Abnormal tissue detection via modal upstream data fusion

Patent metadata

Field	Value
Publication number	US-11361868-B2
Application number	US-201816103987-A
Country	US
Kind code	B2
Filing date	Aug 16, 2018
Priority date	Aug 16, 2017
Publication date	Jun 14, 2022
Grant date	Jun 14, 2022

Abstract

A method for characterizing and locating abnormal tissue in an area of interest in a body is provided. The method may include receiving first and second modality examination data including a plurality of first and second modality examination data segments of the area of interest and determining, by evaluating each first and second modality examination data segment, a set of modality detection candidates based on the first and second modality examination data and a first and second modality feature likelihood model. The method may also include generating a modal or multi-modal correspondence anatomic model of the area of interest by registering the plurality of first and second modality examination data segments, determining one or more modal abnormal tissue detections by fusing the set of modality detection candidates using the modal or multimodal correspondence anatomic model, and providing a modal diagnosis confidence score and a modal anatomic location based on the one or more modal abnormal tissue detections.

First claim

That which is claimed:

1. A method for characterizing and locating abnormal tissue in an area of interest in a body, the method comprising:
receiving first modality examination data of the area of interest captured by a medical imaging device, the first modality examination data comprising a plurality of first modality examination data segments, each first modality examination data segment being associated with a portion of the area of interest or a first modality sensor field of view into the area of interest;
determining, by evaluating each first modality examination data segment with a computing machine, a set of first modality detection candidates based on the first modality examination data and a first modality feature likelihood model, each first modality detection candidate comprising a first modality likelihood score and a first modality relative location;
generating, by a data fusion machine, a modal correspondence anatomic model of the area of interest by registering the plurality of first modality examination data segments;
determining, by the data fusion machine, one or more modal abnormal tissue detections by fusing the set of first modality detection candidates using the modal correspondence anatomic model; and
providing, by the data fusion machine, a modal diagnosis confidence score and a modal anatomic location based on the one or more modal abnormal tissue detections.

2. The method of claim 1 further comprising:
receiving second modality examination data of the area of interest captured by a second medical imaging device, the second modality examination data comprising a plurality of second modality examination data segments, each second modality examination data segment being associated with a portion of the area of interest or a second modality sensor field of view into the area of interest; and
determining, by evaluating each second modality examination data segment, a set of second modality detection candidates based on the second modality examination data and a second modality feature likelihood model, each second modality detection candidate comprising a second modality likelihood score and a second modality relative location;
wherein generating the modal correspondence anatomic model comprises registering the plurality of first modality examination data segments and the plurality of second modality examination data segments;
wherein determining the one or more modal abnormal tissue detections comprises fusing the set of first modality detection candidates and the set of second modality detection candidates based on the modal correspondence anatomic model.

3. The method claim 1 , wherein fusing the set of first modality detection candidates comprises:
generating, by the data fusion machine, a plurality of object association hypotheses, each object association hypothesis being associated with one or more modality detection candidates based at least on the first modality relative locations;
determining a probability of association for each object association hypothesis;
removing one or more object association hypotheses based on the probability of association for the object association hypothesis failing to meet an association threshold; and
determining, by the data fusion machine, the one or more modal abnormal tissue detections based on the remaining object association hypotheses.

4. The method of claim 1 , wherein registering the plurality of first modality examination data segments comprises:
evaluating, by the data fusion machine, each of the modality examination data segments to identify registration features within the modality examination data segments that exceed a characteristic threshold and include a matchable characteristic;
identifying, for a selected registration feature, two or more modality examination data segments that include the matchable characteristic; and
generating a location-based linkage between the two or more modality examination data segments associated with the selected registration feature for inclusion in the modal correspondence anatomic model.

5. The method of claim 1 , further comprising generating, via machine learning, the first modality likelihood model based on first modality examination training data that has been validated based on positive diagnosis and false-positive diagnosis performance.

6. The method of claim 1 , wherein the modal diagnosis confidence score indicates a probability that abnormal tissue located at the modal anatomic location is malignant.

7. The method of claim 1 , wherein the first modality likelihood scores are non-binary, quantitative cancer-likeness scores.

8. The method of claim 1 , wherein each first modality examination data segment is associated with a two-dimensional, cross-section images of the area of interest.

9. The method of claim 1 , further comprising:
associating features and location data between the first modality examination data segments using statistical data association; and
determining, based on associated features and location data, which first modality detection candidates correspond to a same abnormal tissue with a probability of correct association.

10. The method of claim 1 , wherein the medical imaging device is at least one of: ultrasound, positron emission tomography (PET), computed tomography (CT), x-rays, magnetic resonance imaging (MRI), or single-photon emission computed tomography (SPECT) during a first examination of the body.

11. A system for characterizing and locating abnormal tissue in an area of interest in a body, the system comprising processing circuitry configured to:
receive first modality examination data of the area of interest, the first modality examination data comprising a plurality of first modality examination data segments, each first modality examination data segment being associated with a portion of the area of interest or a first modality sensor field of view into the area of interest;
determine, by evaluating each first modality examination data segment, a set of first modality detection candidates based on the first modality examination data and a first modality feature likelihood model, each first modality detection candidate comprising a first modality likelihood score and a first modality relative location;
generate a modal correspondence anatomic model of the area of interest by registering the plurality of first modality examination data segments;
determine one or more modal abnormal tissue detections by fusing the set of first modality detection candidates using the modal correspondence anatomic model; and
provide a modal diagnosis confidence score and a modal anatomic location based on the one or more modal abnormal tissue detections.

12. The system of claim 11 , wherein the processing circuitry is further configured to:
receive second modality examination data of the area of interest, the second modality examination data comprising a plurality of second modality examination data segments, each second modality examination data segment being associated with a portion of the area of interest or a second modality sensor field of view into the area of interest; and
determine, by evaluating each second modality examination data segment, a set of second modality detection candidates based on the second modality examination data and a second modality feature likelihood model, each second modality detection candidate comprising a second modality likelihood score and a second modality relative location;
wherein being configured to generate the modal correspondence anatomic model comprises being configured to register the plurality of first modality examination data segments and the plurality of second modality

Assignees

• Univ Johns Hopkins

Inventors

• Porter David W
• Walton William C
• Peyton Keith S
• Harvey Susan C
• Tsui Benjamin M W

Classifications

• G16H50/30

  for calculating health indices; for individual health risk assessment · CPC title

• G16H50/50 Primary

  for simulation or modelling of medical disorders · CPC title

• G16H30/40

  for processing medical images, e.g. editing · CPC title