a . 2 B . 2c . 2d ), b) in each case the position of the steepest gradient of the gray values along the scan lines (P) is determined and c) from the positions of the steepest gradients the lateral boundary line of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) is determined. A method according to claim 10, characterized in that on the basis of the lateral boundary lines outside of the boundary lines arranged regions of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) are removed. Method according to one of claims 6 to 11, characterized that after the detection of the contour line (K) a contour post processing for smoothing and correction of the contour line (K) is performed. Method according to one of claims 12, characterized that during the contour processing a) the detected contour line (K) is broken down into segments (Y1, Y2, Y3, Y4, X1, X2, X3), b) smoothed the segments become, c) missing segments added and d) faulty Segments are adjusted. Method according to one of claims 6 to 13, characterized in that in the threshold value method a) a histogram of the mammogram ( 2 . 2a . 2 B . 2c . 2d b) from the histogram at least one threshold value (B, B1) for binarizing the mammogram ( 2 . 2a . 2 B . 2c . 2d ), c) the mammogram ( 2 . 2a . 2 B . 2c . 2d ) is binarized on the basis of the at least one threshold value (B, B1) and d) on the binarized mammogram ( 2 . 2a . 2 B . 2c . 2d ) a contour line (K) is determined. A method according to claim 14, characterized in that for determining the at least one threshold value (B) a) the histogram is ascended starting from the lowest gray value, b) a local maximum (Max) of the histogram is determined, c) the steepest gradient ( G) the falling edge of the histogram is determined after the local maximum, d) a threshold value (B) is determined on the steepest gradient (G) and e) the threshold value (B) for binarizing the mammogram ( 2 . 2a . 2 B . 2c . 2d ) is used. Method according to claim 14, characterized in that in that for determining the at least one threshold value (B1) a) go through the histogram descending from the highest gray value becomes, b) determines a local maximum (Max1) of the histogram becomes, c) a valley following the local maximum (Max1) Histogram is determined d) a threshold based on the valley (B1) is determined and e) repeated steps b) to d) until the histogram has passed through. A method according to claim 16, characterized in that different threshold values (B1) are determined, each for binarization of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) and used to determine different contour lines. Method according to claim 17, characterized in that that from the different contour lines the smoothest contour line (K) selected becomes. Method according to one of claims 6 to 18, characterized in that in the scan line method a) starting from a Aufspannpunkt (M) of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) radial scan lines (R) are spanned, b) the mammogram ( 2 . 2a . 2 B . 2c . 2d ) is read out along the scan lines (R), c) based on the gradient of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) is determined along the scan lines (R) for each scan line (R) in each case a contour point (Kp) and d) on the basis of the contour points (Kp) of the individual scan lines (R), the contour line (K) is determined. A method according to claim 19, characterized in that the Aufspannpunkt the image focus (M) of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) corresponds. Method according to one of claims 6 to 20, characterized in that the mammogram ( 2 . 2a . 2 B . 2c . 2d ) for detecting the contour line (K) by means of a locally on a group of pixels of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) acting low-pass filter is filtered. Method according to one of claims 6 to 21, characterized in that based on the determined contour line (K) the mammogram ( 2 . 2a . 2 B . 2c . 2d ) is positioned and scaled. A method according to claim 22, characterized in that for the positioning and scaling of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) one the vertical position of the object area ( 20 ) of the mammogram ( 2 . 2a . 2 B . 2c . 2d ) indicative contour line (H1, H2, H3, H4) and the object area ( 20 ) enclosing rectangle (R1, R2, R3, R4), by means of which the mammogram ( 2 . 2a . 2 B . 2c . 2d ) is scaled and positioned. Method according to one of claims 6 to 23, characterized in that based on the determined contour line (K) an interactive input of a finding (E) in the mammogram ( 2 . 2a . 2 B . 2c . 2d ) is processed, displayed and stored. Method according to Claim 24, characterized that on the basis of the determined contour line (K) the coordinates of the Findings (E) determined and transferred to a findings mask (EM) become. Method according to claim 25, characterized in that that the coordinates of the finding (E) transformed into a time model and output using the clock model.