Memorial Sloan Kettering Cancer Center v. General Electric Co

I. Executive Summary and Procedural Information

• Parties & Counsel:
  • Plaintiff: Memorial Sloan-Kettering Cancer Center (New York)
  • Defendant: General Electric Company (New York); GE HealthCare Technologies Inc. (Delaware); and GE Medical Systems L.L.C. (Delaware)
  • Plaintiff's Counsel: Ashby & Geddes
• Case Identification: 1:26-cv-00633, D. Del., 06/01/2026
• Venue Allegations: Venue is alleged to be proper in the District of Delaware because two of the three defendants (GE HealthCare Technologies Inc. and GE Medical Systems L.L.C.) are incorporated in Delaware. The complaint also alleges all defendants conduct business in the district, have committed acts of infringement there, and that GE has a regular and established place of business, including a manufacturing site, in Delaware.
• Core Dispute: Plaintiff alleges that Defendants' "MotionFree" software, integrated into their Positron Emission Tomography/Computed Tomography (PET/CT) imaging systems, infringes a family of six patents related to methods for retrospectively correcting motion artifacts in medical images using the image data itself.
• Technical Context: The technology addresses motion-induced blurring in medical scans (e.g., from a patient's breathing), which can obscure small lesions critical for cancer diagnosis, by using a software-only method to generate motion-corrected images, eliminating the need for external hardware.
• Key Procedural History: The complaint notes that Defendant GE HealthCare completed a spin-off from Defendant General Electric in January 2023. Plaintiff alleges providing pre-suit notice of infringement to GE as early as November 2019 and to GE HealthCare in May 2023, which may be used to support claims of willful infringement.

Case Timeline

Date	Event
2007-05-04	Earliest Priority Date for all Asserted Patents ('200 Provisional Application)
2017-11-14	U.S. Patent No. 9,814,431 Issued
2018-11-06	U.S. Patent No. 10,117,625 Issued
2019-06-24	Accused Product "Discovery IQ Gen 2" Debut Date (Alleged)
2019-10-22	U.S. Patent No. 10,448,903 Issued
2019-11-06	Plaintiff Alleges GE had Knowledge of '431, '625, and '903 Patents
2020-11-29	Accused Product "Discovery MI Gen 2" Debut Date (Alleged)
2020-12-15	U.S. Patent No. 10,863,950 Issued
2022-06-28	U.S. Patent No. 11,369,322 Issued
2022-10-16	Accused Product "Omni Legend" Debut Date (Alleged)
2023-01-03	GE HealthCare Spin-off from GE Completed
2023-05-30	Plaintiff Alleges GE HealthCare had Knowledge of '431, '625, '903, '950, and '322 Patents
2024-12-24	U.S. Patent No. 12,171,593 Issued
2026-06-01	Complaint Filed

II. Technology and Patent(s)-in-Suit Analysis

U.S. Patent No. 9,814,431 - Methods and Systems for Retrospective Internal Gating

• Patent Identification: U.S. Patent No. 9,814,431, issued November 14, 2017.

The Invention Explained

• Problem Addressed: The patent's background section describes how patient motion during a medical scan, particularly respiratory motion, causes blurring and loss of detail in the resulting images Compl. ¶16 This degradation can limit the ability to detect small lesions and reduce the accuracy of measurements, a significant problem in diagnostic imaging Compl. ¶16 Prior hardware-based solutions for this "respiratory gating," such as pressure belts or cameras, are described as "cumbersome and had a high failure rate" Compl. ¶18 '431 Patent, col. 1:30-41
• The Patented Solution: The invention provides a software-based method to correct for this motion retrospectively, meaning after the image data is acquired Compl. ¶19 Instead of external hardware, it analyzes fluctuations in the image data itself over time to derive a motion signal '431 Patent, abstract The process involves acquiring a series of images, extracting "time-activity" information from individual picture elements (voxels), and algorithmically combining this information to generate a "time varying object motion function" that represents the patient's breathing cycle '431 Patent, FIG. 1 '431 Patent, col. 4:1-8 This function is then used to map the original image data to the correct motion phase and generate a motion-corrected image '431 Patent, col. 4:35-41
• Technical Importance: This approach enabled automated, hardware-free motion correction, potentially making the technique more robust, less intrusive for the patient, and more seamlessly integrated into standard clinical imaging workflows Compl. ¶19

Key Claims at a Glance

• The complaint asserts independent Claim 1 Compl. ¶63
• The essential elements of Claim 1 are:
  • acquiring a series of images at times t1 . . . tn;
  • extracting time-activity information for individual voxels;
  • prioritizing voxels for phase analysis, and assigning weighting factors;
  • applying frequency filter to voxel time-activity curves;
  • using prioritization, combining voxel time-activity information into a time varying object motion function;
  • using the time varying object motion function for the mapping of image data to corresponding motion phases; and
  • using the mapping of image data to corresponding motion phases to generate at least one motion corrected image.
• The complaint reserves the right to assert other claims Compl. ¶64

U.S. Patent No. 10,117,625 - Methods and Systems for Retrospective Internal Gating

• Patent Identification: U.S. Patent No. 10,117,625, issued November 6, 2018.

The Invention Explained

• Problem Addressed: As a continuation of the '431 Patent, the '625 Patent addresses the same problem of image degradation caused by patient motion during image acquisition, which limits diagnostic accuracy '625 Patent, col. 1:9-17
• The Patented Solution: The '625 Patent similarly describes a method for deriving motion information directly from the image data itself '625 Patent, abstract The core concept involves acquiring a series of images, extracting time-based signal information from voxels, generating a "time varying object motion function" from this information, and then using that function to determine the motion phase and generate a corrected series of images '625 Patent, col. 3:13-41
• Technical Importance: This patent continues the development of a hardware-free, automated approach to motion correction in medical imaging, aiming to improve image quality and diagnostic confidence Compl. ¶19

Key Claims at a Glance

• The complaint asserts independent Claim 1 Compl. ¶73
• The essential elements of Claim 1 are:
  • acquiring a series of images at times t1 . . . tn including a moving object;
  • extracting time-activity information for voxels of the images;
  • generating a time varying object motion function based on the extracted time-activity information for the voxels of the images;
  • determining, using the time varying object motion function, phase information for motion of the moving object; and
  • generating an updated series of images correcting for the motion of the moving object using the determined phase information for motion of the moving object.
• The complaint reserves the right to assert other claims Compl. ¶74

U.S. Patent No. 10,448,903 - Methods and Systems for Retrospective Internal Gating

• Patent Identification: U.S. Patent No. 10,448,903, issued October 22, 2019 Compl. ¶22
• Technology Synopsis: This patent, part of the same family, is also directed to methods for correcting motion artifacts in medical imaging without external hardware Compl. ¶16 It claims a method of acquiring a series of images, extracting time-activity information from voxels, determining motion phase information from that data, and generating a corrected image series '903 Patent, Claim 1
• Asserted Claims: Independent Claim 1 Compl. ¶83
• Accused Features: The complaint alleges that the entire MotionFree system, by acquiring PET data and using it to derive a motion signal for gating, practices the claimed method Compl. ¶84

U.S. Patent No. 10,863,950 - Methods and Systems for Retrospective Internal Gating

• Patent Identification: U.S. Patent No. 10,863,950, issued December 15, 2020 Compl. ¶23
• Technology Synopsis: Continuing the same inventive concept, this patent claims a method of using acquired image data to generate a "time varying object motion function," which is then used to determine the phase of motion and generate a corrected image '950 Patent, Claim 1 The language shifts from "voxels" to "a plurality of arrays derived from the image data," which may be a point of construction.
• Asserted Claims: Independent Claim 1 Compl. ¶93
• Accused Features: The MotionFree system's process of analyzing acquired PET data to create and apply a motion correction function is alleged to infringe Compl. ¶94

U.S. Patent No. 11,369,322 - Methods and Systems for Retrospective Internal Gating

• Patent Identification: U.S. Patent No. 11,369,322, issued June 28, 2022 Compl. ¶24
• Technology Synopsis: This patent claims a similar method for retrospective internal gating by acquiring image data, extracting information from arrays derived from that data, generating a motion function, determining phase, and generating a corrected image '322 Patent, Claim 1 The claims are nearly identical to those in the '950 Patent.
• Asserted Claims: Independent Claim 1 Compl. ¶103
• Accused Features: The overall functionality of the MotionFree system is accused of practicing the claimed method Compl. ¶104

U.S. Patent No. 12,171,593 - Methods and Systems for Retrospective Internal Gating

• Patent Identification: U.S. Patent No. 12,171,593, issued December 24, 2024 Compl. ¶25
• Technology Synopsis: The most recent patent in the asserted family claims a method for correcting for "periodic motion" by acquiring a "chronologically ordered image set with temporally cyclical signals," extracting information, generating a motion function, determining phase, and generating a corrected image '593 Patent, Claim 1 The language emphasizes the "periodic" and "cyclical" nature of the motion being corrected.
• Asserted Claims: Independent Claim 1 Compl. ¶113
• Accused Features: The MotionFree system's method of detecting and correcting for respiratory motion is alleged to infringe Compl. ¶114

III. The Accused Instrumentality

Product Identification

• The accused instrumentalities are Defendants' "MotionFree" software (also marketed as "PET Digital Gating") and the PET/CT imaging systems that incorporate it, including the Discovery IQ Gen 2, Discovery MI Gen 2, and Omni Legend product lines Compl. ¶¶2 Compl. ¶26 Compl. ¶46

Functionality and Market Context

• The complaint alleges, based on Defendants' own technical and regulatory documents, that MotionFree is designed to "detect and characterize respiratory motion using acquired PET coincidence data without the use of an external gating device" Compl. ¶27 The described algorithm involves several steps: acquiring PET data ("list-mode data"); generating "reduced-dimension sinogram data"; applying principal components analysis (PCA) and Fast Fourier Transform (FFT) to derive a motion signal; and then using that signal to generate motion-gated images Compl. ¶¶27-30 This process can be applied "retrospectively on any previously acquired scan" Compl. ¶31
• Defendants market MotionFree as "the first-ever respiratory motion management solution that eliminates the need for a gating device" and claim it provides significant improvements in quantitative accuracy and lesion measurement Compl. ¶44 The complaint includes a marketing image demonstrating the improved clarity of an image processed with MotionFree compared to a "Conventional Static" image Compl. ¶49

IV. Analysis of Infringement Allegations

'431 Patent Infringement Allegations

Claim Element (from Independent Claim 1)	Alleged Infringing Functionality	Complaint Citation	Patent Citation
acquiring a series of images at times t1 . . . tn;	The accused products acquire PET coincidence events over time and use them to generate a plurality of sinograms, which the complaint alleges are a "series of images."	¶64	col. 3:15-18
extracting time-activity information for individual voxels;	The accused products analyze the acquired PET list-mode data using principal components analysis (PCA) to compute spatial-temporal variations, which is alleged to be the extraction of time-activity information.	¶64	col. 3:26-30
prioritizing voxels for phase analysis, and assigning weighting factors;	The accused products allegedly use PCA to identify the principal components of motion and FFT to determine the strength of the respiratory signal, which the complaint maps to the claimed prioritizing and weighting steps.	¶64	col. 3:36-48
applying frequency filter to voxel time-activity curves;	The accused products allegedly apply an FFT on the derived signals and calculate values within a specific respiratory frequency range (0.1-0.4 Hz), which is alleged to be a frequency filter.	¶64	col. 3:49-54
using prioritization, combining voxel time-activity information into a time varying object motion function;	The accused products allegedly use the output of the PCA and FFT analysis to derive a 1D signal representing respiratory motion, which the complaint equates to the claimed motion function.	¶64	col. 4:1-8
using the time varying object motion function for the mapping of image data to corresponding motion phases; and	The derived respiratory waveform is allegedly used to create "per-cycle triggers," which enable the sorting of image data into different respiratory gates or bins.	¶64	col. 4:35-41
using the mapping of image data to corresponding motion phases to generate at least one motion corrected image.	The data sorted into gates is allegedly used to generate a motion-corrected image, such as a "quiescent-phase image" or a "4D Gated PET data" set.	¶64	col. 4:35-41

'625 Patent Infringement Allegations

Claim Element (from Independent Claim 1)	Alleged Infringing Functionality	Complaint Citation	Patent Citation
acquiring a series of images at times t1 . . . tn including a moving object;	The accused products acquire a time-series of PET coincidence data of a patient, which contains motion from respiration, and generate a series of sinograms from this data.	¶74	col. 3:15-20
extracting time-activity information for voxels of the images;	The accused products analyze the acquired data, allegedly using PCA, to compute the spatial-temporal variation of the PET list data, which is alleged to constitute extracting time-activity information.	¶74	col. 3:26-30
generating a time varying object motion function based on the extracted time-activity information for the voxels of the images;	The accused products allegedly use PCA and FFT analysis on the extracted data to generate a 1D signal that represents the patient's respiratory motion over time.	¶74	col. 4:1-13
determining, using the time varying object motion function, phase information for motion of the moving object; and	The generated motion function is allegedly used to derive triggers that correspond to phases of the respiratory cycle, enabling subsequent data processing by gate or phase.	¶74	col. 4:35-41
generating an updated series of images correcting for the motion of the moving object using the determined phase information for motion of the moving object.	The image data, having been sorted into motion phases using the derived triggers, is allegedly used to create a motion-corrected image volume. The complaint includes a visual from Defendant's webpage showing a "With motion correction" image with visibly reduced blurring compared to an uncorrected image Compl. ¶57	¶74	col. 4:35-41

Identified Points of Contention

• Scope Questions: A primary question for the court will be one of scope: does the term "extracting time-activity information for individual voxels" as used in the patents read on the accused products' alleged method of generating "reduced-dimension sinogram data" and applying PCA? The defense may argue that sinograms and PCA are fundamentally different from tracking individual voxel values as depicted in the patents' figures.
• Technical Questions: The complaint's infringement theory relies heavily on mapping the specific steps of the accused PCA-based algorithm to the steps recited in the claims. This raises the technical question of whether these steps are truly equivalent. For example, does the accused product's use of PCA to identify eigenvectors with the largest eigenvalues perform the same function as the claimed steps of "prioritizing voxels" and "assigning weighting factors," or is this a legally and technically distinct process?

V. Key Claim Terms for Construction

The Term: "time-activity information for... voxels"

• (from '625 Patent, Claim 1)
• Context and Importance: This term defines the fundamental data input for the patented method. The infringement dispute will likely center on whether the "PET list-mode data" and "sinogram data" allegedly used by the accused product Compl. ¶¶28, 30 qualify as "time-activity information for... voxels." Practitioners may focus on this term because the accused products' use of Principal Component Analysis (PCA) on sinograms may be argued as operating on a transformed representation of the data, not on the direct signal-over-time of individual voxels.
• Intrinsic Evidence for Interpretation:
  • Evidence for a Broader Interpretation: The specification describes this information as the "values v1 ... vn ... of an individual voxel ... in the successive images i1 ... in" '625 Patent, col. 3:26-30 Plaintiff may argue this is a general description and that any data format that captures the signal change of volumetric elements over time, including sinograms which are derived from that data, should be covered.
  • Evidence for a Narrower Interpretation: The patent figures explicitly depict a "Voxel time-activity" curve as a simple 2D plot of a single voxel's value against time '625 Patent, FIG. 2 Defendants may argue this figure, combined with the plain language, defines the term as requiring a direct, untransformed time-series for discrete voxels, which would exclude the more complex, line-integral-based "sinogram data" allegedly processed by the accused system.

The Term: "generating a time varying object motion function"

• (from '625 Patent, Claim 1)
• Context and Importance: This term describes the core output of the data processing stage, which is then used for motion correction. Its definition is critical because the complaint alleges that the "1D signal" derived from PCA in the accused system is this "motion function" Compl. ¶29 The construction will determine if the specific mathematical technique of PCA is equivalent to the method described in the patent.
• Intrinsic Evidence for Interpretation:
  • Evidence for a Broader Interpretation: The patent abstract states the goal is to use "individual voxel signal fluctuations... in combination to create usable respiratory phase information." Plaintiff could argue that "generating a time varying object motion function" is a broad, functional description of this process, and that PCA is merely one way to perform the "combination."
  • Evidence for a Narrower Interpretation: The detailed description explains the function is generated by serially combining filtered voxel curves, where each new voxel is tested and added if it improves the overall signal '625 Patent, FIG. 3 '625 Patent, col. 4:1-13 Defendants may argue this describes a specific iterative, constructive algorithm that is fundamentally different from PCA, which is a matrix decomposition technique that finds orthogonal bases of variation in a dataset.

VI. Other Allegations

• Indirect Infringement: The complaint alleges Defendants actively induce infringement by marketing MotionFree to healthcare providers and providing instructions on how to use the system in an infringing manner Compl. ¶¶26, 66, 76 It also alleges contributory infringement, stating that MotionFree is especially made for practicing the patented methods and is not a staple article of commerce with substantial non-infringing uses Compl. ¶¶26, 67, 77
• Willful Infringement: The complaint alleges willful infringement based on Defendants' alleged pre-suit knowledge of the patents. It specifically pleads that Defendant GE knew of the first three asserted patents since at least November 6, 2019, and that Defendant GE HealthCare had notice of five of the six patents since at least May 30, 2023 Compl. ¶¶58-60

VII. Analyst's Conclusion: Key Questions for the Case

• A central issue will be one of technical equivalence and claim construction: can the patented method, described as an iterative process of combining individual "voxel time-activity curves," be construed to cover the accused product's algorithm, which uses a matrix-based technique (Principal Component Analysis) on "sinogram data"? The court's interpretation of terms like "time-activity information" and "generating a... motion function" will be dispositive.
• A key evidentiary question will be one of functional distinction: does the accused MotionFree system, which allegedly employs PCA to holistically identify dominant motion vectors in a dataset, perform a series of steps that are functionally and sequentially equivalent to the patent's more granular, claimed method of prioritizing, filtering, and combining individual voxel signals?
• The case also raises an implicit but critical question of validity in view of Defendants' own technology. The complaint's detailed explanation of how the accused products allegedly mirror a method described in a GE-assigned patent '132 Patent Compl. ¶¶33-43 immediately frames a potential defense argument that the asserted patents are invalid over prior art developed by the defendant itself. The patentability of the claimed method over such PCA-based approaches will likely be a focal point of the litigation.