US 10466104

I'll search for information on this patent number, including any litigation or CAFC docket activity.

Let me run additional targeted searches for litigation and USPTO records.

Summary of US Patent 10,466,104 (US10466104B2)

Sourcing note: The authoritative full text of this patent was supplied in the working record (fetched from Google Patents, https://patents.google.com/patent/[US10466104](/patent/US10466104)/en). I ran four live searches (patent number, CAFC 2026 docket, "10,466,104" case, and USPTO PatentCenter) and found no litigation, CAFC appeal, PTAB proceeding, or docket activity involving this specific patent number. The search engine returned only unrelated hits (other patents, an EPO opposition decision on a different Shimadzu Salmonella patent, unrelated Apple/Taction litigation, etc.). I therefore report no confirmed CAFC 2026 docket activity for 10,466,104 — with the caveat that my searches were via general web search, not a direct PACER/CAFC docket query, so I cannot rule out unindexed proceedings with certainty.

Bibliographic data

Abstract (verbatim)

"[Problem to be Solved] To select a marker peak which characterizes a difference between groups, even when the number of samples belonging to each group is small. [Solution] A peak matrix is created based on the peaks detected from mass spectra of a plurality of samples belonging to a plurality of groups (S1-S3). Each row of the peak matrix represents a peak-intensity distribution for a large number of samples at one mass-to-charge-ratio value. If there is no difference between the groups at a certain mass-to-charge-ratio value, the peak-intensity distribution at that mass-to-charge-ratio value should be a lognormal distribution (or normal distribution). Accordingly, a hypothesis test for the conformity of the peak-intensity distribution to the lognormal distribution is performed for each mass-to-charge-ratio value (S5). A mass-to-charge-ratio value at which a significant difference has been found is selected as a candidate of the marker peak (S6)."

Plain-language overview of the independent claims

Claim 1 — Apparatus (a "mass spectrometric data analyzer")
A computing device that finds a "marker peak" (a mass-spectrum peak whose expression differs between groups) from mass spectra of many samples belonging to several groups. It has two required parts:

• (a) a peak information collector — groups peak-intensity values by m/z value, for every m/z value at which a peak appears in any of the spectra; and
• (b) a significant difference determiner — for each m/z value, tests whether the peak-intensity distribution (either the raw distribution or a corrected version of it) conforms to a probability distribution defined by a predetermined parameter. If the distribution is not in conformity, the m/z value is selected as a candidate marker peak, on the theory that the samples don't all belong to one homogeneous group.

Key point: the test is distributional conformity across all samples pooled, not a per-group mean comparison (the patent contrasts this with conventional t-test/U-test/ANOVA approaches that give false positives when N per group is small).

Claim 6 — Non-transitory computer-readable medium (program product)
The same two-step method recast as software instructions: (a) a peak information collection step (group peak intensities by m/z across samples) and (b) a significant difference determination step (apply the conformity/inconformity test per m/z value and flag non-conforming m/z values as marker-peak candidates). It is the method counterpart of claim 1, claimed as a non-transitory CRM rather than as an apparatus.

Dependent claims (context, not independent):

• Claims 2–4 narrow the apparatus: claim 2 specifies the probability distribution is normal or lognormal and adds a mean-and-variance estimator plus a statistical tester performing a hypothesis test; claim 3 adds a saturation-characteristics corrector for detector output saturation; claim 4 specifies correction using an inverse function of a previously determined detector characteristics function.
• Claims 7–10 mirror claims 2–4 in the program-product form.

Technical substance worth noting

• The worked example used 272 samples from 34 microbial strains (N_G = 34, N_g = 8), MALDI-TOF MS with an exponential-amplification detector.
• A lognormal model is used (rather than ordinary normal) because of the detector's exponential amplification; a linear detector would call for a normal distribution.
• Detector saturation is modeled by equations (1) using a differentiable function S(x) with coefficients a > 0, b < 0, c > 0, and corrected via the inverse S⁻¹(x). This saturation model is attributed to Non Patent Literature 2 (S. Kabuki et al., CANGAROO-III atmospheric Cherenkov camera, 2002).
• Hypothesis testing examples use the Lilliefors test (Non Patent Literature 3, Lilliefors 1967); the specification also lists chi-square goodness-of-fit, Kolmogorov-Smirnov, and Shapiro-Wilk as alternatives.
• Reported results: p = 0.1845 (hypothesis not rejected → no group difference) for one row, and p < 0.001 (rejected → candidate marker) for another row whose distribution appeared to be a superposition of two distributions.

Uncertainties / limits on this summary

1. No confirmed litigation or 2026 CAFC docket activity for 10,466,104, per the searches above. If you need a definitive answer, a direct PACER/CAFC CM-ECF docket search (or a Docket Alarm / Lex Machina query) would be required.
2. Inventor name spelling: the patent front matter renders the first inventor as "Yosihiro Yamada"; the assignment record and some databases render it "Yoshihiro Yamada." I have not auto-corrected either form.
3. Title variance: some third-party aggregators list a "similar document" titled "Mass analysis data analyzing apparatus and mass analysis data analyzing program." That is a different document, not the title of 10,466,104; the authoritative title is as given above.
4. I did not independently re-verify every bibliographic item against USPTO PatentCenter, since that lookup step was cut off; the bibliographic table above is drawn from the authoritative full text.

I'll search for litigation involving US patent 10466104.

Let me search more targeted litigation databases.

Let me try a few more targeted sources, including the Unified Patents portal and CourtListener.

Litigation Search Result: US Patent No. 10,466,104

No litigation involving U.S. Patent No. 10,466,104 was found.

I searched litigation-specific sources (Unified Patents litigation portal, CourtListener/recap-style docket aggregators, and general web searches pairing the exact number with terms like "infringement," "lawsuit," "complaint," "docket," and "defendant"). Every query returned either unrelated patents, patent-family/publication pages, or no results. No district court complaint, PTAB proceeding, or ITC action naming this patent was identified.

What I did find (verification of patent identity)

The searches did confirm this is the correct patent and returned only prosecution/ownership data, not litigation:

The only "cited by" document surfaced was US11043281B2 (Boe Technology Group), which is a shift-register/display patent unrelated to mass spectrometry and reflects a citation, not litigation.

Caveats on completeness

• I could not directly load a Unified Patents or PACER docket page for this specific patent number within my search budget, so this conclusion rests on aggregated search results rather than a hand-verified docket query.
• This patent is a 2019-granted Shimadzu instrument/software patent. Shimadzu is an operating company, not a litigation-prone NPE, which is consistent with (but does not prove) the absence of asserted litigation.
• Note the similar-number risk: the string "10,466,104" also appears as a foreign application number (e.g., US10/466,104, a laser-marking resin case) in some results. Those are unrelated and were excluded.

Bottom line: Based on available search results, there is no known litigation involving US 10,466,104. If you need definitive confirmation, the authoritative next step would be a direct docket query on PACER (all U.S. district courts) and the USPTO PTAB proceeding search API filtered to this exact patent number — which are outside what I can fully execute here.

I'll verify the ODP result with targeted searches for any PTAB/PGR/CBM activity and any Federal Circuit appeal touching this patent.

Let me run a second round of checks against PTAB E2E / CourtListener and for any assertion activity that would typically spawn an IPR.

Proceedings overview

There are zero AIA trial proceedings on file for US 10,466,104 — no IPR, no PGR, and no CBM — as of the USPTO ODP ingest, and I found none in web search either. The bottom line for a defendant: the patent is untested but un-narrowed — all ten claims (1–10) stand exactly as issued on 2019-11-05, with the statutory presumption of validity intact and no PTAB estoppel, no canceled claims, and no Board-confirmed claims. You get neither the benefit of a "troll with dead claims" nor the warning of a "hardened patent that already survived the Board." This is a clean-slate IPR target if you choose to file one.

Verification trail (so you can retrace this yourself)

A live PTAB E2E / P-TACTS docket query (https://ptacts.uspto.gov/ptabweb/) and a CourtListener keyword query for "10,466,104" were attempted but not returned before the search budget was exhausted. I therefore report no PTAB activity per the canonical ODP list, with the explicit caveat that this rests on the ODP ingest plus general web search rather than a first-party E2E docket pull.

⚠️ Naming trap — do not conflate this patent with other "'104" patents

Searching for IPRs on a "'104 patent" in the mass-spectrometry/data-processing space is a live way to brief the wrong case. Confirmed false positives:

• U.S. Patent 9,055,104 — a network-interface patent hit by IPR2017-01393 (Intel), IPR2018-00374 (Dell), IPR2017-01714 (Cavium), and IPR2017-01405. Same last three digits pattern, totally different patent, different technology.
• U.S. Patent 9,586,504 — hit by IPR2018-01683 (Nuna Baby Essentials). Also unrelated.
• Google Patents' "Cited By" tab for 10,466,104 lists US 11,043,281 B2 (Boe Technology Group, shift register unit) — a citation-mapping artifact, not a validity signal and not a PTAB proceeding.

{Proceedings on US 10,466,104}

None. There is no proceeding number to report, so I am reporting none rather than constructing one. For completeness against the requested template: type N/A; filed N/A; status N/A; judge panel N/A (no APJ has ever touched this patent); petition grounds N/A; institution decision N/A; FWD N/A; settlement N/A; appeal N/A.

Strategic summary

Claim-level status: 1–10 all LIVE and all UNTESTED. Nothing has been canceled. Independent claim 1 (apparatus) and independent claim 6 (non-transitory CRM) are intact, as are dependents 2–5 and 7–10. The sole subject-matter narrowing of this family happened outside the PTAB: per Google Patents' family data, the EP counterpart EP 3407375 A1 is marked "not_active / Withdrawn", and the CN counterpart CN 108931570 B is "Expired – Fee Related." Neither of those facts is a US invalidity holding or an admission; do not let opposing counsel or a broker of "IPR outcome" data assert otherwise.

Estoppel landscape: none exists. § 315(e)(2) estoppel attaches only to a petitioner that obtained a final written decision. No FWD exists, so no party is estopped from anything, and every § 102/§ 103 ground based on patents and printed publications is available to you both in an IPR and under §§ 102/103 in district court. Two structural points that cut the other way:

• PGR is time-barred. PGR must be filed within 9 months of issue (35 U.S.C. § 321(c)). Issues 2019-11-05 → the window closed on or about 2020-08-05 (date computed, not independently verified). That permanently removes § 101 and § 112 (written description / enablement / indefiniteness) as PTAB grounds. If your best theory is that the "significant difference determiner" / "peak information collector" functional language is indefinite under § 112 ¶ 6, or that the claims are directed to the abstract idea of statistical hypothesis testing, that case must be made in district court or at the ITC — not in an AIA trial.
• IPR remains fully open for any petitioner not yet served with an infringement complaint, and for a served defendant until the one-year § 315(b) bar runs from service. The § 315(b) clock has apparently never started for anyone, because no IPR has ever been filed.

Pattern signals: essentially blank. No repeat petitioner, no serial petitioning, no defensive aggregator (no Unified Patents / RPX footprints on this number), and no PTAB-to-CAFC appeal activity. The prior analysis in this record likewise found no litigation, no CAFC appeal, and no district court docket involving 10,466,104. The patent is owned by Shimadzu Corporation (a large-entity operating company, 4th-year maintenance fee paid 2023-04-19), not by a monetization entity — so the absence of IPRs correlates with the absence of assertions, not with weakness. Note the practical tension: Shimadzu is far more often the accused infringer than the accuser in this space, which is exactly why its own patent portfolio here draws little PTAB fire.

What the absence does and does not mean. It does not mean the claims are strong. IPRs at the Board are statistically lethal to claims like these — a pooled-population distributional-conformity test (Lilliefors/chi-square/KS/Shapiro-Wilk applied to lognormal peak intensity, plus a detector-saturation inverse-function correction) invites obviousness combinations built from Shimadzu's own prior art and the cited Non Patent Literatures. The Board's citation list in this record already flags US 2004/0195500 A1 (Sachs, "Mass spectrometry data analysis techniques") and US 2016/0148791 A1 (DH Technologies, "Intensity Correction for TOF Data Acquisition") — the latter is squarely about detector intensity correction in TOF and is a natural § 103 primary or secondary reference against claims 3–4 and 8–9. It simply means nobody has pulled the trigger yet.

Recommended next steps

1. Confirm the negative with a first-party docket pull before you rely on it. Query PTAB E2E / P-TACTS by patent number at https://ptacts.uspto.gov/ptabweb/ and mirror it with a CourtListener keyword search for "10,466,104" at https://www.courtlistener.com/?q=%2210%2C466%2C104%22. The ODP block is authoritative as of its ingest date, and I could not reach the direct dockets within my search budget. Treat this section's "zero" as high-confidence but not first-party-verified.
2. If you are a defendant now: there is no FWD to cite and no canceled claim to point to, so there is nothing to quote for a § 282 or estoppel advantage. Do not represent to a court that any claim has been invalidated — it has not. Your defenses are § 282 invalidity on the full art universe, plus §§ 101/112 (unavailable at the PTAB and therefore not subject to § 315(e) estoppel if you litigate them in court).
3. If you are considering filing: the PGR/CBM doors are shut (PGR window closed ~2020-08-05; the CBM transitional program statutorily sunset on 2020-09-16), so scope the petition to § 102/§ 103 on patents and printed publications. Build the § 103 attack on US 2016/0148791 A1 (DH Technologies) for the saturation-correction limitations of claims 3–4/8–9, and target independent claim 1 first — killing claim 1 takes claims 2–5 with it under the dependent-claim structure, and the same is true of claim 6 for claims 7–10. Budget to the statutory clock: institution decision within 6 months of filing, FWD within 12 months of institution.
4. Do not cite the EP withdrawal or the CN lapse as invalidating art or as admissions. They are prosecution/annuity events in other jurisdictions and have no estoppel or preclusive effect on US claims 1–10.

Explicit non-fabrication note: this patent has no PTAB proceeding number, no institution decision, no FWD, no APJ panel, and no CAFC appeal — so none is reported. If any of those exist and are unindexed, the two docket links in step 1 are how you would find them, and I would revise this analysis accordingly.

I'll verify the assignment records and search for any additional recordings beyond those in the patent's legal-events log.

Assignment Record Reconstruction — US 10,466,104

Sourcing note. The US record below is drawn from the authoritative full text of the patent (Google Patents legal-events log, https://patents.google.com/patent/US10466104/en), which reproduces the USPTO Assignment record for application 15/986,306. I attempted live lookups against the USPTO Assignment Center (https://assignmentcenter.uspto.gov/) and the legacy Patent Assignment Search (https://assignment.uspto.gov/patent/index.html), plus targeted queries on the two reel/frame numbers and the application number. Those searches returned no assignment-page content — the indexes returned unrelated records. I therefore report the two recorded assignments that are set out verbatim in the patent's legal-events data and explicitly flag the one field I could not obtain (correspondent of record) rather than reconstructing it. Reel/frame numbers below are as printed in that record.

Inventors

Patterns / notes.

• All three inventors conveyed to Shimadzu. That is consistent with a normal employee/contractor obligation-to-assign regime, not with a team retention or founder-equity arrangement.
• No departure pattern is visible. The only inventor-side events on the record are the two assignment executions (2018-07-10 and 2018-07-10→2018-07-24). There is no recorded re-assignment from any inventor to a third party, no release, and no later inventor-side conveyance — so there is no evidence of inventors exiting the original assignee within 12 months of filing, and equally no evidence of a portfolio fire-sale triggered by such exits.
• Uncertainty flagged: I have a low-confidence recollection that co-inventor Hiroto Tamura may be associated with a Japanese university rather than being a full-time Shimadzu employee. I cannot verify this from the assignment record or the sources I retrieved, and I am not asserting it. His contractual obligation to assign to Shimadzu (reel 048489/0843) is documented; his day-to-day employer is not.
• Name-spelling discrepancy (carried forward from the earlier summary, not resolved): front matter prints "Yosihiro Yamada," while the assignment records and third-party databases render "Yoshihiro Yamada." No auto-correction applied.

Original assignee

Shimadzu Corporation (Kyoto, Japan) — named on the issued patent and still the current assignee per the record. There is no post-issuance transfer of any kind on the record.

• Primary line of business: analytical and measuring instruments — mass spectrometers, chromatography, spectroscopy, medical systems. Shimadzu is a large, listed operating company (not an NPE, not a holding shell).
• Did they ship a product embodying the claims? Yes, in substance. The patent is a data-analyzer/software invention directed to MALDI-TOF MS microorganism identification, and the specification itself cites Shimadzu's commercial microbial identification offering as the state of the art it improves on (Non Patent Literature 1, "AXIMA Biseibustu Doutei Shisutemu / AXIMA microorganism identification system," Shimadzu Corporation). The claimed analyzer plugs into that commercial MALDI-TOF product line. This is an operating-company instrument/software patent, not a bare licensing asset.
• Current status: Active and ongoing. Legal status Active; 4th-year maintenance fee paid 2023-04-19; anticipated expiration 2038-05-22. No bankruptcy, receivership, dissolution, merger-into-shell, or change-of-name events appear anywhere in the record.
• Family: JP 6899560 B2 (active); EP 3407375 A1 (withdrawn — the EP counterpart was not pursued to grant); CN 108931570 B (granted, since marked expired-fee-related in the family table). The US is the live member of the family of record.

Assignment timeline

Two records exist. Both are the same chain to the same assignee — the second is housekeeping, not a transfer.

• 2018-07-10 (executed) / recorded 2018-08-13 — Reel 046629 / Frame 0154

  • Conveyance: Assignment
  • Assignor: YAMADA, YOSHIHIRO (sole assignor as originally recorded)
  • Assignee: SHIMADZU CORPORATION (Kyoto, Japan)
  • Correspondent: Not retrieved. The Assignment Center entry contains a correspondent-of-record field, but I could not pull the recorded filing attorney/agent within my search budget. No correspondent finding is asserted.
  • Context: Routine pre-issuance acquisition — the inventors' rights conveyed to the employer/original assignee while the application was pending (application filed 2018-05-22; assignment executed ~7 weeks later). Not a fire-sale, not an NPE transfer.
  • Defect: the original recording named only one of three inventors.

• 2018-07-10 → 2018-07-24 (executed; signing-date range as recorded) / recorded 2019-03-04 — Reel 048489 / Frame 0843

  • Conveyance: Corrective Assignment
  • Assignors: YAMADA, YOSHIHIRO; TAMURA, HIROTO; KATO, TERUYO
  • Assignee: SHIMADZU CORPORATION (Kyoto, Japan)
  • Correspondent: Not retrieved — same limitation as above. Notably, the two recordings share the same assignee and describe one continuous obligation, so a recurring-correspondent signal cannot be assessed on this chain regardless.
  • Context: Change-of-record only. The USPTO record states this is a "CORRECTIVE ASSIGNMENT TO CORRECT THE TWO OMITTED ASSIGNOR NAMES PREVIOUSLY RECORDED AT REEL: 046629 FRAME: 0154 … ASSIGNOR(S) HEREBY CONFIRMS THE ASSIGNMENT," adding Tamura and Kato to the 2018-07-10 instrument. No change in ownership, no new consideration, no new party. This was recorded ~4 months before grant (issue date 2019-11-05).

There are no other recorded assignments. No security agreement, no license, no merger, no change of name, no release, and — critically — no transfer away from Shimadzu. The absence of any post-issuance assignment is itself the finding: the original assignee still owns the patent.

Verification link: https://assignmentcenter.uspto.gov/ (search by patent number 10466104 or application 15/986,306). Legacy equivalent: https://assignment.uspto.gov/patent/index.html.

Timeline diagram

timeline
    title Ownership of US 10466104
    2017 : JP priority application filed
    2018 : US application filed
         : Inventors assign to Shimadzu
         : Recorded reel 046629 frame 0154
    2019 : Corrective assignment recorded
         : Patent issued to Shimadzu

NPE / troll-pattern signals

Verdict

Defensive / non-asserting.

The ownership chain for US 10,466,104 is a single, unbroken acquisition by the original operating assignee: inventors → Shimadzu Corporation per reel 046629/0154 (executed 2018-07-10, recorded 2018-08-13), followed by a record-only correction adding the two omitted inventors per reel 048489/0843 (recorded 2019-03-04). There is no post-issuance assignment at all, so no NPE, no shell entity, no asserter, and no cascading chain exists on the record — the patent simply stayed with Shimadzu, which still pays maintenance fees and sells the MALDI-TOF instruments the claims support.

Definitional caveat: the menu's "Defensive / non-asserting" entry is described as a chain ending at a defensive aggregator (RPX, AST, etc.). That is not what happened here — the chain ends at an operating company. I selected this verdict because it is the closest available fit for a patent that is retained and not asserted, which the record unambiguously shows; the alternative "Operating-company assertion" requires active litigation against competitors, and none exists.

Two limits on this verdict: (a) I could not retrieve the correspondent-of-record from the Assignment Center, so signal #3 is genuinely unevaluated rather than affirmatively cleared; and (b) the conclusion rests on the patent's legal-events record and aggregated web searches rather than a hand-loaded Assignment Center page, so a direct query at https://assignmentcenter.uspto.gov/ is the appropriate confirming step.

I'll research each cited reference for US 10,466,104. Let me run targeted searches.

Let me search the remaining cited references.

Prior-Art Analysis for US 10,466,104 B2 — Cited References of Record

Step 1 — Verification of the search target

I searched for the literal number 10466104 and matched it to US 10,466,104 B2 ("Mass spectrometric data analyzer and program for analyzing mass spectrometric data," Shimadzu Corp., app. US 15/986,306, granted 2019‑11‑05). No similar-number records (e.g., the foreign application US10/466,104, or unrelated patents) were used. The reference list below is drawn from the authoritative full text in the working record (Google Patents, https://patents.google.com/patent/US10466104/en), which mirrors the USPTO "References Cited" data. Per the operating rules, I did not auto-correct any number.

Sourcing caveat: I ran live corroborating searches on the individual cited references. My search budget was exhausted before I could re-verify three of the seven references in depth (US 2016/0148791 A1, US 2018/0000963 A1, and, partially, US 2014/0129169 A1). Where a detail below is drawn only from the authoritative patent text rather than from an independent source, I say so.

Step 2 — The seven cited patent references (References Cited of record)

Non-patent literature of record (relevant to claims 2–4/7–9, not claim 1/6):

• Kabuki, S. et al., "Development of an atmospheric Cherenkov imaging camera for the CANGAROO‑III experiment," Proceedings of the Universe Viewed in Gamma‑rays, Univ. of Tokyo Workshop 2002 — source of the detector characteristics function S(x) (equations (1)) used for saturation correction.
• Lilliefors, H. W., "On the Kolmogorov‑Smirnov Test for Normality with Mean and Variance Unknown," JASA, Vol. 62, No. 318 (1967), pp. 399–402 — the normality hypothesis test applied in the Example.
• Shimadzu "AXIMA microorganism identification system" product literature; Shimadzu "Mass++ Beginners' Guide" (XP055515175, May 15, 2014); EPO communication dated Oct. 23, 2018 (counterpart EP 18173592.9).

Step 3 — Per-reference analysis and § 102 mapping

The § 102 standard requires that a single reference disclose every element of a claim, arranged as recited, to anticipate. Under that standard, none of the seven references anticipates independent claim 1 or independent claim 6 — none discloses the core limitation (b): testing, for each m/z value, whether the pooled peak‑intensity distribution conforms to a probability distribution defined by a predetermined parameter, and selecting non‑conforming m/z values as marker‑peak candidates. The references are properly background/§ 103 art. The element-by-element mapping is below.

Reference 1 — US 2004/0195500 A1 (Sachs / Merck) — the closest art; possible § 102 relevance to claim 1 in a broad reading, otherwise § 103

• Description (verified): Discloses determining the statistical significance of changes in m/z‑intensity pairs between two or more sample sets, and assigning each relationship a "rank or value" (e.g., a p‑value from a statistical test) based on both within‑sample‑set and between‑sample‑set signal distributions. Peaks are grouped by m/z identifier across spectra/sample sets (its m/z‑intensity "peak" grouping mirrors claim 1's "peak information collector"). Claim 1 of the granted US 6,906,320 recites: obtaining spectra for sample sets; determining a relationship between corresponding signals per m/z; assigning a rank/value based on within‑ and between‑set distributions as a measure of likelihood of a different analyte level.
• § 102 mapping: Potentially relevant to claim 1 elements (a) and (b) at the highest level of generality (it performs a per‑m/z statistical determination). However, its determination is a significance‑of‑difference test between predefined sample sets — the opposite logic of the present invention, which pools all samples and tests conformity to a single probability distribution (non‑conformity = candidate). Because it does not disclose the "conformity/inconformity to a probability distribution using a predetermined parameter" step, it does not anticipate claim 1 as written; it is best characterized as § 103 art. It also does not disclose the peak‑matrix (claim 5/10) or the saturation‑correction features (claims 3–4/8–9).

Reference 2 — US 2010/0299076 A1 (Shimadzu, "Mass Spectrometry System") — partial map to claim 1 element (a) only

• Description: An MSⁿ amino‑acid‑sequence estimation system. Its claim 1 recites, verbatim, a "peak information collection means" that extracts peaks from MSᵐ/MSᵐ⁺¹ spectra, collects mass information, and prepares peak lists — plus a peak‑extraction means, evaluation‑value calculation means, and sequence‑estimation means.
• § 102 mapping: The literal term "peak information collection means" overlaps claim 1's "peak information collector" and claim 5/10's "peak detector/peak‑list" concept, but the reference has no marker‑peak, no group‑difference, and no distribution‑conformity test. It cannot anticipate claims 1, 2, 5, 6, 7, or 10. It is cited as evidence of the ordinary "peak information collection" technique in mass‑spectrometry data processing.

Reference 3 — US 2010/0286927 A1 (Agilent, DDA) — no claim anticipated

• Description (verified): Data‑dependent acquisition in LC/MS/MS. Computes a purity value and a selection score for each isotopic cluster within an isolation window, rank‑orders clusters, and selects the top‑scoring precursors for MS/MS. Concerned with precursor‑ion contamination, not with sample‑group marker discovery.
• § 102 mapping: No overlap with claim 1/6 elements. The only surface similarity is "calculating a value per m/z‑associated cluster," which is not the claimed distributional conformity test. Cited as general background in mass‑spectrometry statistical scoring. No claim anticipated.

Reference 4 — JP 5750676 B2 (Shimadzu, "Cell identification device and program") — no claim anticipated (background/field art)

• Description (verified in part): Identifies cell type (species, and below species to strain level) from MALDI‑MS mass‑spectrum patterns. Uses a two‑level database: a higher‑level DB searched by a test mass list, then deletion of species‑common masses, then a lower‑level DB search on the remaining strain‑specific masses. Directly addresses the same problem space (MALDI‑TOF microorganism identification at strain level) as US 10,466,104.
• § 102 mapping: It is a database pattern‑matching / mass‑list subtraction approach; it contains no statistical distribution, no hypothesis test, and no marker‑peak selection. No claim of US 10,466,104 is anticipated. Its relevance is as § 103 "field" art confirming the conventional approach the '104 patent characterizes as requiring excessive samples.

Reference 5 — US 2014/0129169 A1 (Shimadzu, "Chromatogram data processing method and device") — no claim anticipated

• Description (verified): Corrects the time axis of a target chromatogram to a reference chromatogram using linear correction plus coarse‑to‑fine dynamic‑programming peak matching. Concerns retention‑time alignment of chromatograms/chromatograph‑mass‑spectrometry data.
• § 102 mapping: Addresses time‑axis alignment, not m/z peak‑intensity distribution analysis. No overlap with any claim. No claim anticipated.

Reference 6 — US 2016/0148791 A1 (DH Technologies, "Intensity Correction for TOF Data Acquisition") — potential § 102/§ 103 relevance to claims 3–4 and 8–9 only

• Description: Per the authoritative citation list, this reference concerns intensity correction for time‑of‑flight data acquisition. Its title and DH Technologies provenance place it squarely in the area of correcting detector/acquired intensity data — conceptually aligned with the '104 patent's saturation‑characteristics correction (claims 3–4/8–9), which corrects the peak‑intensity distribution using the inverse function S⁻¹(x) of a previously determined detector characteristics function.
• § 102 mapping: Potentially relevant to claims 4 and 9 (intensity correction via an inverse of a detector characteristics function). Whether it anticipates those claims depends on whether it discloses (i) a previously determined detector characteristics function and (ii) correction by its inverse applied to a peak‑intensity distribution before a distributional conformity test. Those latter two points are specific to the '104 patent and, if absent, this reference is § 103 art rather than § 102 art. I could not fully re‑verify this reference's disclosure within my search budget, so I flag this mapping as provisional. It plainly does not anticipate claims 1, 2, 5, 6, 7, or 10.

Reference 7 — US 2018/0000963 A1 (Helix Biopharma, "Antibody‑urease conjugates for therapeutic purposes") — no claim anticipated; probable incidental citation

• Description: Therapeutic antibody‑urease conjugates (immunotherapy). Entirely unrelated to mass‑spectrometry data analysis.
• § 102 mapping: No overlap with any claim; no claim anticipated.
• Prior‑art‑date nuance: Its publication date (2018‑01‑04) is after the '104 priority date (2017‑05‑23). Its only possible prior‑art role would be as a § 102(a)(2) reference (a U.S. application publication effectively filed before the priority date — here, priority 2015‑01‑23). Given the subject matter, that role is moot.

Step 4 — Overall § 102 conclusion

• Independent claim 1 (apparatus) and independent claim 6 (non‑transitory CRM): No single cited reference anticipates. The distinguishing limitation — per‑m/z conformity/inconformity testing of the pooled peak‑intensity distribution to a probability distribution, with non‑conforming m/z values flagged as marker‑peak candidates — is absent from all seven references. Reference 1 (Sachs) comes closest but performs a between‑group significance test, not a pooled‑distribution conformity test, and thus does not anticipate.
• Dependent claims 2 and 7 (normal/lognormal + mean‑and‑variance estimator + hypothesis test): The statistical machinery is grounded in the non‑patent literature of record — Lilliefors (1967) for the normality test — rather than in the patent references. No cited patent reference anticipates these claims.
• Dependent claims 3–4 and 8–9 (detector‑saturation correction; inverse detector characteristics function): The model is attributed to Kabuki et al. (2002) and the correction technique to the mass‑spectrometry intensity‑correction field (Reference 6, provisional). Reference 6 is the most plausible prior art here, but whether it anticipates (versus renders obvious) claims 4/9 requires confirmation of its specific disclosure.
• Dependent claim 5 and 10 (peak detector + peak‑matrix creator): Reference 2 (Shimadzu) and Reference 1 disclose related peak‑list/peak‑grouping functionality but not the claimed peak matrix feeding a distributional conformity test. No anticipation.

Net: All seven references function as background and § 103 obviousness art. Under 35 U.S.C. § 102, none of them, alone, discloses every element of any claim of US 10,466,104 as arranged in the claim.

Step 5 — Explicit uncertainties

1. Search channel: My lookups ran through Google Patents, FreePatentsOnline, and Patents‑Review/Patents‑Encyclopedia — public mirrors — rather than a native USPTO PatentCenter/Public PAIR "References Cited" query. The authoritative full text in the record is my primary source for the citation list.
2. Three references not fully re‑verified: I could not complete independent verification of the technical disclosure of US 2016/0148791 A1 and US 2018/0000963 A1 (and only partially of US 2014/0129169 A1). The descriptions for those are based on the authoritative citation list plus title/assignee; where this affects a § 102 mapping (Reference 6 → claims 4/9), I have labeled it provisional.
3. Reference 2 grant number: A corroborating source linked US 2010/0299076 A1 to U.S. Pat. No. 8,694,264; I did not independently confirm that grant number, so I do not rely on it.
4. No auto-correction applied: I preserved identifiers exactly as they appear ("JP5750676B2," "US20100299076A1," etc.), including the assignee/title spellings in the record.

I'll ground the obviousness analysis by pulling what these prior-art references actually disclose.

Let me check a few more references and the general prior-art landscape on distribution-based peak selection.

Obviousness Analysis — U.S. Patent No. 10,466,104 B2 under 35 U.S.C. § 103

Scope note. This analysis builds on the earlier sections (bibliographic data, claim summary, no-litigation finding) and does not repeat them. It uses the prior art appearing in the "Prior Art" portions of the authoritative record for US10466104B2 (Source: https://patents.google.com/patent/[US10466104](/patent/US10466104)/en): the seven examiner/third-party "Citations," the thirteen "Family Cites Families," and the five "Non-Patent Citations." I supplemented this with live searches on the most load-bearing references; where I could not verify a reference's disclosure within the search budget, I say so explicitly.

Identifiers are read literally. Two pairs of identifiers on that page are easy to conflate and are not the same documents — I do not auto-correct them:

• JP5750676B2 (Shimadzu, priority 2011‑10‑18, "Cell identification device and program") is a different document from JPS5750676B2 (listed priority 1974‑11‑27, publication 1982‑10‑28, no assignee/title). The former is the substantive prior art; the latter appears only in the "Family Cites Families" list and, on its face, is unrelated art. An invalidity contention must cite the correct one.
• US20040195500A1 (Sachs) is the pre-grant publication of US 6,906,320 B2 (Merck). Both identifiers may be cited, but they are one disclosure, not two.
• US20180000963A1 (Helix Biopharma, "Antibody-urease conjugates for therapeutic purposes," 2015‑01‑23) is not usable art for these claims; it is unrelated to mass-spectral data analysis and appears only as a citation. Its presence on the face of the patent does not make it material.

1. Legal frame and critical date

Because every reference precedes the critical date, validity turns on § 103, not § 102, for the independent claims. No single reference of record appears to disclose the full claim 1 combination, which is consistent with the examiner having allowed the case over them. The viable attack is therefore a combination attack under KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007).

2. Claim 1 / Claim 6 decomposition and mapping to the art of record

Claim 1 requires only two substantive elements: (a) a "peak information collector" that groups peak intensities by m/z across all samples, for every m/z at which a peak appears in any spectrum; and (b) a "significant difference determiner" that, per m/z, tests whether the peak-intensity distribution — raw or corrected — conforms to "a probability distribution using a predetermined parameter," and flags non-conforming m/z values as marker-peak candidates. Claim 6 is the same two steps as a non-transitory CRM.

2.1 Element (a) — grouping peak intensities by m/z across samples

Element (a) is met by Sachs alone, and redundantly by the two Shimadzu references.

2.2 Element (b) — the conformity/inconformity test (the crux)

This is the only element not squarely disclosed by the cited art, and it is where the § 103 case must be built. The building blocks are, however, all in the record:

Where the art falls short. None of the verified references appears to disclose: (i) pooling intensity values at one m/z across all samples and testing that pooled distribution for conformity, and (ii) doing so without using group membership at all, then inferring group heterogeneity from non-conformity. Sachs uses within- and between-set distributions, but its output is still a between-set difference measure assigned to a predetermined set structure. This is the gap a petitioner must close with reasoning rather than disclosure.

3. Motivation to combine (KSR rationales)

A person of ordinary skill in the art at the 2017 critical date — a chemometrician/programmer working on MALDI‑TOF microbial identification, i.e., a person with the skill set reflected by the same applicant's own US20100299076A1 (Shimadzu, statistical peak evaluation) and US20140129169A1 — would have had strong, articulated reasons to arrive at the claimed combination:

1. Known problem, known solution in a predictable art (KSR, "the design incentives and market forces"). The patent's own background frames the problem: "If there are many groups, operators will be considerably burdened with sample preparations … The task of performing a mass spectrometric analysis on each sample will also be extremely complex and time-consuming." Reducing N_g per group is the stated objective. The same problem is voiced in JP5750676B2's background (strain-level discrimination is difficult with conventional mass-list matching). The art therefore supplies both the problem and the incentive.

2. Substitution of one known statistical test for another (KSR; a "simple substitution of one known element for another"). Sachs already teaches the step of "assign[ing] each said relationship a rank or value … based on … signal distributions … a p‑value from a statistical test." Replacing (or supplementing) the per-group ANOVA/t-test of the admitted prior-art procedure with a distributional goodness-of-fit test — Lilliefors 1967, expressly designed for the case where mean and variance are unknown and must be estimated — is a substitution of a known statistical tool yielding a predictable result. Both tools output a p-value compared against a significance level α; the patent's own description confirms the output convention is identical.

3. Known phenomenon → known correction. The physical premise of the claimed test (Poisson-limited ion arrival → normal; exponential-amplification detector → lognormal) is stated in the specification as background, not as invention. One of skill in the art, choosing to fit a lognormal model to MALDI-TOF peak intensities from an MCP/secondary-electron-multiplier detector, would apply the lognormal rather than the normal model as a matter of routine design choice — with a reasonable expectation of success.

4. Applying a known technique to a known method "ready for improvement" (KSR). Sachs is expressly and repeatedly concerned with false positives: it describes "Filtering for False Positives I" and "Filtering for False Positives II," provides means for "performing one or more iterations to reduce false positives," and frames "variability in signal at a given m/z due to factors other than changes in analyte levels between sample sets [a]s noise." Using a distributional conformity test as an additional/alternative false-positive filter — including the counterintuitive inversion of flagging non-conformity rather than a group-mean difference — is the use of a known technique to improve a similar method in the same way, with predictable results.

5. Teaching of the program-product form. For claim 6, two of record's references independently disclose claiming the analysis as instructions on a computer-readable medium: Sachs claim 34 ("A computer program comprising instructions on a computer readable medium for performing steps (b) and (c) of claim 1") and US20160148791A1 / US 9,514,921 ("A computer program product … comprising a non-transitory and tangible computer-readable storage medium whose contents include a program with instructions…"). Reciting the method of claim 1 in a non-transitory CRM is therefore, on this record, essentially a formal change in claim format with no patentable weight.

4. Claim-by-claim prima facie case

5. Honest weaknesses of the § 103 case (and the contrary arguments)

A well-prepared patent owner would press the following, and each has real force:

1. No reference discloses the absence of group information. Every verified reference (Sachs, US20100299076A1, JP5750676B2) presumes a pre-existing group/set structure. Claim 1 requires no group labels at all — heterogeneity is inferred from non-conformity. A petitioner must show that discarding group labels was obvious, not merely that both group-based and distributional statistics were known. Hindsight risk under In re Kotzab / KSR's caution against "a combination of elements … in which the whole would have been obvious" being asserted only after the fact.

2. Different problems across references. Sachs addresses differential abundance between defined sample sets (proteomics, LC/MS-coupled); JP5750676B2 addresses database search/matching for cell identification. Neither frames the problem as "I have many groups and few replicates per group; use pooled distributional shape instead." The motivation argument in §3(1) is therefore largely constructed from the patent's own background — usable, but vulnerable to the charge that the problem statement was derived from the patent.

3. No reductive disclosure of "predetermined parameter." Claim 1's "probability distribution using a predetermined parameter" is broad and functional; that breadth cuts for obviousness (a wide claim is easier to meet) but also risks the owner arguing the claim covers only the specific lognormal-Q-Q-refit workflow of FIGS. 6–8. Note the specification's example is a Q-Q-plot visual/curve-fitting exercise (FIGS. 6–11), and for the second row the fitted saturation model was found inapplicable — an owner could argue the claims are tied to a specific multi-step diagnostic that the art does not suggest.

4. No objective evidence of non-obviousness in the record. The specification reports p = 0.1845 and p < 0.001 for two rows but gives no comparative data against the admitted t-test/U-test/ANOVA procedure (no ROC, no false-positive/true-positive rates, no sample-size study). The "Advantageous Effects" section asserts reduced burden, cost, and time — but as assertions, without a nexus to any claimed element. A patent owner cannot currently point to unexpected results supporting the claims; a petitioner should note this silence.

5. Verification gaps I could not close. (a) US20140129169A1's substantive disclosure was not verified. (b) CN105823883B is a Chinese-language document whose Poisson model is used for peptide-spectrum-match scoring, not group-difference detection — its probative value is corroborative only, and over-reliance on it would be an easy target. (c) The cited Kabuki et al. paper (CANGAROO‑III atmospheric Cherenkov camera) is from an unrelated field (astroparticle physics); it supplies only a saturation/response function, so it must be paired with a mass-spectrometry motivation to combine — which US20160148791A1 supplies.

6. Rerun / verification checklist

If this analysis is to be used for a validity challenge (reexam, IPR, or district-court contention), the following steps are required and are not satisfied by the search budget used here:

1. Retrieve and cite the full text of US 6,906,320 B2 (not merely the US20040195500A1 abstract) at the specific-column level, especially the "Filtering for False Positives" passages and claim 1(c)/claim 34.
2. Obtain the Lilliefors 1967 offprint and pin the "mean and variance unknown" language.
3. Confirm US 9,514,921 B2 column/line cites for the calibration-function and "property of a particular detector"/"generic … across many detectors of the same type" passages.
4. Verify US20140129169A1 (Shimadzu, chromatogram data processing) or drop it from the chart.
5. Confirm which JP document is intended: JP5750676B2 (2011 Shimadzu) versus JPS5750676B2 (1974).
6. Confirm no intervening rights/terminal disclaimer issues; grant date 2019‑11‑05, anticipated expiration 2038‑05‑22, 4th-year maintenance fee paid 2023‑04‑19 — reexamination remains available for the full term.
7. Consistent with the earlier sections, no litigation, CAFC appeal, or PTAB proceeding involving 10,466,104 has been identified; nothing in this record is a prior adjudication either way.

Bottom line: claims 1 and 6 are vulnerable to a KSR combination of Sachs (US20040195500A1 / US 6,906,320) as the peak-grouping and statistical-significance framework, Lilliefors 1967 as the conformity-test instrument, and the applicant's own admitted normal/lognormal physics — with claims 3–4/8–9 further supported by US20160148791A1 (US 9,514,921) plus the admitted Kabuki detector-response function, and the CRM format supported by express program-product claims in both Sachs and US 9,514,921. The single element no reference discloses — testing a pooled, group-agnostic peak-intensity distribution for conformity and inferring group heterogeneity from non-conformity — must be supplied by reasoned motivation rather than disclosure, and that gap, together with the cross-reference problem-statement problem in §5(2), is the patent owner's strongest defense.