Systems, methods, and computer-readable media for patient image analysis to identify new diseases

Inventors

Gelbman, Dekel • Gurovich, Yaron

Assignees

Fdna Inc

Abstract

Systems, methods, and computer-readable media are disclosed for performing image processing in connection with phenotypic analysis. For example, at least one processor may be configured to receive electronic numerical information corresponding to pixels reflective of at least one external soft tissue image of an individual and access geographically dispersed genetic information stored in a database. The geographically dispersed genetic information may include numerical data that correlates anomalies in pixels in soft tissue images of a plurality of geographically dispersed individuals to specific genes or to specific genetic variants. The at least one processor may also be configured to compare the electronic numerical information for the individual with the numerical data of the geographically dispersed genetic information stored in a database, to determine at least a likelihood that the individual has a specific genetic variant, and prioritize, based on the comparison, one or more genetic variants according to likelihood of pathogenicity.

Core Innovation

The invention provides an electronic image processing system that identifies one or more genetic disorders by analyzing a series of pixels in a plurality of images of external soft tissue. For a first individual, the system receives first electronic data reflective of first values corresponding to pixels of the external soft tissue image, where the first values correspond to relationships between at least one group of pixels. For a plurality of second individuals, it receives second electronic data reflective of second values corresponding to second pixels of external soft tissue images, where the second values correspond to relationships between at least one group of pixels.

The system compares at least some of the first electronic data with at least some of the second electronic data and determines, based on the comparison, a plurality of medical conditions that the first individual is likely to possess. In the disclosed embodiments, the electronic data may be represented as pixel-based data or feature vectors, and the analysis can include processing via neural networks. Region-based neural networks and learned models are described as producing confidence scores or pathogenicity likelihood outputs tied to medical conditions.

The invention further describes de-identification pipelines that use convolutional layers, pooling, normalization, and neural-network components to produce anonymized feature vectors prior to recognition. Geographically dispersed genetic databases are used alongside the phenotypic image analysis to support patient matching and clustering, including identification of shared or unknown genetic disorders, and to prioritize gene variants based on pathogenicity likelihood.

Claims Coverage

The document includes three independent claims that cover an electronic image processing system, a computer-implemented method, and a non-transitory computer-readable medium. Across these independent claims, there are four main inventive features.

Each independent claim centers on receiving pixel-based electronic data that represents relationships between groups of pixels from a first individual and multiple second individuals, comparing the datasets, and determining a plurality of medical conditions likely to be possessed by the first individual.

Stated Advantages

Documented Applications