Systems and methods for platform agnostic whole body image segmentation

Inventors

Richter, Jens Filip Andreas • Johnsson, Kerstin Elsa Maria • Gjertsson, Erik Konrad • Anand, Aseem Undvall

Assignees

Exini Diagnostics AB

Abstract

Presented herein are systems and methods that provide for automated analysis of three-dimensional (3D) medical images of a subject in order to automatically identify specific 3D volumes within the 3D images that correspond to specific anatomical regions (e.g., organs and/or tissue). Notably, the image analysis approaches described herein are not limited to a single particular organ or portion of the body. Instead, they are robust and widely applicable, providing for consistent, efficient, and accurate detection of anatomical regions, including soft tissue organs, in the entire body. In certain embodiments, the accurate identification of one or more such volumes is used to automatically determine quantitative metrics that represent uptake of radiopharmaceuticals in particular organs and/or tissue regions. These uptake metrics can be used to assess disease state in a subject, determine a prognosis for a subject, and/or determine efficacy of a treatment modality.

Core Innovation

The invention automatically processes 3D images to identify cancerous lesions within a subject by combining machine learning-based target volume identification in a 3D anatomical image with hotspot-based lesion detection in a 3D functional image. A processor receives a 3D anatomical image and automatically identifies, for each of a plurality of target tissue regions, a corresponding target volume of interest (VOI) within the 3D anatomical image. The processor determines a 3D segmentation map representing a plurality of 3D segmentation masks, each segmentation mask representing a particular identified target VOI.

The processor receives a 3D functional image obtained using a functional imaging modality and, using the 3D segmentation map, identifies within the 3D functional image one or more 3D volumes corresponding to the identified target VOIs. Within at least a portion of the identified 3D volumes, the processor automatically detects one or more hotspots that represent lesions based on intensities of voxels within the 3D functional image. The workflow links segmentation-defined anatomical target regions to corresponding functional-image locations for lesion hotspot detection.

In additional described embodiments, detected hotspots and reference tissue regions are used to derive intensity-based index values that support assessment of cancer status, severity, prognosis, and treatment efficacy. Reference intensity values and hotspot intensity values are determined using the 3D segmentation map, and hotspot index values are computed from hotspot intensity-derived values and reference intensity-derived values. An overall index value is derived from weighted aggregation of individual hotspot index values, with continuous interpolation scale mapping for hotspot index values, to quantify target-region involvement.

Claims Coverage

The independent claims cover a method and a system for automatically processing 3D anatomical and 3D functional images to identify cancerous lesions. Across the independent claims, there are five inventive features: machine learning-based target VOI identification, generation of a 3D segmentation map, mapping of target VOIs onto functional-image 3D volumes, hotspot lesion detection based on voxel intensities, and processor-memory instructions for the automated workflow.

Both independent claims cover the same core workflow: machine learning modules identify target VOIs in a 3D anatomical image; a 3D segmentation map defines segmentation masks for those VOIs; the segmentation map is used to locate corresponding 3D volumes in a functional image; and lesions are represented by automatically detected hotspots based on voxel intensities within those mapped functional-image volumes.

Stated Advantages

Documented Applications