Optimizing breast core needle biopsy biomarker throughput using an AI-based workflow

Q: How does autonomous, LIS-embedded biomarker ordering actually work without compromising pathologist oversight?

When whole-slide H&E digital images are scanned, they are analyzed by deep-learning algorithms which stratify cases into low, medium, and high likelihood of invasive breast cancer. For cases classified as “high likelihood,” the system automatically generates structured JSON files and triggers requests for the standard breast biomarker panel (ER, PR, HER2, and Ki-67) directly in the LIS. The pathologist’s primary role remains untouched; they review and sign out the final IHC stains, but they receive the slides significantly earlier in the diagnostic cycle because the ordering occurred before their initial review.

Q: Does this automated workflow lead to unnecessary tissue consumption or costly, wasteful staining?

No. The algorithm’s decision thresholds were specifically calibrated to prioritize high positive predictive value (97.28% PPV) for the automated category. During the entire active implementation study, the platform achieved a 100% concordance rate with pathologists’ final diagnoses—meaning zero unnecessary IHC stains were triggered by the system. Any cases where invasive carcinoma is present but falls below this high-confidence threshold are safely triaged to the standard pathologist-driven manual workflow.

Q: How does reducing technical turnaround time translate to real-world clinical and patient benefit?

Accelerating technical laboratory processes directly compresses clinical reporting timelines. In this study, the technical gain of 10.5 hours translated directly into a reduction of approximately one working day for the final pathology report (TATclin). This dramatic reduction directly addresses patient anxiety, optimizes scheduling for multidisciplinary tumor boards, and accelerates time-to-treatment decisions in high-throughput breast oncology clinics.

30 May, 2026

EXECUTIVE BRIEFING

This retrospective, single-center implementation study evaluates the operational impact, diagnostic relevance, and turnaround time efficiency of integrating an artificial intelligence-driven pre-ordering workflow for breast cancer biomarkers within a high-volume pathology laboratory. By utilizing automated deep-learning algorithms integrated with the Laboratory Information System (LIS) to autonomously trigger immunohistochemical staining for high-probability invasive breast cancer specimens, the system bypasses traditional pathologist-initiated order friction. The study demonstrates how pre-emptive workflow automation can dramatically accelerate therapeutic and clinical reporting timelines while preserving absolute diagnostic quality and laboratory safety.

CLINICAL & ENTERPRISE HIGHLIGHTS

Operational Efficiency & Turnaround Time Gains

Technical Turnaround Time Reduction (TATtech): 30% average decrease (dropping from approximately 33.5 hours pre-implementation to 23 hours post-implementation, p < 0.01) measured from the moment of case assignment to technical staining finalization.
Clinical Reporting Turnaround Time Reduction (TATclin): 30% reduction (saving approximately 1 full working day, p < 0.03) measured from case assignment to final diagnostic report.
Technical Performance Stability: 6 consecutive months of sustained TAT reduction following the implementation. Seasonal confounders were ruled out through year-over-year baseline comparison.

Clinical Quality Assurance & Calibration Metrics

Automated Ordering Diagnostic Concordance: 100% appropriate relevance with zero unnecessary immunohistochemical (IHC) stains ordered by the system during the evaluation period.
Calibration Set Positive Predictive Value (PPV): 97.28% for the high-likelihood classification, optimizing automated order confidence on breast core needle biopsies.
Calibration Set Negative Predictive Value (NPV): 99.76% for the low-likelihood category, ensuring high diagnostic safety and minimizing missed carcinomas.
Clinical Cohort Scale: 547 total patient cases analyzed by the system during the active post-implementation period, including 198 confirmed invasive carcinoma cases.

THE BOTTOM LINE

Embedding clinical AI directly into laboratory workflows as an operational engine—rather than a passive diagnostic support tool—delivers massive business and clinical value. By automating routine pre-analytical ordering, laboratories can compress clinical reporting turnaround times by 30%, save pathologists significant administrative time, and eliminate inter-departmental friction. Crucially, this efficiency is unlocked with zero impact on diagnostic accuracy or quality control, enabling high-volume laboratories to confidently scale operations and elevate the standard of patient-centered oncologic care.

CITATION & CTA

APA Citation:

Deman, F., Broeckx, G., Degotte, Q., Siozopoulou, V., del Carmen Tavares, M., Declercq, S., Salgado, R., & Dendooven, A. (2026). Optimizing breast core needle biopsy biomarker throughput using an AI-based workflow. Histopathology, 1–9. https://doi.org/10.1111/his.70181

Frequently Asked Questions

? How does autonomous, LIS-embedded biomarker ordering actually work without compromising pathologist oversight?

↳

When whole-slide H&E digital images are scanned, they are analyzed by deep-learning algorithms which stratify cases into low, medium, and high likelihood of invasive breast cancer. For cases classified as “high likelihood,” the system automatically generates structured JSON files and triggers requests for the standard breast biomarker panel (ER, PR, HER2, and Ki-67) directly in the LIS. The pathologist’s primary role remains untouched; they review and sign out the final IHC stains, but they receive the slides significantly earlier in the diagnostic cycle because the ordering occurred before their initial review.

? Does this automated workflow lead to unnecessary tissue consumption or costly, wasteful staining?

↳

No. The algorithm’s decision thresholds were specifically calibrated to prioritize high positive predictive value (97.28% PPV) for the automated category. During the entire active implementation study, the platform achieved a 100% concordance rate with pathologists’ final diagnoses—meaning zero unnecessary IHC stains were triggered by the system. Any cases where invasive carcinoma is present but falls below this high-confidence threshold are safely triaged to the standard pathologist-driven manual workflow.

? How does reducing technical turnaround time translate to real-world clinical and patient benefit?

↳

Accelerating technical laboratory processes directly compresses clinical reporting timelines. In this study, the technical gain of 10.5 hours translated directly into a reduction of approximately one working day for the final pathology report (TATclin). This dramatic reduction directly addresses patient anxiety, optimizes scheduling for multidisciplinary tumor boards, and accelerates time-to-treatment decisions in high-throughput breast oncology clinics.

Share this article:

Subscribe to our newsletter:

By providing my details I agree that my data will be handled in accordance with IBEX Privacy Policy.

Presented at

Histopathology

Authors

Affiliations

Peer reviewed article

How is artificial intelligence rising to the challenge in breast cancer diagnosis?

How ‘game-changing’ AI is helping discover cancer for the first time in UK | ITV News

The AI revolution in cancer diagnostics with Joseph Mossel, CEO of Ibex Medical Analytics