Resemblance retrieval apparatus, and recording medium for recording resemblance retrieval program

• US 6,292,577 B1
• Filed: 05/01/1998
• Issued: 09/18/2001
• Est. Priority Date: 05/27/1997
• Status: Expired due to Fees

• Current Assignee: Mitsubishi Denki Kabushiki Kaisha (Mitsubishi Electric Corporation)
• Original Assignee: Mitsubishi Electric Corporation
• Inventors: Takahashi, Masanobu
• Primary Examiner(s): CHANG, JON CARLTON
• Application Number:
  US09/070,739
• Time in Patent Office: 1,236 Days
• Field of Search: 382/128, 382/133, 382/190, 382/305, 707/1, 707/3, 707/6
• US Class Current: 382/128
• CPC Class Codes: G06F 16/583 using metadata automatically derived from the content
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06V 20/69 Microscopic objects, e.g. biological cells or cellular parts
• 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 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: Y10S 707/99936 Pattern matching access

Abstract

A resemblance retrieve apparatus comprises a retrieve condition set producing unit, and resemblance vector data with respect to subject vector data is retrieved and displayed with using a weight vector optimized to a retrieve condition as to each of retrieve conditions produced in the retrieve condition set producing unit.

Claims

1. A resemblance retrieval apparatus comprising:

an image database storing subject images corresponding to respective subjects;

a vector database storing vector data for each of the subjects, the vector data including vector elements corresponding to respective attributes characterizing each of the subjects and a feature quantity quantitatively indicating a feature of a corresponding subject image;

a subject vector data forming unit for forming vector data, the vector data being based upon an image and attribute, for a subject of a resemblance retrieval with respect to the vector data stored in the vector database;

a retrieval condition set forming unit for forming a plurality of retrieval conditions;

a vector weight database storing weight vectors corresponding to respective retrieval conditions;

a resemblance retrieval engine communicating with the vector database, the subject vector data forming unit, the retrieval condition set forming unit, and the vector weight database, and retrieving the vector data from the vector database most closely satisfying the retrieval conditions as weighted by the weight vectors and thereby most closely resembling the subject of the resemblance retrieval; and

a retrieval result display unit communicating with the resemblance retrieval engine and the image database for simultaneously displaying the subject images corresponding to the vector data for each of the subjects that is retrieved by the resemblance retrieval engine from the vector database using the corresponding weight vectors and a corresponding image retrieved from the image database.

2. The resemblance retrieval apparatus of claim 1, further comprising at least one of

a feature quantity extracting unit for extracting a feature quantity of an image stored in the image data base, and

a feature quantity input unit for inputting a feature quantity of an image stored in the image data base, communicating with the subject vector data forming unit to supply as a feature quantity for quantitatively indicating a feature characterizing an image, at least one of the feature quantity extracted by the feature quantity extracting unit and the feature quantity input from the feature quantity input unit.

3. The resemblance retrieval apparatus of claim 1, wherein the images stored in the image database are images of biological tissue and a diagnosis title in a pathological tissue diagnosis is employed as a retrieval condition.

4. The resemblance retrieval apparatus of claim 3, wherein the feature quantity of the images of biological tissue is selected from the group consisting of

quantity of a nuclear region;

area of the nuclear region;

shape of the nuclear region;

circular degree of the nuclear region;

color of the nuclear region;

chromaticity of the nuclear region;

quantity of an air space region;

area of the air space region;

shape of the air space region;

circular degree of the air space region;

quantity of an interstitial region;

area of the interstitial region;

shape of the interstitial region;

circular degree of the interstitial region;

color of the interstitial region;

chromaticity of the interstitial region;

quantity of a tubular region

area of the tubular region;

shape of the tubular region;

circular degree of the tubular region;

texture of an image;

feature quantity obtained from a wavelet transformation value;

degree by which an epithelium cell represents a two-layer structure in combination with a muscular epithelium cell;

degree of texture;

existence of nipple pattern;

existence of mesh-shaped pattern;

existence of a necrotic substance;

existence of a fulfill pattern;

color of an image; and

chromaticity of the image.

5. A recording medium having recorded therein a resemblance retrieval program for causing a computer to execute a process including

storing subject images corresponding to respective subjects in an image database;

storing vector data for each of the subjects in a vector database, the vector data including vector elements corresponding to respective attributes characterizing the subjects and a feature quantity quantitatively indicating a feature of a corresponding subject image;

forming vector data, the vector data being based upon an image and attribute, for a subject of a resemblance retrieval with respect to the vector data stored in the vector database;

forming a plurality of retrieval conditions;

storing weight vectors corresponding to respective retrieval conditions in a vector weight database;

retrieving the vector data from the vector database most closely satisfying the retrieval conditions as weighted by the weight vectors and thereby most closely resembling the subject of the resemblance retrieval; and

simultaneously displaying the subject images corresponding to the vector data for each of the subjects that is retrieved from the vector database using the corresponding weight vectors and a corresponding image retrieved from the image database.