AI may cut breast biopsy review time by 44%

A paired-reader study found that pathologists maintained high diagnostic agreement while reviewing breast biopsy slides faster with AI assistance.

Category	Detail
Condition	Breast biopsy pathology review, including assessment for breast cancer, invasive carcinoma, and ductal carcinoma in situ (DCIS)
Key Mechanisms	AI-assisted digital pathology provides decision support during review of digitized hematoxylin and eosin–stained slides; AI-guided immunohistochemistry (IHC) preordering was modeled as a potential way to reduce laboratory turnaround time.
Target Population	194 breast biopsy cases from adult female participants collected at one Australian center between January and July 2024; two experienced pathologists reviewed each case.
Care Setting	Pathology laboratory workflow; findings are from a retrospective, single-center paired-reader study, with proposed IHC-related turnaround-time reductions modeled rather than prospectively tested.

Mean pathologist review time was 40.18 seconds with AI-assisted digital pathology versus 71.12 seconds with standard digital pathology, a reported reduction of about 44%.
Cancer diagnostic agreement was 99% (193/194) with AI assistance and 97% (189/194) with standard digital pathology.
Cancer detection sensitivity was 100% with AI assistance versus 98% without it; specificity was 98% versus 95%, respectively.
For DCIS, agreement was 99% versus 93% and sensitivity was 100% versus 86%; the subgroup included only 21 DCIS-positive cases.
A modeled estimate suggested AI-guided IHC preordering could reduce turnaround time by about 2.5 days for applicable true-positive cases; this was not demonstrated prospectively.

The study compared pathologist diagnoses using standard digital pathology and AI-assisted digital pathology against original pathology diagnoses, with independent adjudication of identified discrepancies.
The standalone AI algorithm was assessed for detection of breast cancer, invasive carcinoma, and DCIS.

The study evaluated AI as decision support in breast biopsy pathology review; it did not establish a clinical management guideline.
The authors stated that confirmation in larger multicenter studies is warranted.

The study assessed pathologist review time, laboratory turnaround time, usability, diagnostic accuracy, and standalone AI performance.
The proposed turnaround-time reduction from AI-guided IHC preordering requires prospective evaluation.

The study was single-center, involved two pathologists, and assessed one representative slide per case.
Cases with inadequate image quality were excluded, and independent adjudication was limited to identified diagnostic discrepancies.
The study was not powered to establish differences in diagnostic accuracy; the small DCIS subgroup limits precision.

The paired-reader analysis included 194 breast biopsy cases from adult female participants. No treatment or prescribing intervention was evaluated.

Not applicable: this study evaluated pathology workflow and diagnostic performance, not patient treatment or medication prescribing.

Interpret the reported review-time and diagnostic-performance results in the context of the study’s retrospective, single-center design and two participating pathologists.
Treat the estimated 2.5-day turnaround-time reduction from AI-guided IHC preordering as modeled, not as a prospectively demonstrated outcome.
Consider the limited number of DCIS-positive cases when interpreting subgroup results.