Rondevoo Tech LLC v. Beholdai Tech Inc

I. Executive Summary and Procedural Information

• Parties & Counsel:
  • Plaintiff: Rondevoo Technologies, LLC (California)
  • Defendant: Behold.AI Technologies, Inc. (Delaware)
  • Plaintiff’s Counsel: The Chong Law Firm, P.A.
• Case Identification: 1:20-cv-00588, D. Del., 04/29/2020
• Venue Allegations: Plaintiff alleges venue is proper in the District of Delaware because Defendant is a Delaware corporation and therefore resides in the district.
• Core Dispute: Plaintiff alleges that Defendant’s "Behold AI. imaging technology" infringes three patents related to generating special-purpose, trainable image analysis algorithms.
• Technical Context: The technology involves computer-assisted image analysis systems that can be "trained" by a user to automatically recognize, classify, and quantify specific features or entities within a digital image.
• Key Procedural History: The complaint notes that the three patents-in-suit share a common specification, which suggests that claim terms may be interpreted consistently across the patents and that arguments regarding the technology's description will be broadly applicable.

Case Timeline

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

U.S. Patent No. 7,088,854 - “Method and apparatus for generating special-purpose image analysis algorithms”

• Issued: August 8, 2006

The Invention Explained

• Problem Addressed: The patent describes the difficulty for existing computer systems to accurately and consistently classify and count different types of objects, or "entities," within complex digital images, such as histological samples for medical research (’854 Patent, col. 1:36-49). Existing systems lacked a mechanism to incorporate and automate the expertise of a skilled human analyst, leading to inconsistent results and labor-intensive manual review (’854 Patent, col. 2:1-26).
• The Patented Solution: The invention is a system that generates a specialized image analysis algorithm by learning from a user's expertise. It uses an "evolving algorithm" that receives an image, partitions the image data (e.g., pixels) into entities, and identifies them based on a user's judgment. This trained "evolving algorithm" is then stored as a "product algorithm," which can be used to automatically classify entities in subsequent images without further user input (’854 Patent, Abstract; Fig. 2).
• Technical Importance: The invention provided a reproducible, automated method for quantifying microscopic features in images, addressing a long-unmet need in fields like Alzheimer's research for accurately counting cellular plaques and tangles (Compl. ¶12).

Key Claims at a Glance

• The complaint asserts independent Claim 1 (’854 Patent, col. 32:20-34; Compl. ¶17).
• Essential elements of Claim 1 include:
  • A computer usable medium with program code configured to:
  • obtain at least one image having a plurality of chromatic data points;
  • generate an evolving algorithm that partitions the chromatic data points within the image into at least one entity identified in accordance with a user's judgment; and
  • store a first instance of the evolving algorithm as a product algorithm that enables the automatic classification of entities within a second image.
• The complaint does not explicitly reserve the right to assert dependent claims.

U.S. Patent No. 7,254,266 - “Method and apparatus for generating special-purpose image analysis algorithms”

• Issued: August 7, 2007

The Invention Explained

• Problem Addressed: As a divisional of the application leading to the ’854 Patent, this patent addresses the same problem of automating expert image analysis (’266 Patent, col. 1:36-49).
• The Patented Solution: The invention claims a specific method for creating the "product algorithm." It details a "training mode" where the system presents an initial set of identified entities to a user, obtains feedback, executes the evolving algorithm using that feedback, and then presents a second, refined set of entities for further feedback or approval. Once approved, the refined evolving algorithm is stored as the product algorithm and can be provided to a second user for application against new image data (’266 Patent, Abstract; col. 10:35-42).
• Technical Importance: This method formalized a process for capturing an expert’s iterative refinement process into a distributable software tool, potentially standardizing complex image analysis across different users and labs (Compl. ¶12).

Key Claims at a Glance

• The complaint asserts independent Claim 1 (’266 Patent, col. 31:19-42; Compl. ¶22).
• Essential elements of Claim 1 include:
  • A method for automating expert quantification of image data using a product algorithm, comprising:
  • obtaining a product algorithm via a training mode that uses iterative input to an evolving algorithm;
  • the training mode comprising: presenting a first set of entities to a user for feedback; obtaining feedback; executing the evolving algorithm with the feedback; and presenting a second set of entities for feedback;
  • obtaining approval from the user;
  • storing the evolving algorithm as a product algorithm; and
  • providing the product algorithm to a second user to apply against a second set of image data.
• The complaint does not explicitly reserve the right to assert dependent claims.

U.S. Patent No. 8,687,879 - “Method and apparatus for generating special-purpose image analysis algorithms”

• Issued: April 1, 2014

Technology Synopsis

Continuing from the same patent family, this patent claims a non-transitory computer program product for automating image quantification. The claimed software generates a "locked evolving algorithm" through an iterative training mode where a user provides feedback on sets of identified entities, which is then stored for subsequent use on other image sets (’879 Patent, Abstract).

Asserted Claims

Independent Claim 1 (Compl. ¶27).

Accused Features

The complaint alleges that Defendant's "Behold AI. imaging technology" is a computer program product that performs the claimed steps of generating a locked algorithm through a training mode (Compl. ¶¶28, 56-64).

III. The Accused Instrumentality

Product Identification

The accused instrumentality is Defendant's "Behold AI. imaging technology," referred to as the "Accused System" (Compl. ¶39).

Functionality and Market Context

The complaint alleges the Accused System "enables image analysis based on product algorithms" (Compl. ¶39). The infringement allegations assert, in a conclusory manner, that the Accused System is a computer program product that practices the claimed methods, including obtaining images, generating an evolving algorithm based on user judgment or feedback, and storing that algorithm for subsequent automatic classification (Compl. ¶¶41-45, 47-54). The complaint does not provide specific technical details about the Accused System's operation or its market context, instead incorporating by reference claim charts that were not included with the complaint filing (Compl. ¶¶40, 46, 55). No probative visual evidence provided in complaint.

IV. Analysis of Infringement Allegations

The complaint references but does not include claim chart exhibits. The following summary is based on the narrative allegations.

’854 Patent Infringement Allegations

’266 Patent Infringement Allegations

Identified Points of Contention

• Evidentiary Questions: The complaint makes conclusory allegations of infringement that refer to non-public exhibits. A central point of contention will be whether Plaintiff can produce evidence that the "Behold AI. imaging technology" actually performs the specific, multi-step training, feedback, and storage processes required by the claims.
• Technical Questions: Does the Accused System generate an "evolving algorithm" that is distinct from a final "product algorithm"? A key question will be whether the accused technology merely uses a conventional machine-learning training process on a static model or if it performs the claimed steps of generating, refining, and then separately storing a distributable algorithm as taught in the patents.

V. Key Claim Terms for Construction

"evolving algorithm"

• Context and Importance: This term is the core of the claimed invention. Its construction will determine the scope of the claims and what type of machine learning or training process constitutes infringement. Practitioners may focus on this term because the patents' shared specification describes a specific interactive process between a user and the algorithm, which may distinguish it from general machine learning models.
• Intrinsic Evidence for Interpretation:
  • Evidence for a Broader Interpretation: The claims describe the term functionally as an algorithm that "partitions" data points and is generated "in accordance with a user's judgment" (’854 Patent, col. 32:25-29). This could support a construction covering a wide range of trainable algorithms.
  • Evidence for a Narrower Interpretation: The specification describes a specific iterative feedback loop where "user input during the evaluation can modify the evolving product algorithm," which is then "locked in place to yield a first product algorithm" (’854 Patent, col. 6:22-31). This language suggests a process of manual, iterative refinement rather than a single, automated training run, potentially narrowing the term's scope.

"user's judgment" (’854 Patent) and "feedback" (’266 Patent)

• Context and Importance: These terms define the nature of the required human interaction. The dispute may turn on whether any user interaction suffices, or if it must be the type of skilled, expert input contemplated by the patent's background section, which focuses on neuropathologists and other scientists (Compl. ¶11; ’854 Patent, col. 4:58-64).
• Intrinsic Evidence for Interpretation:
  • Evidence for a Broader Interpretation: The claim language itself does not explicitly require the user to be an "expert," only that an entity be "identified in accordance with a user's judgment" (’854 Patent, col. 32:28-29) or that the system obtains "feedback from said user" (’266 Patent, col. 31:32).
  • Evidence for a Narrower Interpretation: The specification is grounded in the context of capturing the expertise of "people skilled at identifying a certain entity type," such as a "trained histologist" (’854 Patent, col. 1:64-2:2). This context may support a narrower construction requiring a higher level of substantive, corrective input than simple affirmation or data labeling.

VI. Other Allegations

Willful Infringement

The complaint alleges Defendant has had knowledge of the patents-in-suit "at least as of the service of the present Complaint" (Compl. ¶67). This allegation may support a claim for willful infringement based on conduct occurring after the lawsuit was filed, but it does not allege pre-suit knowledge of the patents.

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

• A core issue will be one of evidentiary proof: Given the complaint’s reliance on conclusory allegations and external exhibits, a key question will be what discovery reveals about the actual architecture and operation of the "Behold AI. imaging technology." The case will likely depend on whether Plaintiff can demonstrate that the accused system performs the specific, iterative training, feedback, storage, and distribution steps recited in the claims.
• A second central issue will be one of claim scope: How broadly will the court construe the term "evolving algorithm"? The dispute will likely focus on whether this term can read on modern, generalized machine-learning training processes, or if it is limited by the specification's disclosure to a more specific, multi-stage method of capturing an expert's iterative judgments to create a distinct, distributable "product algorithm."