patent.fyi
Applications  /  US 2026/0354719
United States Patent Application Publication

US 2026/0354719 A1

Published
Nov 12, 2026
Filed
Apr 26, 2026
Earliest date on the face
May 7, 2025
Classification
A61B 5/14555
First published here
Nov 19, 2026

Generated automatically. Not reviewed by a person. Not legal advice or an opinion.

Section 1 · machine-generated

What if?

Questions about how individual parts of claim 1 might be changed. A question may describe something different from what the claim language recites, or a version of it; no view is offered on which.

1.1Sequential narrowband exposures

What ifWhat if the sensor records one wavelength range per exposure in a timed sequence?
Claim language“simultaneously capturing”

Four narrowband light sources centered at 535, 560, 580, and 600 nanometers illuminate the fundus in turn. A monochrome sensor records one frame per illumination pulse, and the processor registers the frames using retinal vessel landmarks.

1.1Two wavelength ranges

What ifWhat if the sensor records exactly two wavelength ranges in a single exposure?
Claim language“more than three wavelength ranges”

A beam splitter and bandpass filters project fundus views centered at 570 and 600 nanometers onto separate regions of a monochrome sensor. The processor compares vessel and adjacent tissue intensities in each view and converts the ratio of the two optical densities to an oxygen saturation estimate using calibration data.

1.2Spectra sampled along vessels

What ifWhat if the processor returns spectral samples along retinal vessels?
Claim language“an image of the fundus for each wavelength range”

The circuit locates vessel centerlines in a broadband preview and maps them to raw sensor coordinates. It averages wavelength-tagged pixels across each vessel and the adjacent tissue, then passes paired reflectance vectors directly to the determination circuit.

1.2Single spectral ratio map

What ifWhat if the processor produces one spectral ratio map from the raw sensor samples?
Claim language“an image of the fundus for each wavelength range”

Each repeating filter tile contains four spectral channels, and the processor applies dark and gain corrections to the four measurements in each tile. It divides the sum of two oxygen-sensitive channels by the sum of two reference channels and writes one map value per tile for calibrated analysis.

1.3Manual reflectance chart reading

What ifWhat if an operator estimates oxygen saturation from the wavelength images using a printed calibration chart?
Claim language“a determination processing circuit”

The display shows registered wavelength images with calibrated reflectance scales. The operator reads paired vessel and background values from the 570 and 600 nanometer images and uses a printed chart indexed by those reflectance pairs to record an oxygen saturation estimate.

Section 2 · machine-generated

Variations

Further arrangements of the same subject matter, described here as of the publication date at the foot of this page.

Other field of use

Fundus screening in veterinary clinics

A veterinary fundus camera uses an adjustable objective relay and a single image sensor to capture reflected light from the feline fundus in four wavelength bands during one exposure. A four-channel optical splitter projects the filtered views onto separate sensor regions. The image processor registers the four views and restricts analysis to the reflective tapetal region. The determination processor measures the area fraction of pigmented patches using calibrated band ratios and a reference range for feline eyes.

Component substitution

Filter mosaic on a monochrome sensor

A monochrome global shutter sensor with a repeating two-by-two interference filter mosaic replaces a broadband image sensor. Each tile has one filter per band, and all four pixel classes integrate during the same exposure. The image processor applies dark and flat-field corrections and interpolates each pixel class onto a common image grid. The determination processor combines the resulting fundus images to calculate relative macular pigment density from blue and green reflectance.

Parameter or scale variation

Six bands on one sensor

A six-channel beam splitter sends light from the same fundus field through filters centered at 500, 530, 560, 580, 610, and 650 nanometers, each with a 10-nanometer bandwidth. Relay lenses project six 1024-by-1024-pixel views in two rows of three onto a single global shutter sensor. The image processor extracts all six views from each exposure and registers them using geometric calibration maps. The determination processor fits a two-component hemoglobin absorption model to vessel reflectance relative to adjacent tissue to estimate retinal oxygen saturation.

Combination with existing systems

Fundus imaging with retinal tomography

A shared eye-facing objective couples a four-band snapshot fundus camera to a near-infrared optical coherence tomography scanner through a dichroic beam splitter. The fundus path projects four filtered views onto one sensor in a single exposure, and the tomography path scans the same retinal field to produce depth-resolved images. Software registers the fundus images to a projection of the tomography volume using vessel branch points. The determination processor reports the area of reduced short-wavelength reflectance within the macular region located from the tomographic foveal profile.

Automated or data-driven variant

Automated retinal vessel caliber measurement

A four-channel optical relay projects four filtered fundus views onto separate regions of a single sensor during one exposure. The image processor corrects the four views for geometric distortion, registers them, and passes them as separate input channels to a segmentation model trained on annotated fundus images. The model identifies the optic disc and classifies retinal vessels as arteries or veins. The determination processor measures vessel widths perpendicular to their centerlines in a ring around the disc and computes the ratio of median arterial width to median venous width.

Automatic frame selection for fundus measurements

A snapshot fundus camera records a burst of eight frames at 30 frames per second, with four wavelength bands captured on one sensor in each exposure. The image processor reconstructs the four band images per frame and scores the target region for sharpness, saturated pixels, and coverage. It aligns the three highest-scoring frames by a transform shared across bands within each frame and takes a pixelwise median for each band. The determination processor measures the optic disc rim-to-cup area ratio from the resulting images.

Specific version of a part

Four filtered sensor quadrants

A relay with beam splitters forms four images of the same fundus field on four quadrants of a monochrome global shutter sensor. Each optical branch contains a bandpass filter with a different center wavelength, and all quadrants integrate during the same exposure.

Six band filter mosaic

A repeating three by two filter tile places six wavelength-selective filters over a monochrome pixel array. Broadband illumination supplies all six bands, and a global shutter integrates the six pixel groups over the same interval.

Bandwise mosaic reconstruction

The circuit applies a dark offset and a flat field gain to each pixel, then groups corrected samples by their filter position in the repeating tile. Bilinear interpolation within each group yields a fundus image for each wavelength range on a common coordinate grid.

Hemoglobin fit across bands

The circuit computes optical density for each band as the natural logarithm of adjacent tissue intensity divided by vessel intensity. A constrained least squares fit uses band-averaged hemoglobin absorption coefficients and a baseline offset, then reports the oxygenated contribution divided by the sum of the oxygenated and deoxygenated contributions.

Section 3 · quoted from USPTO records

Other work

Passages reproduced word for word from two earlier-published U.S. patent documents, placed beside the language of claim 1. No characterization is offered. Each passage is marked A or B for the document it comes from; ¶ numbers are the paragraph numbers in the USPTO’s published full text of that document.

Document AUS 2013/0107002 A1
Published
May 2, 2013
Filed
Aug 8, 2012
Class
H04N 13/02
Document BUS 2023/0015951 A1
Published
Jan 19, 2023
Filed
Jun 27, 2022
Class
A61B 3/1208
Claim 1
Document A
Document B
1.1an imaging unit including an image sensor, the imaging unit simultaneously capturing, by the image sensor, light components in more than three wavelength ranges contained in light reflected from a fundus of the subject to generate image information used to determine the biometric information of the subject;
The imaging apparatus includes: a multiband image sensor with a light receiving surface which includes two-dimensional arrays of at least three visible light pixels and an infrared pixels, the visible light pixels having sensitivity in different visible light wavelength ranges, the infrared pixels having sensitivity in the infrared wavelength range; a coded aperture which is disposed on the light receiving surface side of the multiband image sensor and serves as an aperture i
A · US 2013/0107002 A1 · abstract
1.2an image processing circuit that generates an image of the fundus for each wavelength range from the image information; and
Further, with reference to, as a reference image, an interpolated image obtained from the image data of the second pixel which is blurred through a coded aperture, image data obtained through other pixels, namely, the first and third to sixth pixels (visible light pixels) is subjected to demosaicing processing, so as to generate a 5-band full color image (Step S503).
A · US 2013/0107002 A1 · ¶ 78
In some further examples, a plurality of spectral bands (e.g., red, green, blue, etc.) are captured all in one image frame, and are then separated in post processing performed by the image processor 106.
B · US 2023/0015951 A1 · ¶ 64
1.3a determination processing circuit that determines the biometric information of the subject based on the image of the fundus for each wavelength range.
Another aspect relates to a method of eye disease screening, comprising: flooding an eye fundus with illumination in one or more predetermined spectral bands; capturing at least one multispectral fundus image; analyzing the at least one multispectral fundus image for identification of one or more artifacts symptomatic of an eye disease; and providing an output based on the identification of one or more artifacts in the at least one multispectral fundus image, wherein the outp
B · US 2023/0015951 A1 · ¶ 5

An empty cell means only that no passage was reproduced there.

Statements in these documents
It is difficult to obtain data, in particular, depth map data satisfying these conditions, from a single-shot image, and therefore, there has been disclosed a method of taking a plurality of images of the same subject using different coded apertures or with a plurality of different cameras (see, for example, JP 2008-542863 A).
A · US 2013/0107002 A1 · ¶ 6
The use of this technology makes it possible to perform image processings including processing for generating an all-in-focus image by extending the depth of field after photographing and refocusing processing of changing the focus position.
A · US 2013/0107002 A1 · ¶ 3
Drusen may sometimes go undetected in color fundus images.
B · US 2023/0015951 A1 · ¶ 2
In one possible configuration, the fundus imager generates multispectral images that allow for extraction of details such as drusen and other artifacts that are not detectable in color fundus images.
B · US 2023/0015951 A1 · ¶ 3
Cite aspatent.fyi, US 2026/0354719 A1, claim 1, first published 2026-11-19T02:47:19-05:00, https://patent.fyi/US20260354719/, SHA-256 e707f05145884a2fcd28b9efab12400e01d79310014ee087550e4dfba9da1a8d
First published
Permanent addresshttps://patent.fyi/US20260354719/
Content fingerprint (SHA-256)e707f05145884a2fcd28b9efab12400e01d79310014ee087550e4dfba9da1a8d
Methodv2026-11-18.2 · automated, not reviewed by a person
Correctionscorrections@patent.fyi