Whole Slide Tiling via Thumbnail Pre-screening

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for whole slide tissue tiling are inefficient, especially at high magnification and with overlap, as they require reading pixels at high resolution, leading to slow processing times.

Innovation Solution

The proposed system performs a reduction operation on the whole slide image to generate a thumbnail, applies thresholding to identify pixels corresponding to tissue presence or absence, and selects a subset of tiles associated with tissue presence, significantly speeding up the tile extraction process.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional high-resolution pixel reading methods are used for whole slide tissue tiling, then measurement precision is maintained, but processing speed deteriorates significantly

Engineering Contradiction:
Improvetile selection accuracyVSAvoidprocessing speed
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent segments the whole slide image processing into two distinct stages: (1) a fast pre-screening stage using downscaled thumbnails to identify candidate tile regions, and (2) a precise verification stage using original high-resolution pixels only for selected tiles. This segmentation allows most of the image to be processed quickly at low resolution while maintaining high precision only where needed.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a dimensional transformation by creating downscaled thumbnail versions of the whole slide image. This creates a new dimension (scaled-down representation) that enables rapid processing and tile candidate identification, while the original high-resolution dimension is reserved only for final tile extraction and analysis, thus resolving the speed-precision tradeoff.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Reliability

If high magnification tiling is performed with overlap, then tile coverage and overlap quality are improved, but processing time increases dramatically

Engineering Contradiction:
Improvetile coverage qualityVSAvoidprocessing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent performs preliminary actions by generating downscaled thumbnails and identifying candidate tile regions before actual tile extraction. This preliminary screening eliminates the need to process every possible tile position at high resolution, thus reducing processing time while maintaining coverage quality through subsequent verification of candidate tiles.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent applies local quality by using high-resolution processing only in specific local regions (candidate tile areas identified from thumbnails) rather than uniformly across the entire slide. This allows high magnification and overlap requirements to be met locally where needed, while saving computational resources in regions already identified as non-candidates.

Inventive Principle:
Principle #3Local quality

Data Source

PatentUS20250111511A1Fast whole slide tissue tiling method
Publication Date: 2025.04.03 MEMORIAL SLOAN KETTERING CANCER CENT
  • US20250111511A1 patent drawing
  • US20250111511A1 patent drawing
  • US20250111511A1 patent drawing

AI summary

The present application relates generally to image tiling, including but not limited to systems and methods of fast whole slide tissue tiling. A computing system may identify a first dimension of a first image from which to select one or more tiles. The computing system may perform a reduction operation on the first image to generate a second dimension of a second image. The computing system may perform a smoothening operation on the second image. The computer system may identify a first set of pixels in the second image corresponding to a presence of a feature and a second set of pixels corresponding to an absence of the feature. The computing system may select, from a plurality of tiles corresponding to the first image, a subset of tiles corresponding to the first set of pixels identified from the second image.