Endoscopic Device AI Patents: Who Leads, Where Gaps Are 2026
- Filing already peaked. Activity crested at 59 families in 2021 and has run flat to declining since, with the 2022 midpoint at 21 — this is not a technology still accelerating into its growth phase.
- Image processing is the real battleground, not AI models. G06T image data processing appears in 138 of 255 records, well ahead of G06N AI-model computing at 57 — most of the contested claim space is about how images are prepared and read, not the learning architecture itself.
- Filing is US-centred but not US-exclusive. 116 of the tracked records entered through the United States, with China, the EPO and WIPO's PCT route each in the 30-36 range — a genuinely multi-jurisdiction field rather than a single-market one.
Filing growth compares 2021 (59 records) with 2024 (20) — a three-year span. 2024 is the most recent year we treat as complete: publication lags filing by roughly 18 months, so 2025 onwards are still filling in and any growth rate that ends there would understate the field. Top-5 share is the combined record count of the five largest assignees divided by all 255 records in scope (CR5), not by the ranked leaders only.
What this landscape covers
This dataset tracks 255 patent families at the intersection of endoscopic imaging hardware and computer-aided diagnosis — filings that combine endoscope or endoscopy terminology with AI-assisted detection, polyp or lesion classification, or computer-aided diagnosis claim language, classified under diagnostic/surgical (A61B1/00), neural-network computing (G06N3) and health informatics (G16H50) codes. The scope spans hardware-adjacent claims — the endoscope apparatus itself — through to software-only claims covering detection and classification pipelines that run on captured endoscopic video.
Coverage runs from 2015 through the 2026-07-31 data cut-off. Because publication typically lags filing by around eighteen months, the 2025 and 2026 counts in any trend line understate real filing activity and should be read as provisional rather than final.
Filing trends and technology composition
Two views of the same 255 families: how filing volume has moved year over year, and which IPC subclasses carry the claim density.
A peak already behind us
Filings rose from 4 in 2017 to a peak of 59 in 2021, then eased back through the 2022 midpoint of 21 and beyond — a pattern consistent with an early land-grab phase that has already run its course, rather than a field still in its growth curve. The most recent one to two years will revise upward somewhat as pending applications publish, but the shape of the curve past the 2021 peak is unlikely to reverse.
Image processing outweighs the AI model layer
A61B diagnosis-and-surgery codes anchor the corpus at 214 of 255 records, as expected given the search scope. Below that, G06T image data processing (138) dominates the technical layer, more than double G06N AI-model computing (57) and G16H healthcare informatics (51). G06V image/video recognition (49) and the smaller H04N, G06K and G06F counts round out a field where image capture, preprocessing and recognition claims outnumber claims on the learning model itself.
Shares are the percentage of the 255 records in scope. A patent can carry several IPC classes, so the shares add up to more than 100%.
Go deeper on Endoscopic Device AI and Machine Learning with Eureka
This page is one run against one query. Ask Eureka your own question about endoscopic device ai and machine learning and every answer comes back with the patent numbers behind it.
Try EurekaThe most-cited prior art
US5235510A — Computer-aided diagnosis system for medical use
A picture archiving communication system for storing and transferring digital image data across one or more hospitals, built from a modality unit, a database, a workstation display and a connecting network. The modality covers multiple diagnosis apparatuses including a film digitizer, angiography apparatus, CT scanner, MRI system, nuclear medicine apparatus, ultrasound apparatus and an electric endoscope, with an examination-ordering system tied into the network and a workstation that outputs computer-aided diagnosis results.Filed by Kabushiki Kaisha Toshiba; granted 1993-08-10. With 726 citations, it is the single most-referenced record in this corpus.
| # | Publication no. | Patent title | Citations |
|---|---|---|---|
| 1 | US5235510A | Computer-aided diagnosis system for medical use | 726 |
| 2 | US6174291B1 | Optical biopsy system and methods for tissue diagnosis | 611 |
| 3 | US20180253839A1 | A system and method for detection of suspicious tissue regions in an endoscopic procedure | 167 |
| 4 | US20180098690A1 | Endoscope apparatus and method for operating endoscope apparatus | 97 |
| 5 | US20150010878A1 | Dental demineralization detection, methods and systems | 86 |
| 6 | US20190297276A1 | Endoscopy Video Feature Enhancement Platform | 66 |
| 7 | US9345389B2 | Additional systems and methods for providing real-time anatomical guidance in a diagnostic or therapeutic pro… | 65 |
| 8 | US20190252073A1 | System and method for diagnosing gastrointestinal neoplasm | 64 |
| 9 | US20190080454A1 | Methods for polyp detection | 59 |
| 10 | WO1999045838A1 | Optical biopsy system and methods for tissue diagnosis | 58 |
Citation counts accumulate over time and favour older filings; treat this table as a map of foundational and heavily-referenced prior art, not of current commercial importance.
Each row carries its publication number; clicking a row searches Eureka by that number.
Put your own technology through the same analysis
Eureka on the web
When you want the answer in the next five minutes.
The agent works the prompt against patents and technical literature, citing every source.
Run your analysis now →MCP server & REST API
When it has to run inside your own pipeline.
Patent search, landscape analysis and assignee resolution as MCP tools. Drop them into any agent framework, or call REST directly.
Browse MCP servers →What the numbers mean for filing strategy
Four read-outs from the trend and classification data that matter for where to file next and where prior art risk concentrates.
The land-grab phase has passed
Filing volume rose steadily from 2017 through a 2021 peak, then declined toward the 2022 midpoint of 21 and has not recovered. New entrants now face a more mature claim landscape than the raw novelty of AI-assisted endoscopy might suggest.
Image pipeline claims outnumber model claims
G06T image data processing appears in well over half of all records, more than double the count for G06N AI-model computing. Freedom-to-operate risk concentrates in how endoscopic images are captured, enhanced and segmented before any classifier runs, not in the classifier architecture itself.
Foundational art predates the AI wave
The most-cited record in this corpus, US5235510A, is a 1993 picture-archiving and computer-aided diagnosis system — not an endoscopy-specific AI filing. High citation counts here mark broad, early digital-imaging infrastructure patents rather than recent lesion-classification breakthroughs.
A multi-jurisdiction filing pattern
The United States leads but does not dominate; China, the EPO and the WIPO PCT route each sit in a comparable 30-36 range, with Japan and India further behind. Clearing art in one jurisdiction alone will not clear it globally.
Eureka can read the same corpus for gaps instead of for coverage: under-claimed branches adjacent to endoscopic device ai and machine learning, with the prior art for and against each one.
Who holds the ground
Assignee activity in this corpus includes established endoscope manufacturers and a long tail of smaller research and technology entrants; recent-year momentum has slowed across the tracked assignees rather than concentrating in one leader.
Incumbent filers have gone quiet recently
Assignees including Fujifilm and Olympus show zero filings in the latest tracked year, alongside several smaller research-oriented entrants. This is consistent with the broader post-2021 filing decline rather than any single company's retreat from the space.
Collaboration is sparse
The strongest co-assignee pair in the dataset shares only a single joint filing, between an AI medical services entity and a cancer research foundation. Most families in this corpus are filed by a single assignee rather than through joint ventures or research consortia.
A broad but not deeply concentrated field
With 255 families spread across manufacturers, hospitals, research institutes and smaller technology firms, no single assignee's recent activity dominates the current picture. New entrants should expect to compete against a distributed set of prior filers rather than a single gatekeeper.
| Assignee | Recent year | YoY |
|---|---|---|
| Fujifilm Corporation | 0 | — |
| Olympus Corporation | 0 | — |
| Artificial Digestive Aid Intelligence R&D Co., Ltd. | 0 | — |
| Verily Life Sciences LLC | 0 | — |
| Penn State Research Foundation | 0 | — |
| SPECTRASCI | 0 | — |
| AI Medical Service Inc. | 0 | — |
| GE Healthcare (formerly Ashaimo/Getinge-related entity) | 0 | -100% |
Where to take this analysis
The trend and classification data point to specific next steps depending on whether the goal is freedom-to-operate, licensing or new filing strategy.
Run a freedom-to-operate check on image-pipeline claims
Given G06T claims outnumber G06N claims by more than two to one, an FTO review focused only on AI model architecture will miss most of the actual claim density in this field.
Explore claim mapping in Eureka →Watch the jurisdictions beyond the US
With China, the EPO and WIPO each carrying a comparable share of filings, a US-only prior art search will understate risk in this field materially.
Compare jurisdictions in Eureka →Reassess timing given the post-2021 decline
Filing has already passed its peak year; a new entrant should weigh whether remaining claim space is genuinely open or simply less contested because incumbents have moved on.
Model filing timing in Eureka →Common questions about this landscape
This dataset tracks 255 patent families published between 2015 and the 2026-07-31 data cut-off that combine endoscope or endoscopy terminology with AI-assisted detection, polyp or lesion classification, or computer-aided diagnosis claim language. The true figure is somewhat higher because publication typically lags filing by around eighteen months, so the most recent one to two years are still filling in. Counting families rather than raw document counts also corrects for the fact that a single invention can generate several published documents across continuations and jurisdictions.
No — filing peaked at 59 families in 2021 and has since eased back, with the 2022 midpoint at 21 and no recovery visible in the years since. This suggests the initial land-grab phase for AI-assisted endoscopy claims has already passed, even though publication lag means the very latest years will revise upward slightly. A new filer today is entering a field with an established claim base rather than a wide-open one.
A61B (diagnosis and surgery) anchors the corpus at 214 of 255 records, which is expected given the search scope. Within the technical layers beneath that, G06T (image data processing) leads at 138 records, more than double G06N (AI-model computing) at 57, meaning image capture, enhancement and segmentation claims carry more density than claims on the learning model architecture itself. G16H healthcare informatics and G06V image recognition follow at 51 and 49 respectively.
US5235510A is a 1993 Toshiba filing for a picture archiving and communication system that includes an electric endoscope as one of several imaging modalities feeding a shared database and diagnostic workstation. With 726 citations it is the most-referenced record in this corpus, but its claims cover general medical-image archiving and workstation-based diagnosis output rather than any specific AI lesion-classification method. Companies should treat it as foundational infrastructure prior art relevant to image storage and workstation display architecture, not as a direct block on modern deep-learning classification claims.
The assignee base includes established endoscope manufacturers such as Fujifilm and Olympus alongside a longer tail of hospitals, research institutes and smaller AI-focused entrants, though the recent-year momentum data shows several of the tracked assignees — including these larger manufacturers — recording zero filings in the latest tracked year. Co-filing between assignees is rare in this corpus, with the strongest co-assignee pair sharing only a single joint filing. This points to a field where individual entities file independently rather than through joint research consortia.
Research Endoscopic Device AI and Machine Learning in depth with Eureka
Go past this page: query the whole endoscopic device ai and machine learning corpus yourself, in your own scope.
Every answer comes back with patent numbers you can open.
Disclaimer. This page is generated from Patsnap Eureka data drawn from a limited snapshot of global patent and scientific-literature records, and is provided for general information and reference only.
Patent data carries inherent limitations: recent filings (typically the most recent 18–24 months) are under-counted due to standard publication lag; counts may be reported at either a patent-family or a patent-record basis and are not always directly comparable; classification, applicant-name, and citation data may contain errors, duplicates, or omissions; and the underlying search query defines and constrains the scope shown. As a result, the analysis may be incomplete or inaccurate and may not reflect the full technology landscape.
Nothing on this page constitutes an exhaustive prior-art, novelty, freedom-to-operate, or validity search, nor does it constitute legal, financial, investment, or professional advice, and it should not be relied upon as such. Any patent, commercial, or strategic decision should be verified independently and reviewed with qualified patent, legal, and domain professionals. Patsnap makes no warranties, express or implied, as to the accuracy, completeness, or fitness for any particular purpose of the information presented.
Machine translation. Assignee and organisation names originally recorded in Chinese, Japanese or Korean have been rendered into English by an AI translation step so that the tables stay readable. These renderings are best-effort and may not match a company’s registered English name; the original name is what the underlying patent record carries, and it is what any Eureka query launched from this page uses.