Systems and methods for automated identification and classification of lesions in local lymph and distant metastases
Inventors
Brynolfsson, Johan Martin • Sahlstedt, Hannicka Maria Eleonora • Richter, Jens Filip Andreas
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12597127-B2
Publication Date
2026-04-07
Expiration Date
Abstract
Presented herein are systems and methods that provide automated analysis of 3D images to classify representations of lesions identified therein. In particular, in certain embodiments, approaches described herein allow hotspots representing lesions to be classified based on their spatial relationship with (e.g., whether they are in proximity to, overlap with, or are located within) one or more pelvic lymph node regions in detailed fashion.
Core Innovation
The invention provides an automated method for processing aligned 3D functional images and 3D anatomical images of a subject to identify and/or characterize cancerous lesions within a pelvic region. The 3D functional image includes one or more hotspots representing potential lesions, and the 3D anatomical image is segmented to identify representations of one or more pelvic bones, thereby creating a 3D segmentation map aligned with the anatomical image.
A 3D pelvic atlas image is received that includes one or more pelvic lymph sub-regions and one or more reference pelvic bone regions. The pelvic atlas is transformed to co-register with the 3D segmentation map using the reference pelvic bone regions and the pelvic bone regions, thereby creating a transformed 3D pelvic atlas image aligned to the 3D anatomical image and segmentation thereof.
For each hotspot of the 3D functional image, the method determines a pelvic lymph classification using the pelvic lymph sub-regions within the transformed 3D pelvic atlas image. In one framework, the transforming includes determining a first transformation and a second transformation that align subsets of the reference pelvic bone regions to corresponding subsets of the pelvic bone regions, and then determining a final overall transformation based on the first transformation and the second transformation.
In another framework, a plurality of prospective 3D pelvic atlas images is used, with selection of a best-fit pelvic atlas image and refinement of co-registration using a left-side transformation, a right-side transformation, and a weighted transformation based on the left-side transformation and the right-side transformation.
Claims Coverage
The independent claims cover automated processing of aligned 3D functional and anatomical pelvic images, segmentation of pelvic bones to create a 3D segmentation map, atlas-based co-registration using reference pelvic bone regions, and pelvic lymph classification for each hotspot. The claims also include variants using two transformation subsets, prospective atlas best-fit selection, left/right weighted transformation, and a second refinement transformation.
Automated processing of aligned 3D functional and anatomical pelvic images for hotspot pelvic-lymph classification
Receiving a 3D functional image and a 3D anatomical image of a subject with aligned pelvic region representation, segmenting the anatomical image to create a 3D segmentation map, receiving a 3D pelvic atlas image comprising pelvic lymph sub-regions and reference pelvic bone regions, transforming the pelvic atlas to co-register it with the segmentation map using the reference and segmented pelvic bone regions, and determining for each hotspot a pelvic lymph classification using the pelvic lymph sub-regions within the transformed 3D pelvic atlas image.
Two transformation subset-to-subset registration with final overall transformation for atlas co-registration
The transforming comprises determining a first transformation that aligns a first subset of the reference pelvic bone regions to a corresponding first subset of the pelvic bone regions, determining a second transformation that aligns a second subset of the reference pelvic bone regions to a corresponding second subset of the pelvic bone regions, and determining a final overall transformation based on the first transformation and the second transformation to transform the 3D pelvic atlas image.
System embodiment of automated pelvic-bone segmentation, atlas co-registration, and hotspot pelvic lymph classification
A system comprising a processor and memory with instructions causing the processor to receive aligned 3D functional and 3D anatomical images including a pelvic region representation and hotspots, segment the anatomical image to create a 3D segmentation map of pelvic bone regions, receive a 3D pelvic atlas image comprising pelvic lymph sub-regions and reference pelvic bone regions, transform the pelvic atlas to co-register with the segmentation map using the reference and pelvic bone regions, and determine for each hotspot a pelvic lymph classification using the pelvic lymph sub-regions within the transformed 3D pelvic atlas image.
Prospective pelvic atlas plurality with best-fit selection and refined atlas registration using left/right weighted transformation
Receiving a plurality of prospective 3D pelvic atlas images; for each prospective pelvic atlas image determining a corresponding coarse registration transformation that aligns a reference pelvic bone region to a target pelvic bone region; transforming each prospective pelvic atlas image to create transformed prospective pelvic atlas images; selecting a best-fit pelvic atlas image; determining a fine registration transformation by determining a left-side transformation aligning a left reference hip bone region with a left hip bone region, determining a right-side transformation aligning a right hip bone region with a right hip bone region, and determining a weighted transformation based on the left-side transformation and the right-side transformation; transforming the best-fit pelvic atlas image using the weighted transformation to create a final transformed pelvic atlas image; and determining for each hotspot a pelvic lymph classification using pelvic lymph sub-regions within the final transformed pelvic atlas image.
Best-fit 3D pelvic atlas selection via transforming prospective atlases and using transformed best-fit atlas
Performing steps (c) and (d) for a plurality of prospective 3D pelvic atlas images to determine a corresponding transformed version for each prospective atlas and selecting a particular one as a best-fit 3D pelvic atlas image, and using the transformed version of the best-fit 3D pelvic atlas image as the transformed 3D pelvic atlas image to determine the pelvic lymph classification for each hotspot.
System embodiment of prospective pelvic atlas selection with initial best-fit and second refinement transformation
A system in which the instructions cause the processor to receive a plurality of prospective 3D pelvic atlas images; for each prospective pelvic atlas image determine a first registration transformation to co-register it with the 3D segmentation map using reference pelvic bone regions and pelvic bone regions and transform the prospective atlas to create a plurality of transformed prospective pelvic atlas images; select a particular one as an initial best-fit pelvic atlas image; determine for the initial best-fit pelvic atlas image a second registration transformation to refine co-registration using reference pelvic bone regions and pelvic bone regions and transform the initial best-fit pelvic atlas image to create a final transformed pelvic atlas image; and use the final transformed pelvic atlas image to determine the pelvic lymph classification for each hotspot.
Overall, the claims cover an automated 3D imaging workflow that segments pelvic bones, co-registers a 3D pelvic atlas to a 3D segmentation map using reference bone regions, computes a final overall transformation derived from first and second subset alignments, and classifies each functional hotspot via pelvic lymph sub-regions in the transformed atlas.
Stated Advantages
Documented Applications
No documented applications found
Interested in licensing this patent?