# US 6,731,782 B2 · Method and system for automatic identification and quantification of abnormal anatomical structures in medical images

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 6,731,782 B2
- Document ID: US6731782B2
- Status: Granted
- Kind: B2
- Application number: 10/262,672
- Filed: 2002-10-02
- Published: 2004-04-08
- Granted: 2004-05-04
- Earliest claimed: 2002-10-02
- Assignees: VirtualScopics, LLC
- Inventors: Edward Ashton
- Web page: https://patentgrep.com/patent/US6731782B2

## CPC

- G06T 7/0012: {Biomedical image inspection}
- G06T 5/20: using local operators
- G06T 5/70: Denoising; Smoothing
- G06T 7/11: Region-based segmentation
- G06T 7/143: involving probabilistic approaches, e.g. Markov random field [MRF] modelling
- G06T 7/187: involving region growing; involving region merging; involving connected component labelling
- G06T 7/62: of area, perimeter, diameter or volume
- G06V 10/267: {by performing operations on regions, e.g. growing, shrinking or watersheds}
- G06T 2207/10081: Computed x-ray tomography [CT]
- G06T 2207/10088: Magnetic resonance imaging [MRI]
- G06T 2207/30016: Brain
- G06V 2201/032: of protuberances, polyps nodules, etc.
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

Abnormal regions in volumetric image sets are detected and delineated through the following technique. Noise is suppressed in the original data. The background is classified into one or more background classes. An exemplar is identified. Essentially similar structures throughout the volume are identified; a directed clustering technique has been developed for doing so. Quantitative information (lesion volume, shape, etc.) is extracted and output to database.

## Claims

### Claim 1

A method for detection, identification and delineation of a structure in a volume from original data, the method comprising:
(a) performing background characterization in the original data to determine at least one background class in the original data;
(b) identifying an exemplar in the original data;
(c) identifying structures throughout the volume in accordance with the at least one background class determined in step (a) and the exemplar identified in step (b); and
(d) extracting quantitative information from the structures identified in step (c).

### Claim 2 (depends on 1)

The method of claim 1, further comprising suppressing noise in the original data to produce noise suppressed data, and wherein steps (a) and (b) are performed using the noise suppressed data.

### Claim 3 (depends on 2)

The method of claim 2, wherein the noise is suppressed using median filtering.

### Claim 4 (depends on 2)

The method of claim 2, wherein the noise is suppressed using low-pass filtering.

### Claim 5 (depends on 1)

The method of claim 1, wherein step (a) is performed with an adaptive Bayesian classifier.

### Claim 6 (depends on 1)

The method of claim 1, wherein step (b) is performed through manual tracing.

### Claim 7 (depends on 1)

The method of claim 1, wherein step (b) is performed through semi-automated tracing.

### Claim 8 (depends on 1)

The method of claim 1, wherein step (b) is performed through statistical region growth.

### Claim 9 (depends on 1)

The method of claim 1, wherein step (b) is performed through geometrically constrained region growth.

### Claim 10 (depends on 1)

The method of claim 1, wherein step (c) is performed through directed clustering.

### Claim 11 (depends on 10)

The method of claim 10, wherein:
the volume comprises at least one background class and at least one target class; and
the directed clustering comprises:
(i) calculating a statistical descriptor for each of the at least one background class and the at least one target class, the statistical descriptor being used to calculate a discriminant; and
(ii) assigning each voxel to a class for which the discriminant is a minimum.

### Claim 12 (depends on 11)

The method of claim 11, wherein the directed clustering further comprises (iii) re-estimating background class statistics after step (ii).

### Claim 13 (depends on 12)

The method of claim 12, wherein, in the directed clustering, steps (ii) and (iii) are performed iteratively until a stable solution is achieved.

### Claim 14 (depends on 1)

The method of claim 1, wherein step (d) comprises outputting the quantitative information to a database.

### Claim 15 (depends on 1)

The method of claim 1, wherein the volume comprises an area of interest in a living body, and the structure is an abnormal region in the area of interest in the living body.

### Claim 16

A system for detection, identification and delineation of a structure in a volume from original data, the system comprising:
an input device for receiving a user input comprising the original data;
a processor, in communication with the input device, for:
(a) performing background characterization in the original data to determine at least one background class in the original data;
(b) identifying an exemplar in the original data;
(c) identifying structures throughout the volume in accordance with the at least one background class determined in step (a) and the exemplar identified in step (b); and
(d) extracting quantitative information from the structures identified in step (c); and
an output device, in communication with the processor, for outputting a result of step (d).

### Claim 17 (depends on 16)

The system of claim 16, wherein the processor suppresses noise in the original data to produce noise suppressed data and performs steps (a) and (b) using the noise suppressed data.

### Claim 18 (depends on 17)

The system of claim 17, wherein the processor suppresses the noise through median filtering.

### Claim 19 (depends on 17)

The system of claim 17, wherein the processor suppresses the noise through low-pass filtering.

### Claim 20 (depends on 16)

The system of claim 16, wherein the processor performs step (a) with an adaptive Bayesian classifier.

### Claim 21 (depends on 16)

The system of claim 16, wherein the user input comprises a manual tracing of the exemplar, and wherein the processor performs step (b) by receiving the manual tracing from the input device.

### Claim 22 (depends on 16)

The system of claim 16, wherein the processor performs step (b) through semi-automated tracing based on the user input.

### Claim 23 (depends on 16)

The system of claim 16, wherein the processor performs step (b) through statistical region growth.

### Claim 24 (depends on 16)

The system of claim 16, wherein the processor performs step (b) through geometrically constrained region growth.

### Claim 25 (depends on 16)

The system of claim 16, wherein the processor performs step (c) through directed clustering.

### Claim 26 (depends on 27)

The system of claim 27, wherein:
the volume comprises at least one background class and at least one target class; and
the processor performs the directed clustering by:
(i) calculating a statistical descriptor for each of the at least one background class and the at least one target class, the statistical descriptor being used to calculate a discriminant; and
(ii) assigning each voxel to a class for which the discriminant is a minimum.

### Claim 27 (depends on 26)

The system of claim 26, wherein the processor performs the directed clustering further by further (iii) re-estimating background class statistics after step (ii).

### Claim 28 (depends on 27)

The system of claim 27, wherein, in the directed clustering, the processor performs steps (ii) and (iii) iteratively until a stable solution is achieved.

### Claim 29 (depends on 16)

The system of claim 16, wherein the out put device comprises a database, and wherein step (d) comprises outputting the quantitative information to the database.
