Systems and methods for evaluating hyperspectral imaging data using a two layer media model of human tissue

Patent metadata

Field	Value
Publication number	US-9619883-B2
Application number	US-201414776681-A
Country	US
Kind code	B2
Filing date	Mar 17, 2014
Priority date	Mar 15, 2013
Publication date	Apr 11, 2017
Grant date	Apr 11, 2017

Abstract

The disclosure provides computer systems, non-transitory computer-readable storage medium, and methods for determining the presence of a skin indication in a subject. In some aspects the method includes acquiring a hyperspectral imaging data set from a region of interest of a human subject using a hyperspectral imager, and applying the hyperspectral imaging data set against a classifier including a two layered media model, the two layered media model including a first layer of a modeled human tissue overlying a second layer of the modeled human tissue, where the two layered media model has been trained by application of simulated data across a set of optical and geometric properties associated with the presence or the absence of a skin indication, where the two layered media model computes tissue reflectance R from the modeled human tissue by a relationship provided herein.

First claim

What is claimed is:

1. A computer implemented method, performed by a computer system having one or more processors and memory storing one or more programs for execution by the one or more processors, the method comprising:
acquiring a hyperspectral imaging data set from a region of interest of a human subject using a hyperspectral imager; and
applying the hyperspectral imaging data set against a classifier comprising a two layered media model, the two layered media model comprising a first layer of a modeled human tissue overlying a second layer of the modeled human tissue, wherein the two layered media model has been trained by application of simulated data across a set of optical and geometric properties associated with the presence or the absence of a skin indication, wherein the two layered media model computes tissue reflectance R from the modeled human tissue by the relationship:

R
=

∑

i
=
0

∞

⁢

∑

j
=
0

∞

⁢

exp
⁡

(

ⅈ
⁢

⁢

ω
~

1

+

j
⁢

⁢

ω
~

2

)

⁢

P
⁡

(

ⅈ
,

j
❘

L
1

)

wherein,
L 1 is the thickness of the first layer,
{tilde over (ω)} 1 is a modified single scattering of the first layer,
{tilde over (ω)} 2 is a modified single scattering of the second layer,
exp is the exponential function, and
P(i,j|L 1 ) is a joint probability density function that a photon will experience i interactions with the first layer and j interactions with the second layer given that the first layer thickness is L 1 , wherein the application of the hyperspectral imaging data set against the classifier causes the classifier to produce a determination as to whether the region of interest has the skin indication.

2. The computer implemented method of claim 1 , wherein the determination as to whether the region of interest has the skin indication is a likelihood that the region of interest has the skin indication.

3. The computer implemented method of claim 1 , wherein the determination as to whether the region of interest has the skin indication is either a determination that the region of interest has the skin indication or a determination that the region of interest does not have the skin indication.

4. The computer implemented method of claim 1 , the method further comprising training the model using the trajectory vector {right arrow over (x)} p of each respective photon p in a set of photons comprising a million or more photons.

5. The computer implemented method of claim 4 , wherein, for each respective photon in the set of photons, the trajectory vector {right arrow over (x)} p of the respective photon was calculated such that {right arrow over (x)} p,k = x, y, z k represents the location of the k th interaction between the photon and the two layer model and wherein the trajectory vector {right arrow over (x)} p was stored in a non-transitory data file.

6. The computer implemented method of claim 5 , wherein P(i,j|L 1 ) is approximated as

N
⁡

(

ⅈ
,

j
❘

L
1

)

P

wherein N(i,j|L 1 ) is a count of the set of photons in the set of photons that experience i interactions with the first layer and j interactions with the second layer given L 1 , and has the form:

N
⁡

(

ⅈ
,

j
❘

L
1

)

=

∑

p
=
1

p
=
P

⁢

{

1

N

p
,
1

=

ⅈ
⁢

⁢
and
⁢

⁢

Assignees

• Hypermed Imaging Inc

Inventors

• Yudovsky Dmitry

Classifications

• G06T2207/30088

  Skin; Dermal · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T7/0016 Primary

  involving temporal comparison · CPC title

• G06T2207/10004

  Still image; Photographic image · CPC title

• G06F19/321

  Physics · mapped topic