MétaCan
Menu
← Back to cohort

Advancing Breast Cancer HER2 FISH Quality by Image Analysis.

2009· article· en· W2319533240 on OpenAlexaff
Marianne Rogers, Jean Doré, Michael Grunkin, Kenneth P. H. Pritzker

Bibliographic record

VenueCancer Research · 2009
Typearticle
Languageen
FieldMedicine
TopicRadiomics and Machine Learning in Medical Imaging
Canadian institutionsUniversity of TorontoMount Sinai Hospital
Fundersnot available
KeywordsBreast cancerMedicinePathologyStage (stratigraphy)CancerBiologyInternal medicine

Abstract

fetched live from OpenAlex

Abstract Background: A critically important factor in the correlation of Her2 amplification and clinical outcome in breast cancer is the quality of histologic Her2 amplification detection. FISH is the “gold standard” for HER2 amplification detection but presently most labs use manual methods which are labour intensive and restricted by analysis time to small samples (60 cells or less if amplification status appeared clear to the observer).An ideal FISH detection system should be rapid (less than 30 minutes per sample), identify HER2/neu and CEP17 copy number objectively in each cell, count sufficient cells to be representative statistically, count Her2/neu in cell clusters objectively, and have a permanent record of each cell counted. Our objective is to develop an image analysis system for Her2/neu FISH that is superior to current manual and automated methods in all the above criteria.Materials and Methods: The Visiopharm Integrator System software and a Leica DM6000B fluorescence microscope equipped with a Prior 8-slide capacity motorized stage (Mac 5000 ps system stage control) and Hamamatsu camera (Model CA4742-80-12AG) were used for image analysis. Forty breast cancer biopsies, 15 core biopsies included, were examined for Her2 FISH previously assessed manually on the same slide. For manual detection, at least 3 representative fields were selected by the observer. For image analysis, unbiased tumour sampling, typically 16 fields, was assessed within a region of interest previously identified. Methods were compared by ASCO/CAP amplification criteria and by assessment of technical time, cells counted and objectivity of counting criteria.Results:Comparison manual vs image analysisMethodManualImage AnalysisTechnical time: minutes, average, range50 (45-135)25 (20-40)Cells counted: average, range58 (20-217)238 (18-1151)HER2 in clusterssubjectiveobjective and reproducibleASCO/CAP amplified/equivocal19/217/3ASCO/CAP not amplified/uninterpretable18/120/0 Seven discordant cases were observed. Five cases were downgraded by image analysis. Two cases, one uninterpretable manually and one seen as not amplified manually, were seen as amplified by image analysis.Discussion: Image analysis FISH with Visiopharm software allows for establishment of finite cell inclusion criteria reflecting size, circularity and other measurable cell features. Image analysis facilitates higher cell counts without observer selection bias in less time, and with smaller increases in technical time as more cells are counted. Discordance may be attributable to heterogeneity with larger sample of cells assessed and objective assessment of clusters in the image analysis method. Our image analysis protocol demonstrated successfully the quantification of HER2 in separate signals, in clusters and in split signals. HER2/CEP17 copy numbers were determined for each cell and for more cells in much less time while providing a permanent image record of all cells assessed.Image analysis has promise to improve substantially the quality of Her2 FISH assessment in breast cancer biopsies.Supported in part by an unrestricted grant from Hoffman-LaRoche Ltd. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 6015.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

How this classification was reachedexpand

Full frame machine prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.

metaresearch head score (Codex)0.005
metaresearch head score (Gemma)0.004
Version: metacan-v3-hybrid-931329e0061cValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Bench or experimental · Consensus signal: Bench or experimental
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.009
Threshold uncertainty score0.030

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.0050.004
Meta-epidemiology (narrow)0.0010.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0020.001
Science and technology studies0.0000.001
Scholarly communication0.0010.001
Open science0.0010.001
Research integrity0.0010.001
Insufficient payload (model declined to judge)0.0090.003

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.030
GPT teacher head0.475
Teacher spread0.445 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designBench or experimental
Domainnot available
GenreEmpirical

How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".

Quick stats

Citations0
Published2009
Admission routes1
Has abstractyes

Explore more

Same venueCancer Research→Same topicRadiomics and Machine Learning in Medical Imaging→French-language works237,207→