Similar case searching apparatus and similar case searching method

• US 8,958,613 B2
• Filed: 05/29/2012
• Issued: 02/17/2015
• Est. Priority Date: 11/04/2011
• Status: Active Grant

• Current Assignee: Panasonic Corporation (Panasonic Holdings Corporation)
• Original Assignee: Panasonic Corporation (Panasonic Holdings Corporation)
• Inventors: Kondo, Kenji, Takata, Kazutoyo, Kozuka, Kazuki
• Primary Examiner(s): Koziol, Stephen R
• Assistant Examiner(s): SCHWARTZ, RAPHAEL M
• Application Number:
  US13/482,129
• Publication Number:
  US 20130114867A1
• Time in Patent Office: 994 Days
• Field of Search: None
• US Class Current: 382/128
• CPC Class Codes: G06F 18/2113 by ranking or filtering the set of features, e.g. using a measure of variance or of feature cross-correlation
• CPC Class Codes: G06V 30/1985 Syntactic analysis, e.g. using a grammatical approach syntactic image representation G06V30/18171
• CPC Class Codes: G16H 30/20 for handling medical images, e.g. DICOM, HL7 or PACS
• CPC Class Codes: G16H 30/40 for processing medical images, e.g. editing
• CPC Class Codes: G16H 50/70 for mining of medical data, e.g. analysing previous cases of other patients

Abstract

A similar case searching apparatus comprises: an image feature extracting unit; an image interpretation item fitting degree calculating unit which calculates the fitting degree of each image feature quantity to each of image interpretation items, based on first image interpretation knowledge indicating the range of values of image feature quantities of each type; an image interpretation item selecting unit; a weight determining unit which determines the weight to each image feature quantity based on second image interpretation knowledge defining correlations between image feature quantities, such that the weight is larger as the correlation with the image interpretation item is higher; a similar case searching unit which searches for the similar case data items, by adding the determined weight to each image feature quantity of the interpretation target image and a corresponding image feature quantity of the medical images in a case database, and comparing the weighted image feature quantities.

Claims

1. A similar case searching apparatus which searches a case database for similar case data items including similar images similar to an interpretation target image, the case database storing a plurality of case data items, and each of the plurality of case data items and a case data item of the interpretation target image including one or more medical images and an image interpretation report that is a document data item indicating interpretations of the medical images, said similar case searching apparatus comprising:

an image feature quantity extracting unit configured to extract a plurality of image feature quantities from one of the medical images that is the interpretation target image;

an image interpretation item fitting degree calculating unit configured to calculate a fitting degree of the image feature quantities extracted by said image feature quantity extracting unit with respect to each of image interpretation items, based on first image interpretation knowledge composed of information items each indicating a presence range of values of image feature quantities of a corresponding type extracted from medical images whose image interpretation reports include the image interpretation item, each of the image interpretation items being a character string verbally indicating a feature of the medical images, the information items being prepared for the respective image interpretation items, and the values being calculated with respect to each of the image interpretation items;

an image interpretation item candidate display unit configured to display one of (a) image interpretation items each having a fitting degree larger than a predetermined threshold value and (b) a predetermined number of image interpretation items selected in a descending order of their fitting degrees, both (a) and (b) being included in the image interpretation items;

an image interpretation item selecting unit configured to allow a user to select one or more of the image interpretation items displayed by said image interpretation item candidate display unit;

a weight determining unit configured to determine, based on second image interpretation knowledge, a weight that is added to each of the image feature quantities extracted by said image feature extracting unit, such that the weight to the image feature quantity is larger as a correlation between the image feature quantity and the one of the image interpretation items selected by said image interpretation item selecting unit is higher, the second image interpretation knowledge being prepared information defining correlations between image feature quantities extracted from medical images and image interpretation items selected from the image interpretation reports of the medical images; and

a similar case searching unit configured to search the case database for the similar case data items, by weighting each of the image feature quantities in a first set extracted from the interpretation target image by said image feature quantity extracting unit and a corresponding one of the image feature quantities in a second set extracted from the medical images included in the case data items registered in the case database, using the weight to the image feature quantity determined by said weight determining unit, and comparing the weighted image feature quantities in the first set and the weighted image feature quantities in the second set.

2. The similar case searching apparatus according to claim 1,

wherein said image interpretation item fitting degree calculating unit is configured to:

obtain, from the first image interpretation knowledge, a presence range information item indicating values of all of the image feature quantities related to each of target image interpretation items with respect to which the fitting degrees are calculated;

calculate matching degrees of the respective image feature quantities extracted by said image feature quantity extracting unit with respect to the obtained presence range information item; and

calculate the fitting degree of the image feature quantities extracted by said image feature quantity extracting unit with respect to each of the target image interpretation items, by (i) weighting the image feature quantity using a weight calculated with respect to the matching degree of the image feature quantity such that the weight to the image feature quantity is larger as the image feature quantity is determined, based on the second image interpretation knowledge, to be more highly related to the target image interpretation item, and (ii) integrating the matching degrees calculated for the image feature quantities extracted by said image feature quantity extracting unit.

3. The similar case searching apparatus according to claim 1,

wherein said image interpretation item candidate display unit is further configured to display the fitting degrees with respect to the image interpretation items, together with the image interpretation items.

4. The similar case searching apparatus according to claim 1,

wherein said image interpretation item candidate display unit is further configured to:

determine a co-occurrence probability of each of pairs of the image interpretation items selected by said image interpretation item selecting unit, based on prepared co-occurrence probability information defining a co-occurrence degree of the pair of the image interpretation items in the image interpretation reports attached to the medical images; and

display information indicating a possibility that an inappropriate image interpretation item is included by mistake in one of the pairs having the determined co-occurrence probability when the determined co-occurrence probability is smaller than or equal to a predetermined value.

5. The similar case searching apparatus according to claim 1, wherein said image interpretation item candidate display unit is further configured to display, based on a prepared co-occurrence probability information, information indicating that it is impossible to select the image interpretation item having a co-occurrence probability calculated as being smaller than or equal to the predetermined value with respect to one of the image interpretation items selected by said image interpretation item selecting unit, the prepared co-occurrence probability information defining the co-occurrence degrees of the pairs of the image interpretation items included in the image interpretation reports attached to the medical images.

6. The similar case searching apparatus according to claim 1,

wherein said image interpretation item candidate display unit is further configured to:

estimate a disease name having a highest correlation with each of the selected image interpretation items, based on correlation information extracted from the image interpretation reports attached to the medical images, the correlation information being prepared as information defining correlations between disease names and image interpretation items each of which is a character string verbally indicating a feature of medical images; and

display the estimated disease name.

8. The similar case searching apparatus according to claim 1,

wherein said image interpretation item setting unit is configured to set one or more of the image interpretation items for use in a similar case search, based on history information of image interpretation items previously selected by the user from among the sets of image interpretation items, the one or more of the image interpretation items being obtained by performing the same selection as in the history information from among the image interpretation items having fitting degrees larger than or equal to the predetermined threshold value calculated by said image interpretation item fitting degree calculating unit.

7. A similar case searching apparatus which searches a case database for similar case data items including similar images similar to an interpretation target image, the case database storing a plurality of case data items, and each of the plurality of case data items and a case data item of the interpretation target image including one or more medical images and an image interpretation report that is a document data item indicating interpretations of the medical images, said similar case searching apparatus comprising:

an image feature quantity extracting unit configured to extract a plurality of image feature quantities from one of the medical images that is the interpretation target image;

an image interpretation item fitting degree calculating unit configured to calculate a fitting degree of the image feature quantities extracted by said image feature quantity extracting unit with respect to each of image interpretation items, based on first image interpretation knowledge composed of information items each indicating a presence range of values of image feature quantities of a corresponding type extracted from medical images whose image interpretation reports include the image interpretation item, each of the image interpretation items being a character string verbally indicating a feature of the medical images, the information items being prepared for the respective image interpretation items, and the values being calculated with respect to each of the image interpretation items;

an image interpretation item setting unit configured to set, as an image interpretation item for use in a similar case search, an image interpretation item having a fitting degree that is a value calculated by said image interpretation item fitting degree calculating unit as being larger than or equal to a predetermined threshold value and;

a weight determining unit configured to determine, based on second image interpretation knowledge, a weight that is added to each of the image feature quantities extracted by said image feature extracting unit, such that the weight to the image feature quantity is larger as a correlation between the image feature quantity and the one of the image interpretation items set by said image interpretation item setting unit is higher, the second image interpretation knowledge being prepared information defining correlations between image feature quantities extracted from medical images and image interpretation items selected from the image interpretation reports of the medical images;

and a similar case searching unit configured to search the case database for the similar case data items, by weighting each of the image feature quantities in a first set extracted from the interpretation target image by said image feature quantity extracting unit and a corresponding one of the image feature quantities in a second set extracted from the medical images included in the case data items registered in the case database, using the weight to the image feature quantity determined by said weight determining unit, and comparing the weighted image feature quantities in the first set and the weighted image feature quantities in the second set.

9. A similar case searching method of searching a case database for similar case data items including similar images similar to an interpretation target image, the case database storing a plurality of case data items, and each of the plurality of case data items and a case data item of the interpretation target image including one or more medical images and an image interpretation report that is a document data item indicating interpretations of the medical images, said similar case searching method comprising:

extracting a plurality of image feature quantities from one of the medical images that is the interpretation target image;

calculating a fitting degree of the image feature quantities extracted in said extracting with respect to each of image interpretation items, based on first image interpretation knowledge composed of information items each indicating a presence range of values of image feature quantities of a corresponding type extracted from medical images whose image interpretation reports include the image interpretation item, each of the image interpretation items being a character string verbally indicating a feature of the medical images, the information items being prepared for the respective image interpretation items, and the values being calculated with respect to each of the image interpretation items;

displaying one of (a) image interpretation items each having a fitting degree larger than a predetermined threshold value and (b) a predetermined number of image interpretation items selected in a descending order of their fitting degrees, both (a) and (b) being included in the image interpretation items;

allowing a user to select one or more of the image interpretation items displayed in said displaying;

determining, based on second image interpretation knowledge, a weight that is added to each of the image feature quantities extracted in said extracting, such that the weight to the image feature quantity is larger as a correlation between the image feature quantity and the one of the image interpretation items selected in said selecting is higher, the second image interpretation knowledge being prepared information defining correlations between image feature quantities extracted from medical images and image interpretation items selected from the image interpretation reports of the medical images; and

searching the case database for the similar case data items, by weighting each of the image feature quantities in a first set extracted from the interpretation target image in said extracting and a corresponding one of the image feature quantities in a second set extracted from the medical images included in the case data items registered in the case database, using the weight to the image feature quantity determined in said determining, and comparing the weighted image feature quantities in the first set and the weighted image feature quantities in the second set.

11. A non-transitory computer-readable recording medium for use in a computer, said recording medium having a computer program recorded thereon for causing the computer to execute the similar case searching method according to claim 9.

10. A similar case searching method of searching a case database for similar case data items including similar images similar to an interpretation target image, the case database storing a plurality of case data items, and each of the plurality of case data items and a case data item of the interpretation target image including one or more medical images and an image interpretation report that is a document data item indicating interpretations of the medical images, said similar case searching method comprising:

extracting a plurality of image feature quantities from one of the medical images that is the interpretation target image;

calculating a fitting degree of the image feature quantities extracted in said extracting with respect to each of image interpretation items, based on first image interpretation knowledge composed of information items each indicating a presence range of values of image feature quantities of a corresponding type extracted from medical images whose image interpretation reports include the image interpretation item, each of the image interpretation items being a character string verbally indicating a feature of the medical images, the information items being prepared for the respective image interpretation items, and the values being calculated with respect to each of the image interpretation items;

setting, as an image interpretation item for use in a similar case search, an image interpretation item having a fitting degree that is a value calculated in said calculating as being larger than or equal to a predetermined threshold value and;

determining, based on second image interpretation knowledge, a weight that is added to each of the image feature quantities extracted in said extracting, such that the weight to the image feature quantity is larger as a correlation between the image feature quantity and the one of the image interpretation items selected in said selecting is higher, the second image interpretation knowledge being prepared information defining correlations between image feature quantities extracted from medical images and image interpretation items selected from the image interpretation reports of the medical images; and

searching the case database for the similar case data items, by weighting each of the image feature quantities in a first set extracted from the interpretation target image in said extracting and a corresponding one of the image feature quantities in a second set extracted from the medical images included in the case data items registered in the case database, using the weight to the image feature quantity determined in said determining, and comparing the weighted image feature quantities in the first set and the weighted image feature quantities in the second set.

12. A non-transitory computer-readable recording medium for use in a computer, said recording medium having a computer program recorded thereon for causing the computer to execute the similar case searching method according to claim 10.