DCT
8:25-cv-02329
Artificial Intelligence Imaging Association Inc v. Mvision Ai Inc
I. Executive Summary and Procedural Information
- Parties & Counsel:
- Plaintiff: Artificial Intelligence Imaging Association, Inc. (Florida)
- Defendant: Mvision AI, Inc. (Finland)
- Plaintiff’s Counsel: The Strategic Legal Group, PLLC
- Case Identification: 8:25-cv-02329, S.D. Fla., 09/03/2025
- Venue Allegations: Plaintiff alleges venue is proper because Defendant has committed acts of infringement in the district, including selling its GBS Contour+ software to Florida healthcare institutions. Plaintiff also suggests venue may be proper based on Defendant being a foreign corporation.
- Core Dispute: Plaintiff alleges that Defendant’s GBS Contour+ software, used for radiotherapy planning, infringes a patent related to generating synthetic image data for training machine learning models.
- Technical Context: The dispute is in the field of artificial intelligence for medical imaging, where generating large, high-quality datasets is critical for training accurate diagnostic and planning models.
- Key Procedural History: Plaintiff alleges it sent Defendant a formal demand letter on June 9, 2025, providing pre-suit notice of the asserted patent and the infringement allegations.
Case Timeline
| Date | Event |
|---|---|
| 2019-04-25 | ’272 Patent Priority Date |
| 2022-02-22 | U.S. Patent No. 11,257,272 Issues |
| 2024-01-01 | Defendant allegedly sold GBS Contour+ to Florida hospitals (date inferred from "in 2024") |
| 2025-05-15 | Date of technical/marketing materials reviewed by Plaintiff |
| 2025-06-09 | Plaintiff sends formal demand letter to Defendant |
| 2025-09-03 | Complaint Filed |
II. Technology and Patent(s)-in-Suit Analysis
U.S. Patent No. 11,257,272 - “Generating Synthetic Image Data for Machine Learning”
The Invention Explained
- Problem Addressed: The patent addresses the challenge of acquiring the vast amounts of specialized and varied image data required to effectively train machine learning models for computer vision tasks. The patent notes that manually capturing and annotating real-world images, especially stereo images or those with additional data channels like depth information, is expensive, time-consuming, and often results in limited datasets. (’272 Patent, col. 2:19-38).
- The Patented Solution: The invention provides systems and methods to automatically generate large, diverse sets of synthetic image data. The core concept, illustrated in system diagrams like Figure 1, involves programmatically assembling a virtual "image scene" from component parts, such as a background image database, a 3D object database, and a library of textures. A "virtual camera," with parameters mimicking a real-world camera, is then placed within this scene to capture a synthetic image. This process can be repeated with varying parameters to quickly generate thousands of unique images and corresponding data like depth maps. (’272 Patent, col. 3:22-44, Abstract).
- Technical Importance: This automated approach is intended to overcome the data-sourcing bottleneck in machine learning development by providing a scalable method for creating customized, application-specific training datasets. (’272 Patent, col. 2:39-51).
Key Claims at a Glance
- The complaint asserts infringement of at least Claim 1. (Compl. ¶¶11, 14, 15).
- Independent Claim 1:
- Receiving a database of background images, an object database including 3D models, a library of texture materials, and a library of camera setting files.
- Constructing a synthetic image scene by selecting a background image, selecting a 3D model, arranging them in foreground/background portions, and covering the 3D model with a texture material.
- Placing a virtual camera in the foreground of the scene, where the camera captures a view defined by a camera settings file.
- Rendering projection coordinates from the image plane as pixel coordinates of a synthetic image.
- Appending the synthetic image to a training dataset for a machine learning system.
III. The Accused Instrumentality
Product Identification
- Defendant’s GBS Contour+ software. (Compl. ¶1).
Functionality and Market Context
- The complaint describes GBS Contour+ as a "comprehensive radiation oncology solution that employs artificial intelligence for automated contouring and treatment planning optimization." (Compl. ¶13).
- Plaintiff alleges, based on technical documentation and marketing materials, that the software "incorporates synthetic data generation and machine learning training methods" that practice the steps of the asserted patent. (Compl. ¶13). The alleged purpose of this functionality is to generate synthetic CT data for training purposes using inputs like MR data and anatomical models. (Compl. ¶18).
- The software is allegedly marketed and sold to healthcare providers in the United States, including specific hospitals in Florida. (Compl. ¶¶1, 3).
IV. Analysis of Infringement Allegations
No probative visual evidence provided in complaint.
’272 Patent Infringement Allegations
| Claim Element (from Independent Claim 1) | Alleged Infringing Functionality | Complaint Citation | Patent Citation |
|---|---|---|---|
| receiving...a database of background images, an object database including 3D models... | The GBS Contour+ software allegedly receives and processes MR image data and medical imaging libraries as training input, which are analogous to background imaging data and 3D anatomical models. | ¶14-15 | col. 16:21-25 |
| constructing a synthetic image scene...selecting a background image...selecting a 3D model...and covering the 3D model with a texture material | The software is alleged to construct synthetic CT image scenes using 3D modeling and rendering techniques, which includes applying texture or density patterns. | ¶14-15 | col. 15:35-40 |
| placing a virtual camera in the foreground portion of the synthetic image scene, the virtual camera capturing the camera view...defined by a camera settings file... | The software allegedly applies "virtual imaging parameters that function similarly to camera settings" to create the synthetic data. | ¶15, ¶19 | col. 18:6-21 |
| rendering projection coordinates included in the image plane as pixel coordinates of a synthetic image | The software is alleged to render the generated scenes as "pixel-based outputs." | ¶15 | col. 19:21-36 |
| appending the synthetic image to a training dataset...used for training a machine learning system... | The software allegedly "compiles the resulting images into training datasets used to develop or improve deep learning models." | ¶15 | col. 48:61-65 |
- Identified Points of Contention:
- Scope Questions: A central dispute may arise over whether the patent's more general computer vision terminology can be read onto the specific medical imaging context of the accused product. For instance, does "receiving...a database of background images" as described in the patent encompass the accused software's alleged use of "MR image data"? Similarly, does a "3D anatomical model" from a "medical imaging library" constitute a "3D model" from an "object database" as claimed?
- Technical Questions: The complaint's allegations are based on "publicly available descriptions" and "information and belief." (Compl. ¶15). A key question for discovery will be whether the GBS Contour+ software's internal architecture actually performs the discrete, sequential steps recited in Claim 1 (e.g., constructing a scene, placing a virtual camera, rendering) or if it employs a different, more integrated AI process for generating synthetic data that does not map to the claim elements.
V. Key Claim Terms for Construction
The Term: "virtual camera"
Context and Importance: This term is foundational to the claimed method. The infringement theory hinges on the allegation that the accused software's application of "virtual imaging parameters" (Compl. ¶15) satisfies this limitation. The scope of "virtual camera" will determine what level of simulation is required to infringe.
Intrinsic Evidence for Interpretation:
- Evidence for a Broader Interpretation: The specification describes the term's function broadly, stating a camera view is a "perspective for viewing an image scene provided by the direction, position, orientation, camera intrinsics, camera calibration metadata, and/or camera capture settings of a virtual camera device." (’272 Patent, col. 14:17-22). This could support an interpretation covering any system that defines a computational viewpoint using a set of parameters.
- Evidence for a Narrower Interpretation: The patent provides extensive detail on specific parameters that precisely model physical camera components, such as focal lengths, distortion centers, lens field of view, and zoom settings. (’272 Patent, Figs. 5-7; col. 38:58-col. 43:20). A defendant may argue that a "virtual camera" must incorporate this level of detailed, physics-based simulation, potentially narrowing the claim scope away from a more abstract set of imaging parameters.
The Term: "background image"
Context and Importance: Plaintiff's infringement theory appears to map the accused software's use of "MR image data" to this claim term. The viability of this theory depends on whether a medical scan can be considered a "background image" in the context of the patent.
Intrinsic Evidence for Interpretation:
- Evidence for a Broader Interpretation: The specification states that background images "may be virtual or real world images depicting cityscapes, landscapes, out of focus scenes, or other image types." (’272 Patent, col. 27:36-39). The phrase "or other image types" could be argued to be open-ended enough to include medical images.
- Evidence for a Narrower Interpretation: The patent's examples and detailed descriptions primarily relate to general-purpose computer vision, such as autonomous vehicles (col. 20:57-65) and scenes with common objects like faces, cars, and animals (col. 16:1-6). A defendant may argue that the term should be construed in light of these disclosed embodiments, limiting its scope to non-medical visual scenes.
VI. Other Allegations
- Indirect Infringement: The complaint alleges inducement based on Defendant's "detailed technical documentation, training materials, and customer support services that explicitly teach" the infringing methods. (Compl. ¶8). It alleges contributory infringement on the basis that the accused "synthetic data generation module" is a material component of the invention that is not a staple article of commerce and has no substantial non-infringing use. (Compl. ¶¶8, 19).
- Willful Infringement: Willfulness is alleged based on both pre-suit and post-suit knowledge. Post-suit knowledge is grounded in a formal demand letter sent June 9, 2025. (Compl. ¶10). Pre-suit knowledge is alleged on information and belief, based on Defendant's purported "industry monitoring, such as attendance at medical imaging conferences and review of competitor offerings." (Compl. ¶24).
VII. Analyst’s Conclusion: Key Questions for the Case
- A core issue will be one of definitional scope: can terms rooted in general computer graphics, such as "background image" and "object database," be construed to cover the specialized "MR image data" and "3D anatomical models" allegedly used in the accused radiotherapy planning software?
- A key evidentiary question will be one of operational correspondence: as discovery proceeds, what evidence will emerge to show that the GBS Contour+ software actually performs the discrete, sequential method steps of Claim 1—constructing a scene, placing a virtual camera, rendering pixels—as opposed to utilizing a different AI-driven workflow for synthetic data generation?
- A final question relates to intent: for the claims of indirect and willful infringement, what evidence beyond general industry monitoring can Plaintiff produce to establish that Defendant had the requisite knowledge of the patent and specific intent to infringe, particularly before receiving the demand letter?