ARTIFICIAL INTELLIGENCE


fact that 92.3% of CE-marked pathology algorithms reported by Matthews et al. received approval through self-certification under the now-superseded IVDD directive, which did not involve review by a notified body.23


While a few tools have


since achieved certification under the more stringent IVDR framework,27,28


which


ensures that devices undergo rigorous validation and meet the highest clinical and safety standards, these remain the exception rather than the rule. However, with the forthcoming European AI Act, the use of ethical and unbiased AI systems will become mandatory, offering a regulatory framework that may further strengthen public and professional trust in AI- supported diagnostics.


The hidden costs of postponing AI-assisted pathology While concerns around costs and reliability are valid, delaying the implementation of AI in pathology incurs substantial hidden costs – clinical, ethical, and financial. Economically, significant savings have been reported in several assessments of AI-assisted pathology, including prostate cancer diagnosis29


and


detection of breast cancer metastases in sentinel lymph nodes.30


The latter was


associated with potential cost savings that range from €1,500 to €12,500 (approx. £1,250 – £10,500) per 100 evaluations.30 Beyond financial benefits, postponing the use of AI tools allows high inter- observer variability to persist. This variability can lead to under- and over- treatment, negatively impacting patient outcomes. Missing an early-stage malignancy may result in more aggressive interventions later, while overestimating tumour grade can lead to unnecessary treatments with harmful side effects, reducing wellbeing and life expectancy.4,31 As mentioned above and depicted in Figure 3, studies have shown that in some conditions, up to three in 10 patients may receive suboptimal or incorrect therapy due to diagnostic discrepancies.7 Delays in integrating AI not only postpone potential savings but also perpetuate clinical inconsistencies that undermine patient care. From an ethical


A critical obstacle is the relatively low rate of full digital pathology adoption, which constrains the deployment of AI tools


Unassisted AI-assisted


Fig 2. Enhanced consistency with Aiosyn Mitosis Breast in clinical performance study. (a) Heatmap visualisation illustrating the consistency of hotspot selection (areas counted by pathologists). Colours range from green (minimal overlap between readers) to red (maximum overlap), indicating high agreement. (b) Graphical representation of the consistency improvements observed in the study, showing the increased alignment in mitosis counting when using Aiosyn Mitosis Breast. Source: Tessier et al.11


perspective, if technology exists that can reduce variability and improve outcomes, there is a collective responsibility to ensure its timely adoption.


Bridging the gap: strategies to facilitate adoption The current healthcare landscape, marked by budgetary constraints and a limited pool of tools certified under the new IVDR, has placed many digital pathology laboratories in a holding pattern. Institutions are following new developments but delaying investment until more favourable conditions emerge, such as improved reimbursement mechanisms and a broader selection of certified solutions.


In the meantime, digital pathology and


AI providers can take proactive steps to reduce the adoption barrier: n Conduct multi-centre validation studies to verify clinical performance in different settings.


n Pursue IVDR certification under notified body review for tools previously approved under IVDD, to enhance trust and ensure market access.


n Design modular and interoperable software solutions that integrate easily with existing laboratory systems and workflows.


WWW.PATHOLOGYINPRACTICE.COM OCTOBER 2025 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1


AI-assisted mean (0.48) Unassisted mean (0.33)


Unassisted AI-assisted Consistency (kappa)


n Implement pilot programmes that allow institutions to test AI tools in real-world settings before full-scale adoption.


n Publish ROI-focused analyses that quantify cost-effectiveness and operational benefits.


Although evidence shows that pathologists working alongside AI achieve levels of consistency and efficiency that surpass traditional methods,11,12,14-16


there


is a significant delay in bringing these benefits to patient care. While software providers must continue to address the points above, the responsibility also lies with end users. Whether we see broader adoption in the near future will depend on whether more laboratories choose to adopt digital workflows, explore pilot programmes, include AI in procurement roadmaps, and embrace certified AI tools as part of their diagnostic repertoire.


A practical case: making Aiosyn’s AI clinically ready Aiosyn Mitosis Breast is an IVDR-certified AI solution developed to address the challenge of inter-observer variability in mitotic figure assessment. Designed for breast cancer grading, this algorithm identifies mitotic figures in haematoxylin


41 0.0 1.0 0.9


Cohen’s kappa (linearly weighted)


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46  |  Page 47  |  Page 48  |  Page 49  |  Page 50  |  Page 51  |  Page 52  |  Page 53  |  Page 54  |  Page 55  |  Page 56