System, method and computer product for differentiating between tissue states and types

Inventors

TOLEDANO, Shani • Tshuva, Moshe • GAT, Sharon

Assignees

HT Bioimaging Ltd

Abstract

A method for differentiating between tissue states or types including receiving a sequence of thermal data of a tissue, wherein the sequence is sampled at at least one location of the tissue while the tissue is being thermally disturbed, deriving from the thermal data at least one tissue-related thermal variable associated with each of the tissue locations, segmenting the tissue into segments comprising the locations having corresponding the at least one thermal variable and generating an output indicating the tissue segments. The disclosure also includes a system and computer product for differentiating between tissue states or types.

Core Innovation

The invention relates to a thermography-based method for differentiating tissue states or tissue types. It controls a heating source to heat at least a surface of at least a portion of tissue from an initial, base temperature over a predetermined first period of time, and samples thermal data over the first period of time and over a second period of time while the tissue is allowed to passively return to the base temperature.

From the sampled thermal data, the invention derives a set of thermal distribution related features with respect to each of a plurality of locations within the tissue. The set of features includes calculating features for the first period of time and for the second period of time, and the tissue is segmented into segments, each comprising one or more of the plurality of locations that have corresponding sets of features.

Based on the sets of features, the invention generates an output that indicates the tissue segments and identifies various tissue states in the tissue segments. Tissue state identification is performed using the derived thermal distribution features, including correlation to predefined signature values and/or a machine learning classifier trained on labeled thermal feature sets, and the described tissue-state categories include normal, inflamed, neoplasm, dysplasia, and cancerous tissue.

In the described implementations, thermal data is captured as thermal imaging in which each location corresponds to a pixel or a voxel. The derived features include thermal distribution characteristics calculated for heating and passive return, such as variance/correlation based features and curve-peak or growth/decay related signatures, and the system output can be presented as a thermal map or distribution map with overlay on RGB imagery to support tissue-type identification and tissue segmentation.

Claims Coverage

The document contains three independent claims: a method claim, a computer program product claim, and a system claim. Across these independent claims, there are four main inventive feature themes.

Across the independent claims, the coverage centers on controlled heating with time-separated thermal sampling, derivation of thermal distribution related features for multiple tissue locations across both periods, tissue segmentation according to corresponding feature sets, and output generation that indicates segments and identifies tissue states.

Stated Advantages

Documented Applications