WO2000005677A1 - Systeme de detection automatise de masses cancereuses dans les cliches mammaires - Google Patents

Systeme de detection automatise de masses cancereuses dans les cliches mammaires Download PDF

Info

Publication number
WO2000005677A1
WO2000005677A1 PCT/US1998/015350 US9815350W WO0005677A1 WO 2000005677 A1 WO2000005677 A1 WO 2000005677A1 US 9815350 W US9815350 W US 9815350W WO 0005677 A1 WO0005677 A1 WO 0005677A1
Authority
WO
WIPO (PCT)
Prior art keywords
roi
context data
mammogram
neural net
data comprises
Prior art date
Application number
PCT/US1998/015350
Other languages
English (en)
Inventor
Gary Lee Shapiro
Davis Regoalt Oliver, Jr.
Original Assignee
Lockheed Martin Corporation
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Lockheed Martin Corporation filed Critical Lockheed Martin Corporation
Priority to AU85859/98A priority Critical patent/AU8585998A/en
Priority to PCT/US1998/015350 priority patent/WO2000005677A1/fr
Publication of WO2000005677A1 publication Critical patent/WO2000005677A1/fr

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]

Definitions

  • the present invention relates generally to the field of computer-aided diagnosis of medical images. More specifically, the present invention discloses an automated system for detecting cancerous masses in digital mammograms.
  • the prior art includes a variety of systems to detect both microcalcifications and mass lesions.
  • the prior art techniques include either the use of features extracted by human observers or computer-extracted features.
  • the latter features for microcalcifications include shape analysis such as compactness, Fourier descriptors of image boundaries, and average distance between calcifications applied to the extracted features.
  • features used include spiculations or irregular masses that are identified by local radiating structures or analysis of gradient histograms generated by seed growing and local thresholding methods, morphologically-based features and image texture. These features are classified using either neural networks or binary decision trees. Common to all of these approaches is limited testing, with the computationally- intensive nature of the process implied as one reason for less than full, comprehensive testing.
  • Kegelmeyer et al. have reported promising results when applying algorithms using binary decision trees and a dense feature map approach in identifying spiculated lesions and calcifications. Kegelmeyer et al. have achieved 100% sensitivity and 82% specificity when merging edge information identifying spicules with local texture measures, thus eliminating false-positive detections.
  • the present approach is based on an innovative concept to detect patterns in medical images (e.g., mammograms) using a Fourier transform optical correlator for image analysis followed by a series of neural networks hosted on a digital computer to analyze the results.
  • An optical processor allows higher- order bandwidth, multi-resolution, multi-orientation approaches for feature extraction and enhancement that are not feasible in real time on a digital computer.
  • a hybrid optical/digital computer approach ensures sufficient processing power at a moderate cost to accommodate discriminating algorithms and yet analyze a mammogram in a matter of seconds.
  • This invention provides a system for automated detection of cancerous masses in mammograms.
  • the mammogram is digitized and regions of interest (ROIs) are detected using Fourier analysis (e.g., by means of an optical correlator).
  • ROIs regions of interest
  • the pixels in the ROI are averaged together to create a smaller array of super-pixels that are input into a first neural net.
  • Context data is extracted from the mammogram for each ROI, such as size, location, ranking, brightness, density, and relative isolation from other ROIs.
  • a second neural net receives the output values from the first neural net and the context data as inputs and generates an output score indicating whether the ROI contains a cancerous mass.
  • the second neural net can also be provided with context data from another view of the same breast, the same view of the other breast, or a previous mammogram for the same patient.
  • FIG. 1 is a simplified block diagram of the present invention.
  • FIG. 2 is a flow chart of the present invention.
  • FIG. 3 is a section of a mammogram showing a region of interest that is potentially cancerous.
  • FIG. 4 is a section of a mammogram corresponding to FIG. 3 that has been overlaid with a super-pixel grid.
  • FIG. 5 is a section of a mammogram corresponding to FIG. 3 after the region of interest has been reduced to super-pixels.
  • FIG. 6 is a section of a mammogram corresponding to FIG. 3 after the region of interest has been enhanced.
  • FIG. 7 is an example of a report consisting of two views in a mammogram with the suspicious regions of interest marked.
  • FIG. 1 provides a simplified block diagram of the overall system. A corresponding flow chart of the process used in the present invention is illustrated in FIG.
  • FIG. 2 is acquisition of a digital mammogram 20 for analysis.
  • a wide variety of hardware 11 can be used for image acquisition.
  • a high-resolution scanner can be used to scan a conventional mammogram film, or a direct digital image can be acquired.
  • Mammograms are typically paired by views (mediolateral and craniocaudal).
  • raw digitized images 20 are received by the workstation. These images are raw, single byte images in the 35-70 micron resolution range. Either a fixed size is used for the image or information on the pixel width and height of the image is included. A header is added to the image that allows the system to know the size and data type of the image.
  • a database is used to track the image, its size, and orientation (left breast, right breast, craniocaudal, or mediolateral). The quality of imagery is also examined. This is primarily a check that the image is a mammogram (i.e., it has the features of a mammogram) and that it meets specifications indicative of a mammogram ready to be read by a radiologist.
  • Image processing hardware 12 reduces the scale of the image to a resolution of approximately 230 microns. This reduces the processing requirement while maintaining sufficient resolution to identify regions of interest (ROIs).
  • a blurring/contrast reduction algorithm is used where the image is superimposed on the background, to reduce "edge effects" (i.e., erroneous high frequency signals created by a sudden drop-off or edge).
  • Regions of interest are identified in the image by
  • bandpass filters are used to detect feature sizes. These bandpass filters are as follows:
  • Peaks that are too small or too large relative to the scale of the bandpass are immediately rejected. A more detailed size determination is made on the remaining peaks.
  • a radial size is calculated by measuring the distance from the peak in several directions until the pixel brightness is:
  • FIG. 3 is an example of a section of a mammogram showing a region of interest that is potentially cancerous. These ROIs are extracted from the original high-resolution image in squares approximately 2-3 times feature size as measured by the first criteria and scaled to fixed size (256 x 256 pixels). There are eight possible ROI resolutions, ranging from a minimum size of 9.7 mm (38 micron resolution) to a maximum size of 7.8 cm (305 micron resolution).
  • Each region of interest is analyzed by means of a neural network using a super-pixeled image as an input.
  • This grid is a radial-polar grid with the angles evenly spaced and constant radial increments.
  • the grid consists of 10 equally-spaced radial and 32 equally-spaced angular bands.
  • the pixels in each grid space are averaged together to create 320 "super-pixels" (reference numeral 24 in FIG. 2).
  • the super-pixels, along with two inputs determined from the approximate size of the feature, are used as the inputs to the first neural net 14.
  • FIG. 4 is a section of a mammogram corresponding to FIG. 3 that has been overlaid with a super-pixel grid.
  • FIG. 5 is a section of a mammogram corresponding to FIG. 3 after the region of interest has been reduced to super-pixels.
  • the first neural network 14 is trained on both cancerous and non-cancerous regions of interest.
  • Regions of Interest are selected from known (verified normal or verified cancerous) mammograms. These ROIs are ranked according to their appearance, with +1 being most normal-like and -1 being most lesion-like.
  • the ROIs are then converted to super-pixels for input to the first neural network. A random scaling, rotation and translation of the super-pixel grid is performed, and half of the image (angularly) is chosen in a way that excludes any parts of the ROI outside the original area of the mammogram (These edges are generated by superimposing an odd-sized image on a fixed-sized background).
  • the values of the half-grid are normalized in two steps: A stretching of the values over a fixed range increases the contrast within the half-grid, while a normalization of the sum of all the half-grid inputs assures consistency in brightness over all possible half-grids. A statistical thresholding of the ROI is used to determine the approximate "size" of the main feature in the ROI. This size measurement is converted to an equivalent diameter, perturbed by the same scaling perturbation used on the super-pixel grid, then converted into a pair of inputs, using a sine-cosine transformation. These two values, along with the grid super-pixels, are input to the first neural network using a randomly initialized set of neural network weights.
  • the output of the first neural network (restricted between +1 and -1) is compared with the chosen ranking.
  • An error value is calculated and used to correct the neural network weights using a back-propagation learning algorithm known as REM (Recursive Error Minimization).
  • the REM algorithm uses the derivatives of the propagated error to create a more stable convergence than conventional back-propagation.
  • a new set of perturbations is then applied to the next ROI, creating a new half-grid, which is again normalized and applied to the network. This process is repeated until the RMS of the error values drops to a small value, or the change in the error over several iterations becomes very small, whichever occurs first. This signifies that the network is now trained.
  • the number of iterations depends on several conditions, but in general will take about 1000-3000 cycles through the complete set of ROIs.
  • a set of ROIs selected from "unknown" mammograms is processed through the neural network (reference numeral 25 in FIG. 2). No perturbations are done on these inputs, although the grid is rotated through all possible angles, one angular grid space at a time, such that no half-grid includes any "edges".
  • Each half-grid in turn is normalized as above, and applied to the trained network along with the two size inputs from the ROI.
  • the outputs for the rotations are combined, producing two or three statistical outputs for each ROI. These outputs are then used as part of the set of inputs for the second neural network 15, as discussed below.
  • FIG. 6 is a section of a mammogram corresponding to FIG. 3 after the ROI has been enhanced to emphasize the contours and lines of the features in the image.
  • these enhanced images are used, along with the original ROIs, to generate context inputs for the second neural net 15 (or "context network").
  • the context inputs for the context network 15 includes the location, ranking, brightness and density profiles, sizes of inner and outer features, relative isolation from other ROIs, roughness of contours, fuzziness, and the number and length of any spicules.
  • the distance of the center of each ROI from the nipple is determined by:
  • the context neural net 15 is initially trained, again using verified cancerous and non-cancerous cases with random deviations for each input to generate framing examples.
  • the context inputs for each ROI are determined, including the outputs from the trained first neural network 14.
  • Each input is perturbed by a random value (e.g., +/-10%) and then converted into a pair of inputs using a sine-cosine transformation.
  • the full set of inputs is applied to the second neural network 15 using a randomly initialized set of neural network weights.
  • the output of the second neural network 15 is compared to the same ranking value used for the first neural network 14.
  • the weights are corrected by the method cited above, and the second network 15 is trained to convergence using about 1000-3000 cycles through the set of ROIs.
  • the "unknown" ROIs are passed once through the trained second network 15 without perturbations.
  • the output of this network is then combined with that of other ROIs belonging to the same mammogram (including other views, other breast and/or other years) to characterize the ROI and the mammogram.
  • the trained context neural network 15 can then be used to evaluate unknown cases (reference numeral 26 in FIG. 2).
  • the context neural net 15 is a second network separate from the first neural network 14 used in step 3.
  • the context neural net 15 could be substituted.
  • a single neural net could receive both the ROI super-pixels and context data as inputs.
  • Threshold regions of interest After training is complete, an ROC curve is generated from the output score produced by the second neural network in response to sets of known mammograms. An appropriate threshold output score can be determined from the resulting ROC curve which will achieve the desired probability of detection and false alarm rate.
  • This threshold can then be used to classify unknown cases during normal operation of the system (reference numeral 27 in FIG. 2). For example, the threshold score is used by the decision software 16, as depicted in FIG. 1, to generate a report for the radiologist.
  • FIG. 7 is an example of a report consisting of two views in a mammogram with the suspicious regions of interest marked. This report typically shows both views of the mammogram, along with a classification determined by the analysis of the neural networks. The locations of suspect masses are marked on the display.
  • the final classifications might include the following: (a) "Suspect" mammogram has at least one ROI detected on either view with a value less than a predetermined threshold (i.e., more lesion-like), but more than a fixed level (e.g., 0.2) below the threshold.
  • a predetermined threshold i.e., more lesion-like
  • a fixed level e.g., 0.2
  • the present system can be used by a radiologist in either of two modes of operation.
  • the system merely aides the radiologist by flagging any ROIs falling into the "Suspect” or “Cancer” classifications.
  • the radiologist continues to thoroughly review each mammogram and remains responsible for diagnosis.
  • the system is equipped with an image archive 18 shown in FIG. 1.
  • the archive can store each mammogram or only those mammograms designated by the radiologist.
  • the second mode (“screening mode"), the system analyzes each mammogram and only those classified as "Suspect” or “Cancer” are brought to the attention of the radiologist for further review. "Normal" mammograms are sent directly to the image archive 18 for archival storage.
  • a number of additional types of context data can be input into the context neural net to enhance the accuracy of the system.
  • the ROIs can be analyzed in the context of information from the other view of same breast. This step can be combined with the previous step using the second neural network 15 as described above, or performed separately using a third neural network.
  • the additional context data can include information on the size and brightness of the ROI under question as well as other ROIs in the same breast and ROIs in the alternate view which have nearly the same perpendicular distance to the nipple. For training purposes, each value is allowed to vary over a specified range to gain additional test cases.
  • the second neural network 15 is then trained using both cancerous and non-cancerous ROIs. The resulting connections are then used to evaluate ROIs outside of the training set.
  • the ROIs can be analyzed in the context of the same view of other breast for the same patient.
  • This step is implemented by performing the same operations on the other breast, then cross-correlating the information to determine asymmetries between the two breasts.
  • additional context inputs can be input into the context neural net 15 corresponding to the sizes and density network results for ROIs in similar locations on the other breast.
  • the ROIs can also be analyzed in the context of prior year mammograms for the same patient. Additional context inputs can be input into the context neural net 15 corresponding to ROIs in similar locations in prior-year mammograms for the same patient. This is similar to the use of context inputs derived from contemporaneous alternate views, as described above, but instead relies on detecting changes that have occurred since previous mammograms. These additional context inputs should produce a significant reduction in the false alarm rate in light of the temporal development properties of cancers in contrast to non-cancers. Here again this implementation requires a additional sets of mammograms for adequate training of the context neural net 15.
  • the present invention could be employed to detect abnormalities in a wide variety of medical images.
  • the present system could be used to detect cancer or other types of lesions in a chest x-ray film; or cancerous and pre-cancerous cells in a pap smear or biopsy.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Medical Informatics (AREA)
  • Radiology & Medical Imaging (AREA)
  • Quality & Reliability (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Multimedia (AREA)
  • Apparatus For Radiation Diagnosis (AREA)

Abstract

L'invention concerne un système (10) de détection automatisé de masses cancéreuses dans les clichés mammaires identifiant d'abord les régions d'intérêt au moyen d'une analyse Fourier (au moyen d'un corrélateur optique). On extrait des données de contexte du cliché mammaire pour chaque région d'intérêt, telles que les dimensions, l'emplacement, le classement, la clarté, la densité et l'isolement relatif d'autres régions d'intérêt. On fait des moyennes des pixels dans la région d'intérêt pour créer un réseau plus petit de superpixels (24) qui sont saisis dans un premier réseau neuronal (14). Un second réseau neuronal (15) reçoit les valeurs de sortie en provenance du premier réseau neuronal et les données de contexte sous forme d'entrées, puis génère un résultat indiquant si les régions d'intérêt contiennent ou non une masse cancéreuse. Le second réseau neuronal peut être pourvu de données de contexte d'une autre vue du même sein, de la même vue de l'autre sein ou d'un cliché mammaire antérieur pour le même patient.
PCT/US1998/015350 1998-07-23 1998-07-23 Systeme de detection automatise de masses cancereuses dans les cliches mammaires WO2000005677A1 (fr)

Priority Applications (2)

Application Number Priority Date Filing Date Title
AU85859/98A AU8585998A (en) 1998-07-23 1998-07-23 System for automated detection of cancerous masses in mammograms
PCT/US1998/015350 WO2000005677A1 (fr) 1998-07-23 1998-07-23 Systeme de detection automatise de masses cancereuses dans les cliches mammaires

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/US1998/015350 WO2000005677A1 (fr) 1998-07-23 1998-07-23 Systeme de detection automatise de masses cancereuses dans les cliches mammaires

Publications (1)

Publication Number Publication Date
WO2000005677A1 true WO2000005677A1 (fr) 2000-02-03

Family

ID=22267542

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US1998/015350 WO2000005677A1 (fr) 1998-07-23 1998-07-23 Systeme de detection automatise de masses cancereuses dans les cliches mammaires

Country Status (2)

Country Link
AU (1) AU8585998A (fr)
WO (1) WO2000005677A1 (fr)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2001057777A3 (fr) * 2000-02-04 2002-08-22 Arch Dev Corp Procede, systeme et support lisible par ordinateur destines a un poste de travail de recherche intelligente pour interpretation assistee par ordinateur d'images medicales
FR2870447A1 (fr) * 2004-05-21 2005-11-25 Gen Electric Dispositif de classification de pixels en mammographie a contraste accentue
WO2008055760A1 (fr) * 2006-11-09 2008-05-15 Siemens Aktiengesellschaft Procédé pour produire une radiographie pendant une mammographie
US10595805B2 (en) 2014-06-27 2020-03-24 Sunnybrook Research Institute Systems and methods for generating an imaging biomarker that indicates detectability of conspicuity of lesions in a mammographic image
CN111402335A (zh) * 2020-03-18 2020-07-10 东软睿驰汽车技术（沈阳）有限公司 深度学习模型的评价方法、装置、电子设备及存储介质

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5003979A (en) * 1989-02-21 1991-04-02 University Of Virginia System and method for the noninvasive identification and display of breast lesions and the like
US5212637A (en) * 1989-11-22 1993-05-18 Stereometrix Corporation Method of investigating mammograms for masses and calcifications, and apparatus for practicing such method
US5491627A (en) * 1993-05-13 1996-02-13 Arch Development Corporation Method and system for the detection of microcalcifications in digital mammograms
US5537485A (en) * 1992-07-21 1996-07-16 Arch Development Corporation Method for computer-aided detection of clustered microcalcifications from digital mammograms
US5754693A (en) * 1991-02-18 1998-05-19 Sumitomo Osaka Cement Company Limited Method of optical recognition and classification of patterns
US5815591A (en) * 1996-07-10 1998-09-29 R2 Technology, Inc. Method and apparatus for fast detection of spiculated lesions in digital mammograms

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5003979A (en) * 1989-02-21 1991-04-02 University Of Virginia System and method for the noninvasive identification and display of breast lesions and the like
US5212637A (en) * 1989-11-22 1993-05-18 Stereometrix Corporation Method of investigating mammograms for masses and calcifications, and apparatus for practicing such method
US5754693A (en) * 1991-02-18 1998-05-19 Sumitomo Osaka Cement Company Limited Method of optical recognition and classification of patterns
US5537485A (en) * 1992-07-21 1996-07-16 Arch Development Corporation Method for computer-aided detection of clustered microcalcifications from digital mammograms
US5491627A (en) * 1993-05-13 1996-02-13 Arch Development Corporation Method and system for the detection of microcalcifications in digital mammograms
US5815591A (en) * 1996-07-10 1998-09-29 R2 Technology, Inc. Method and apparatus for fast detection of spiculated lesions in digital mammograms

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
HARA TAKESHI ET AL: "Digital Mammography, Proceedings of the 3rd International Workshop on Digital Mammography", 9 June 1996 (1996-06-09) - 12 June 1996 (1996-06-12), pages 257 - 262 *

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2001057777A3 (fr) * 2000-02-04 2002-08-22 Arch Dev Corp Procede, systeme et support lisible par ordinateur destines a un poste de travail de recherche intelligente pour interpretation assistee par ordinateur d'images medicales
US6901156B2 (en) 2000-02-04 2005-05-31 Arch Development Corporation Method, system and computer readable medium for an intelligent search workstation for computer assisted interpretation of medical images
US7184582B2 (en) 2000-02-04 2007-02-27 Arch Development Corporation Method, system and computer readable medium for an intelligent search workstation for computer assisted interpretation of medical images
FR2870447A1 (fr) * 2004-05-21 2005-11-25 Gen Electric Dispositif de classification de pixels en mammographie a contraste accentue
US7298884B2 (en) 2004-05-21 2007-11-20 General Electric Company Method and apparatus for classification of pixels in medical imaging
WO2008055760A1 (fr) * 2006-11-09 2008-05-15 Siemens Aktiengesellschaft Procédé pour produire une radiographie pendant une mammographie
US8326009B2 (en) 2006-11-09 2012-12-04 Siemens Aktiengesellschaft Method for producing an X-ray image during a mammography
US10595805B2 (en) 2014-06-27 2020-03-24 Sunnybrook Research Institute Systems and methods for generating an imaging biomarker that indicates detectability of conspicuity of lesions in a mammographic image
CN111402335A (zh) * 2020-03-18 2020-07-10 东软睿驰汽车技术（沈阳）有限公司 深度学习模型的评价方法、装置、电子设备及存储介质
CN111402335B (zh) * 2020-03-18 2023-07-28 东软睿驰汽车技术（沈阳）有限公司 深度学习模型的评价方法、装置、电子设备及存储介质

Also Published As

Publication number Publication date
AU8585998A (en) 2000-02-14

Similar Documents

Publication Publication Date Title
US6246782B1 (en) System for automated detection of cancerous masses in mammograms
US7646902B2 (en) Computerized detection of breast cancer on digital tomosynthesis mammograms
US6760468B1 (en) Method and system for the detection of lung nodule in radiological images using digital image processing and artificial neural network
US5657362A (en) Automated method and system for computerized detection of masses and parenchymal distortions in medical images
US7903861B2 (en) Method for classifying breast tissue density using computed image features
US5133020A (en) Automated method and system for the detection and classification of abnormal lesions and parenchymal distortions in digital medical images
AU2005207310B2 (en) System and method for filtering a medical image
US4907156A (en) Method and system for enhancement and detection of abnormal anatomic regions in a digital image
JP4184842B2 (ja) 画像判別装置、方法およびプログラム
Qian et al. Image feature extraction for mass detection in digital mammography: Influence of wavelet analysis
CA2744690C (fr) Analyse d'image
US20120014578A1 (en) Computer Aided Detection Of Abnormalities In Volumetric Breast Ultrasound Scans And User Interface
US6738500B2 (en) Method and system for detecting small structures in images
JP2001510360A (ja) 二段分類器を使用するディジタル放射線画像中の塊検出
JPH09508815A (ja) 乳房ｘ線写真における、改良された、コンピュータによる、腫瘤の発見および分類のための自動的方法およびシステム
CN101103924A (zh) 基于乳腺x线摄片的乳腺癌计算机辅助诊断方法及其系统
Rizzi et al. A supervised method for microcalcification cluster diagnosis
Lado et al. Real and simulated clustered microcalcifications in digital mammograms. ROC study of observer performance
WO2000005677A1 (fr) Systeme de detection automatise de masses cancereuses dans les cliches mammaires
WO2002015113A2 (fr) Examen mammographique assiste par ordinateur servant a detecter et a classifier des microcalcifications
JP2004195080A (ja) 異常陰影検出装置
ABUBACKER et al. AN IMPROVED PERIPHERAL ENHANCEMENT OF MAMMOGRAM IMAGES BY USING FILTERED REGION GROWING SEGMENTATION.
Nguyen et al. Detect abnormalities in mammograms by local contrast thresholding and rule-based classification
Haldorai et al. Survey of Image Processing Techniques in Medical Image Assessment Methodologies
Sample Computer assisted screening of digital mammogram images

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A1

Designated state(s): AL AM AT AU AZ BA BB BG BR BY CA CH CN CU CZ DE DK EE ES FI GB GE GH GM HR HU ID IL IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MD MG MK MN MW MX NO NZ PL PT RO RU SD SE SG SI SK SL TJ TM TR TT UA UG UZ VN YU ZW

AL Designated countries for regional patents

Kind code of ref document: A1

Designated state(s): GH GM KE LS MW SD SZ UG ZW AM AZ BY KG KZ MD RU TJ TM AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE BF BJ CF CG CI CM GA GN GW ML MR NE SN TD TG

121 Ep: the epo has been informed by wipo that ep was designated in this application
DFPE Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101)
REG Reference to national code

Ref country code: DE

Ref legal event code: 8642

122 Ep: pct application non-entry in european phase
NENP Non-entry into the national phase

Ref country code: CA