Method and device for case-based decision support
AssigneeKoninklijke Philips N.V.
InventorsLalitha Agnihotri, Lilla Boroczky, Luyin Zhao, Michael Chun-chieh Lee
Life of this patent
Estimating the term…
Abstract
This invention relates to a method and device for case-based decision support. It proposes that a case-based decision support system is trained on inputs from several radiologists in order to have a “baseline” system, and then the system provides an option to a radiologist to refine the baseline system based on his/her inputs which either refine weights of features for similarity distance computation directly or provide new similarity ground truth clusters. By enabling modifying the similarity distance computation based on user inputs, this invention adapts similarity ground truth to different users with different experience and/or different opinions.
Claims20
A method comprising the steps of:
- performing a similarity distance computation between an input query case and a set of cases to retrieve a first subset of similar cases by using a default set of features and weights for assessment of similarity,
- wherein each feature comprises a case characterization and each weight corresponds to a feature;
- presenting the first subset of similar cases and the default set of features and weights to a user;
- receiving first input from the user, including:
- an indication of a number of cases to be retrieved;
- an indication of specific data or a specific image to be presented with each retrieved case;
- and
- an indication of a specific case to exclude from retrieval;
- (i) receiving second input from the user, including:
- an indication that an adaptation to the set of features and weights is needed, and at least one of (a) a modified weight for at least one of the default set of features, wherein the user directly modifies the weight or (b) at least one new feature in addition to the default set of features, wherein the user directly adds the new feature;
- (ii) modifying the similarity distance computation with the at least one new feature, and modified weights for retrieving a second subset of similar cases for the user;
- (iii) presenting the retrieved second subset of similar cases with the specific data or specific image for each retrieved case;
- repeating steps (i)-(iii) until receiving third input from the user indicating acceptance.
A processing device including a non-transitory computer-readable storage medium and comprising:
- a retrieving unit arranged for performing similarity distance computation between an input query case and a set of cases to retrieve a first subset of similar cases using a default set of features and weights for assessment of similarity,
- wherein each feature comprises a case characterization and each weight corresponds to a feature;
- a presenting unit arranged for presenting the first subset of similar cases and the default set of features and weights to a user;
- a receiving unit arranged for receiving first input from the user, including:
- an indication of a number of cases to be retrieved;
- an indication of specific data or a specific image to be presented with each retrieved case;
- and
- an indication of a specific case to exclude from retrieval;
- (i) receiving second input from the user, including:
- an indication that an adaptation to the set of features and weights is needed, and at least one of (a) a modified weight for at least one of the default set of features, wherein the user directly modifies the weight or (b) at least one new feature in addition to the default set of features, wherein the user directly adds the new feature;
- and
- (ii) a modifying unit arranged for modifying the similarity distance computation with the at least one new feature and-modified weights by providing a new similarity ground truth for retrieving a second subset of similar cases for the user;
- (iii) presenting the retrieved second subset of similar cases with the specific data or specific image for each retrieved case;
- repeating steps (i)-(iii) until receiving third input from the user indicating acceptance.
Dependencies are inferred from the claim text.