Accessing medical image databases using medically relevant terms

Patent metadata

Field	Value
Publication number	US-9953040-B2
Application number	US-67125608-A
Country	US
Kind code	B2
Filing date	Jul 25, 2008
Priority date	Aug 1, 2007
Publication date	Apr 24, 2018
Grant date	Apr 24, 2018

Abstract

A system for accessing a database of a plurality of image data sets includes an acquisition unit which acquires a query for searching the database for an image data set or an image data subset comprised in an image data set. The query includes at least one medically relevant term which defines a search criteria. A determining unit determines the image data set or the image data subset included in the image data set based on the strength of semantic matches between the at least one medically relevant term and (a) corresponding medical annotation(s) describing the image data set. A retrieving unit retrieves the determined image data set or image data subset from the database.

First claim

The invention claimed is:

1. A system for accessing an image set database, the system comprising:
one or more processors configured to:
receive a query including at least one medical term based on a medical ontology and defining a search criteria;
segment an image data set into a plurality of image data subsets based on an adapted mesh shape model, wherein at least one medical annotation annotates at least one of the segmented image data subsets, and the at least one medical annotation includes at least one of an anatomical annotation and a diagnostic annotation, and wherein the adapted mesh shape model includes the at least one medical annotation located within the mesh and an anatomical structure defined by the mesh;
determine a strength of semantic matches between the at least one medical term and the at least one medical annotation generated by the segmentation;
determine an image data set based on the determined strength;
find a plurality of different types of relations between the at least one medical term and the at least one medical annotation;
define a number, the number being the number of found different types of relations in the plurality of different types of relations;
rank the determined image data set based on the defined number;
retrieve a retrieval set based on the ranking;
generate the anatomical annotation for the image data set based on a volume of at least one anatomical structure in a region identified by the adapted mesh shape model;
extract the at least one medical term from the query;
anchor the extracted at least one medical term into medical ontology;
extract medical annotations describing an image data set for determination;
anchor the extracted medical annotations into the medical ontology; and
examine the strength of semantic matches between the anchored medical term and the anchored medical annotation based on the number of relations in a path between the two anchors;

wherein the one or more processors are configured to generate the annotations by being configured to:
generate a first annotation that segments a first anatomical structure in the diagnostic medical images;
generate a second annotation that segments a second anatomical structure in the diagnostic medical images; and
automatically generate a third annotation, the third annotation comprising a ratio between the first annotation and the second annotation;

wherein the anchored medical term includes semantically related diagnostic terms and anatomical terms, and the anchored medical annotation includes semantically related diagnostic annotations and anatomical annotations; and
wherein the medical ontology comprises the Foundational Model of Anatomy system and the Unified Medical Language System.

2. The system as claimed in claim 1 , wherein the one or more processors are further configured to:
generate a diagnostic annotation for the determined image data set based on:
a comparison of an anatomical annotation with a predefined criteria for use in a medical diagnosis; and
a comparison of two anatomical annotations of the determined image data set.

3. The system as claimed in claim 1 , wherein the at least one medical term includes at least one of an anatomical term and a diagnostic term; and
wherein the at least one medical annotation includes at least one of an anatomical annotation and a diagnostic annotation;
wherein the one or more processors are further configured to:
extract the at least one medical term from the query;
anchor the extracted at least one medical term into the medical ontology;
extract the medical annotations describing an image data set for determination;
anchor the extracted medical annotations into the medical ontology; and
examine the strength of semantic matches between the anchored medical term and the anchored medical annotation based on the number of relations in a path between the two anchors;

wherein the anchored medical term includes semantically related diagnostic terms and anatomical terms, and the anchored medical annotation includes semantically related diagnostic annotations and anatomical annotations.

4. The system as claimed in claim 1 , wherein image data sets that are related in at most one query term are discarded from the retrieval set.

5. A method of accessing an image set database, the method comprising:
receiving a query comprising a medical term based on a medical ontology and defining a search criteria;
determining a strength of semantic matches between the medical term and a medical annotation which is generated by a segmentation of the image data set into a plurality of image data subsets based on an adapted mesh shape model and the medical annotation annotates a segment of the segmented image data subsets;
determining a path between the medical term and the medical annotation; and
defining a number, the number being the number of relations in the path between the medical term and the medical annotation;
wherein the strength is further determined based on the defined number;
wherein the method further comprises retrieving an image data set based on the determined strength;
wherein the medical annotation includes a diagnostic annotation, and the diagnostic annotation is generated by:
generating a first anatomical annotation that segments a first anatomical structure in diagnostic medical images;
generating a second anatomical annotation that segments a second anatomical structure in the diagnostic medical images;
automatically generating a third anatomical annotation, the third anatomical annotation comprising a ratio between the first annotation and the second annotation; and
comparing the ratio between the first anatomical annotation and the second anatomical annotation to a predefined ratio to generate the diagnostic annotation;
wherein an additional anatomical annotation is generated for the image data set based on at least one measure of at least one anatomical structure in a region identified by the adapted mesh shape model;

wherein the at least one measure includes a volume calculation; and
wherein the medical ontology comprises the Foundational Model of Anatomy system and the Unified Medical Language System.

6. The method as claimed in claim 5 , wherein the medical term comprises an anatomical term and a diagnostic term.

7. An electronic data processing device including at least one processor configured to perform the method according to claim 5 .

8. A system for accessing an image set database, the system comprising:
one or more processors configured to:
receive a query including at least one medical term based on a medical ontology and defining a search criteria;
segment an image data set into a plurality of image data subsets based on an adapted mesh shape model, wherein at least one medical annotation annotates at least one of the segmented image data subsets, and the at least one medical annotation includes at least one of an anatomical annotation and a diagnostic annotation, and wherein the adapted mesh shape model includes the at least one medical annotation located within the mesh and an anatomical structure defined by the mesh;
determine a strength of semantic matches between the at least one medical term and the at least one medical annotation generated by the segmentation;
determine an image data set based on the determined strength;
find a plurality of different types of relations between the at least one medical term and the at least one medical annotation;
define a number, the number being the number of found different types of relations in the plurality of different types of relations;
rank the determined image data set based on the defined number;
retrieve a retrieval set based on the ranking;
generate the anatomic

Assignees

• Weese Juergen
• Lehmann Helko
• Qian Yuechen
• Ten Kate Warner Rudolph Theophile
• Koninklijke Philips Nv

Inventors

• Weese Juergen
• Lehmann Helko
• Qian Yuechen
• Ten Kate Warner Rudolph Theophile

Classifications

• G06F16/58 Primary

  Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually · CPC title

• G16H30/20

  for handling medical images, e.g. DICOM, HL7 or PACS · CPC title

• G06F16/5866

  using information manually generated, e.g. tags, keywords, comments, manually generated location and time information · CPC title

• G06F19/321

  Physics · mapped topic

• G06F17/30268

  Physics · mapped topic