Aleksandra Zuraw from Digital Pathology Place discusses digital pathology from the basic concepts to the newest developments, including image analysis and artificial intelligence. She reviews scientific literature and together with her guests discusses the current industry and research digital pathology trends.
Site
RSS
Apple
Data updated on 19/11/2026
Recent rankings
Latest chart positions across Apple Podcasts and Spotify rankings.
Shared links between episodes and podcasts
Links found in episode descriptions and other podcasts that share them.
Discover shows related to Digital Pathology Podcast, based on actual content similarities. Explore podcasts with similar topics, themes, and formats, backed by real data.
There is no related content for this show.
257: Digital Cytology and AI in Practice | CorePlus x Epredia at PathVisions 2026
Can a lab run cytology fully digital, with AI, and still keep rescans near 1%?
CorePlus in Puerto Rico has been reading histology digitally since January 1, 2020. Then they took on the hard part: digital cytology. In this talk from the Epredia breakfast workshop at PathVisions 2026, CorePlus founder Mariano de Socarraz walks through six years of building one scanner platform for histology, Pap smears and low-cellularity urine cytology, and where AI paid off.
Every AI project in pathology has to pay off three times: for the pathologist, for the business, and for the patient.
You'll learn:
Why CorePlus went digital on vision and gut feeling, not an ROI spreadsheet, and what "return on innovation" means
How a prostate AI moved from second read to front-end triage: 99.2% negative predictive value across 9,300+ prostate cases and about 122,000 slides
Why cytology is harder to scan than histology, and how water immersion scanning cut scan time by 66%
How 100% Pap smear QC fits in the time a manual 10% review used to take, and the 245 cases it caught
Why Mariano calls the 10% review standard "an illusion" (England's NHS has required 100% review since 2003)
Urine cytology with AI: 92.5% sensitivity, 98.4% specificity, quantitative N:C ratios, and the fight to bring rescans down from 12%
What pathology residents should learn now, and his advice for labs that haven't started
Episode Highlights
00:00 - Every AI project has to pay off three times
01:15 - Six years of digital pathology at CorePlus
02:31 - Starting with vision, not ROI
02:57 - Decoupling the pathologist from the glass
04:06 - Goals: one scanner platform, redundancy, rescans below 1%
07:25 - Prostate AI as a second read: 99.2% negative predictive value
09:13 - AI triage on the front end
10:12 - Five AI algorithms in routine use
10:46 - Why cytology is harder to scan: single layer, extended focus, Z-stacking
14:14 - Water immersion cuts scan time by 66%
14:41 - 100% Pap smear quality control
22:32 - "10% review is an illusion"
23:06 - Low-cellularity urine cytology
24:28 - 92.5% sensitivity, 98.4% specificity
25:41 - From qualitative to quantitative: N:C ratios
28:49 - Rescan rates: where they still are
32:06 - Takeaways: go digital first, add AI where volume and variability are highest
35:53 - Q&A: what should pathology residents learn?
40:56 - Return on innovation
43:34 - Advice for labs just starting
CorePlus publications discussed
Rivera Rolon MDM, Gustafson E, Cole R, et al., de Socarraz M, Santa-Rosario JC. Implementing 100% quality control in a cervical cytology workflow using whole slide images and artificial intelligence provided by the Techcyte SureView System. Cancer Cytopathology, 2025. doi:10.1002/cncy.70019
Santa-Rosario JC, Gustafson EA, Sanabria Bellassai DE, Gustafson PE, de Socarraz M. Validation and three years of clinical experience in using an artificial intelligence algorithm as a second read system for prostate cancer diagnosis - real-world experience. Journal of Pathology Informatics, 2024. doi:10.1016/j.jpi.2024.100378
A glass slide is a data point: one person, one place, one moment, then a box in the archive. A digital slide is a data source: it can be shared, measured, reused and revisited. That idea runs through the whole second edition of Digital Pathology 101, and in this live launch I walk you through what changed since the 2023 edition and why.
I'm Dr. Aleks Zuraw, a board-certified veterinary and toxicologic pathologist and the host of the Digital Pathology Podcast. I wrote Digital Pathology 101 as the bridge between pathology and computer science that I needed in my first job, as the pathologist on a team of image analysis scientists. The second edition is free as a PDF.
What's new in the second edition
Five chapters became six, with new formatting and new images (about 130 pages in the first edition, 179 now)
AI gets two chapters: Understanding AI in Pathology (computer vision to pathology vision, machine learning fundamentals, foundation models, transfer learning) and Implementing AI in Pathology (generative AI, validation, FDA pathways, bias and ethics, running a digital pathology operation)
A rewritten chapter on digital pathology in toxicologic pathology: whole slide imaging, image analysis and AI in nonclinical drug development
QR codes that take you to the matching podcast episodes and courses
Episode Highlights
0:00 - Welcome: the second edition and the Tissue Science Club
3:23 - Why I wrote Digital Pathology 101 in the first place
4:30 - What changed since 2023: remote pathology, AI at scale, foundation models
5:53 - Five chapters became six: what's in each one
9:18 - Glass slide vs digital slide: data point vs data source
255: Are Robots the Next Big Shift in Digital Pathology | with Trang Vu, PhD
Robots are reaching the pathology lab, whole slide scanners are back in the spotlight, and pathology AI companies are getting bought. Dr. Aleks Zuraw, board-certified veterinary pathologist and host of the Digital Pathology Podcast, recaps PathVisions 2026, the Digital Pathology Association's annual meeting, with Trang Vu, who comes from the spatial biology side and tracks digital pathology vendors closely.
You'll learn:
Why hardware and robotics took over the PathVisions exhibit hall this year
What makes Grundium's new Ocus R different: independent scanning heads you add over time, a slide tray instead of a door
Why the biggest whole slide scanner isn't always the right one for your lab's slide volume
How AxLab's robotic arm feeds slides into a scanner, and why robotics is only now reaching pathology
Why Philips' DICOM-ready scanner files matter for storage and cloud AI
Why standalone pathology AI companies are partnering with image management systems (IMS)
What this year's deals mean: AstraZeneca and Modella, Roche and PathAI, Agilent and Proscia
Where virtual staining is ready, and where it still needs validation for every marker
Elsie, the humanoid lab assistant from Tactus AI
Who small and remote scanners are for, including a $4,500 option for rural hospitals
Chapters
00:00 Elsie: a robot that reloads your scanner overnight
01:05 PathVisions 2026 in 30 seconds
01:42 Meet Trang Vu
03:01 Why PathVisions is "USCAP for the cool kids"
05:03 Hardware is back on the exhibit floor
07:18 Grundium's Ocus R: launch strategy and design
13:17 Do you need a 600-slide scanner?
254: Digital Pathology Q&A: How to Get Started | PathVisions 2026
Is a digital slide as good as glass? Live from the Hamamatsu booth at PathVisions 2026, Dr. Aleks Zuraw and Amy Welch, Product Marketing Manager for NanoZoomer at Hamamatsu, open the question box and answer the beginner questions attendees dropped in all week. Elizabeth Blakely joins as a surprise guest.
You'll learn:
Whether a digital slide is as good as glass for diagnosis
How long it takes to scan a slide, and why the 15 x 15 mm benchmark means little to a pathologist
How a whole slide scanner turns a glass slide into a digital image
Whether AI will replace pathologists
Why a digital slide is a data source, not a data point
The parts of the digital pathology workflow, from pre-analytics to AI
What to learn first if you're new to the field
Chapters
00:00 Live from PathVisions 2026 with Amy Welch
00:56 Is a digital slide as good as glass?
02:49 What the NanoZoomer numbers mean
03:22 How long does it take to scan a slide?
05:14 What surprised us most when we started in digital pathology
09:29 Is there a scanner for small volumes?
10:15 Will AI replace pathologists?
16:42 Pathology clinics: patients seeing their own slides
17:44 How does a whole slide scanner work?
21:11 What is digital pathology?
25:14 Digital pathology in 25 years
26:07 Surprise guest: Elizabeth Blakely
29:33 Direct-to-digital tissue imaging
253: What Happens After Your Whole Slide Scanner Arrives? | PathVisions 2026
What happens after your whole slide scanner arrives in the lab? Live from the Hamamatsu booth at PathVisions 2026, Dr. Aleks Zuraw meets the people who help labs go digital, plus a surprise guest from the DICOM standard.
In this episode:
David Clunie of DICOM Working Group 26 on the PathVisions 2026 DICOM Connectathon: scanners, archives, viewers and annotation tools from different vendors working together, and the new work on annotations with holes
Stephanie Fullerton introduces two new members of the Hamamatsu team
Jessica Calafati, NanoZoomer national sales manager, on why going digital is easier than it was 10 years ago, what to do about a high rescan rate, and why your scanner and image management system (IMS) vendors need to talk before every upgrade
Gary Davis, business development manager for MoxiePlex, on multiplex fluorescence imaging and spectral unmixing
Chapters
00:00 What happens after your whole slide scanner arrives
00:26 Live from the Hamamatsu booth at PathVisions 2026
01:07 David Clunie on the DICOM Connectathon results
02:48 DICOM annotations: contours within contours
05:06 Hamamatsu's new team members
06:09 Jessica Calafati: why going digital is easier now
07:47 How to fix a high rescan rate
09:29 Why scanner and software teams need to talk
09:59 Gary Davis on MoxiePlex
11:18 Spectral unmixing in three sentences
12:41 Send us your scanner questions
252: Digital Pathology Trends 2026: Agilent Buys Proscia | PathVisions
What are the biggest digital pathology trends of 2026, and what does Agilent buying Proscia tell us?
It's the clearest sign yet that the field is moving from shiny algorithms to full, connected workflows. In this live recap from Pathology Visions 2026 in San Diego, Dr. Aleks Zuraw (DVM, PhD, DACVP), board-certified veterinary pathologist and host of the Digital Pathology Podcast, sits down with Imogen Fitt of Signify Research to break down the biggest news of the week and the trends behind it.
Imogen tracks the digital pathology market for a living. Aleks brings the pathologist's view. Together they cover the Agilent and Proscia deal, why software is finally being built around pathologist ergonomics, how scanner differences can change AI results, and whether every AI tool needs FDA clearance or can live as a lab-developed test (LDT).
Key takeaways - Agilent's acquisition of Proscia is the peak of a partnership trend: everyone wants to be compatible with everyone. - This year's software news centers on ergonomics, voice dictation and voice AI, not just new algorithms. - PathVisions posters compared scanners and showed that image differences between models can change AI output. Purpose and intended use have to be defined within a whole system. - "Good enough" can be the right specification when AI is a safety net that flags cases for a human. - Not everything needs to be a medical device. The LDT pathway lets single labs start, gather data, and decide later whether to pursue clearance. - Pay-per-scan models and implementation work, not hype, are what will get digital pathology to more labs.
Episode Highlights 00:36 - Agilent acquires Proscia - The biggest digital pathology news heading into PathVisions. 01:10 - Aperture and Concentriq - AI that finds clinical trial patients and an AI-enabled image management system. 01:43 - Ergonomics and voice AI - Why nobody likes pop-up windows, and what's replacing them. 02:16 - From staining to sign-out - Aleks recalls first seeing Agilent's full workflow at USCAP.03:39 Scanners change AI output - What this year's posters showed, and why it surprised a pathologist.05:09 - Good enough vs perfect - Imogen plays devil's advocate on accuracy targets.07:15 - The pixel pathway - How locked systems ended up driving interoperability.08:02 - FDA clearance or LDT? - Does every pathology AI tool need to be a medical device?11:03 - Implementation over hype - Why the "boring" topics are the real progress.12:03 - Pay-per-scan - Getting digital pathology to labs that can't buy a scanner upfront.Resources Mentioned- Signify Research: https://www.signifyresearch.net/- Pathology Visions 2026: https://digitalpathologyassociation.org/pathology-visions-conference- Free Digital Pathology 101 book: https://www.aleksandrazuraw.com/bookConnect with Imogen Fitt on LinkedIn: https://www.linkedin.com/in/imogen-fitt-194566127/Enjoyed this episode? Share it with one colleague who's still deciding whether to go digital.
251: Pathology Foundation Models: PathBench, Scanner Shift, and the HER2 Slide Swing
What if changing the scanner or choosing another slide from the same tumor changes your pathology AI results?
In DigiPath Digest #52, I review three studies on pathology foundation models, scanner shift, and slide selection to ask how much of “model performance” comes from the model itself.
We look at a benchmark comparing 32 foundation models across 41 tasks, a study testing 10 models across scanners and centers, and a breast cancer study examining how predictions change when different slides represent the same patient.
The findings challenge familiar assumptions about bigger models, benchmark rankings, and the idea that one slide can represent a heterogeneous tumor. Combining information from multiple slides improved performance in the breast cancer study, without changing the model architecture.
My takeaway: evaluate the model, the data, and the workflow together. I share three practical questions to ask before accepting an AI benchmark score.
Highlights with Timestamps
00:42 — PathVisions updates and plans for conference activities.
05:09 — Trailblazer podcast interviews: sharing real implementation experience.
08:02 — How much of foundation model performance comes from the model?
13:16 — The 32-model benchmark and closely grouped leading scores.
16:06 — Why bigger models do not consistently perform better on unfamiliar data.
18:58 — Scanner shift and testing models on scanners held out during training.
22:18 — Why AUROC alone can miss important performance losses.
24:55 — Use-case-specific validation and the “pixel pathway” from scanning to diagnosis.
28:37 — Slide selection, tumor heterogeneity, and HER2 prediction.
33:46 — Combining slides from the same patient: performance gains and computational cost.
35:31 — Three questions to ask before trusting a benchmark.
37:27 — Audience Q&A on available datasets and checking AI-generated answers.
250: AI for Scientific Writing: 4 Rules from a Pathologist
Can you use AI to write scientific papers without fabricated references, journal trouble or learning to code? I recorded this episode at the airport on my way to ESTP 2026 in Kraków. It's my practice run of the talk I gave there.
When deep learning came into pathology, people asked me if I was going to learn to code. I decided no. Large language models kind of proved me right. In this episode I walk you through how I use LLMs for writing papers, responding to co-authors, building slides and even running code, and the four rules that keep it safe:
Domain expertise is the prompt
Supply the sources
Encode corrections
The subject matter expert verifies
In this episode:
What the LLM does vs. what the pathologist does, task by task
What Toxicologic Pathology, Veterinary Pathology and Elsevier journals require when you disclose AI use
Pathology LLM tools already working on tox study reports, image analysis and whole slide images
The reference I didn't check that almost cost me a publication
Why an LLM will happily justify switching patients from Tylenol to acetaminophen
Automation bias, model bias, and how to start with the tools your institution has approved
What happens when AI models that perform almost perfectly on scattered cervical cells become less reliable than a coin flip on crowded cell groups?
In DigiPath Digest #51, I examine what this performance gap tells us about artificial intelligence in cytopathology.
The first paper evaluated six convolutional neural network models trained to distinguish benign from high-grade lesions using scattered cervical cytology cells. The models achieved AUCs ranging from 0.950 to 0.996 on the original images.
When the same models were applied to hyperchromatic crowded cell groups without retraining or threshold recalibration, their AUCs fell to 0.385-0.683. Different architectures also failed differently. Some overcalled benign clusters, while others missed high-grade clusters.
This isn’t simply a technical problem. It illustrates a practical rule for pathology AI: a model should only be trusted for the morphology, specimen type, imaging system, and intended use on which it has been directly validated.
I also review the current evidence for large language models in cytopathology. Potential applications include structured reporting, diagnostic support, quality control, education, research, and workflow integration.
Some early results appear promising, but each comes with important limitations. One structured-reporting application achieved 99.4% accuracy at a single institution. A diagnostic-support model included the correct answer among its top 10 differentials in 59.1% of general medicine cases. A hybrid quality-control system flagged 84% of errors associated with amended reports, but its false-positive rate wasn’t reported.
Most importantly, the review found no language model specifically trained and clinically validated on cytopathology reports.
The takeaway is straightforward: we’re still working with narrow AI. Strong performance in one setting doesn’t guarantee performance when the cells, preparation, scanner, institution, or clinical task changes.
248: Are Foundation Models Really Better for Digital Pathology? Podcast with Panu Kauppila
What good is a powerful foundation model if it slows the pathologist down, can’t explain its result, or doesn’t fit the clinical workflow?
Foundation models are gaining attention across digital pathology. But they’re not finished clinical tools by themselves.
In this episode of the Digital Pathology Podcast, I speak with Panu Kauppila, Chief Product Officer at Aiforia, about what foundation models are, how they differ from traditional convolutional neural networks, and what it takes to make them useful for pathologists.
Panu describes a foundation model as a large, context-aware building block. To perform a specific pathology task - such as grading, segmentation, or mitotic counting - it must be combined with an adapter, a task-specific head, curated annotations, a usable interface, and integration with the laboratory workflow.
We also discuss one of the biggest practical constraints: speed.
A pathologist shouldn’t have to click a button and wait for an analysis. Panu explains why AI should run automatically in the background so the results are already available when the case reaches the pathologist’s worklist.
The conversation also examines the tradeoff between model size, computational cost, and clinical performance. Larger models may improve robustness and generalizability, but they can also require more processing power. For clinical applications, Aiforia focuses on smaller and medium-sized foundation models that provide the necessary quality without making the workflow slower or unnecessarily expensive.
Annotated data remains central. The foundation model supplies the underlying image understanding, while the task-specific head and controlled annotations determine how the model performs on a particular pathology problem. This structure also raises important questions about bias, data provenance, ownership, regulatory documentation, and explainability.
Not on Amazon yet. I'll let you know when the print version is out.
Thank you for learning with me, trailblazer. If this episode helped you, share it with one colleague who's just starting their digital pathology journey.
23:20 Pathology AI companies and image management systems
27:55 Modella and AstraZeneca
30:27 "Acquisition row": PathAI, Roche and Proscia
32:26 Agilent's full digital pathology stack
35:16 Virtual staining: where it's ready and where it isn't
39:51 Elsie, the humanoid lab assistant from Tactus AI
44:10 Direct-to-digital imaging and why digitization pays off
47:54 Small and remote whole slide scanners
51:08 A $4,500 remote scanner for rural hospitals
54:00 What's next: USCAP 2027 in Vancouver
Good to know: Agilent has signed a definitive agreement to acquire Proscia; the deal hasn't closed yet. Philips' "50% smaller files" is Philips' own up-to figure based on preliminary data.
Follow Trang Vu on LinkedIn: https://www.linkedin.com/in/trang-nguyen-vu/
Dr. Beth Plocharczyk's episode on serving two community hospitals with remote scanners:
Low-risk applications may offer the most practical starting point. Text extraction, completeness checks, report consistency review, and quality-control flagging could reduce repetitive work without asking an unvalidated model to make the final diagnosis.
Highlights with timestamps
00:00 - Welcome to DigiPath Digest #51 and the new lunch-and-learn time
01:40 - How two image models performed worse than a coin flip on cell clusters
02:29 - Image models, language models, and vision-language models
04:40 - Why AI adoption in cytopathology remains low
05:25 - Scattered single cells versus hyperchromatic crowded cell groups
08:14 - AUCs fall from 0.950-0.996 to 0.385-0.683
10:02 - How ResNet-50 and GoogLeNet failed differently
11:04 - What the attention maps revealed
13:22 - The intended-use lesson for pathology AI
17:02 - Current applications of large language models in cytopathology
18:39 - Structured reporting and the 99.4% accuracy result
19:32 - Diagnostic support, AMIE, and the top-10 limitation
21:09 - Quality-control applications and the missing false-positive rate
23:30 - Why cytopathology still needs domain-specific language models
24:54 - Retrieval-augmented generation, education, and research support
27:07 - Two AI families, one requirement: direct validation
28:35 - Context of use and intended-use validation
29:12 - Human-reviewed training data and destructive book scanning
32:02 - FDA and European approaches to evolving AI models
35:00 - Protecting patient and practitioner well-being
36:21 - Cytopathology-specific benchmarks and shared test sets
38:16 - Why low-risk AI applications should come first
Finally, we look at multimodal AI models that combine pathology images with reports, genomic data, molecular information, and clinical outcomes. These tools could support more interactive, predictive, and prognostic applications - but only if they’re introduced through secure, controlled workflows with clear audit trails.
Episode Highlights
00:00 — Where does bias enter a foundation model workflow? Panu distinguishes the underlying foundation model from the annotated dataset used to build the task-specific application.
00:27 — Meet Panu Kauppila An introduction to Aiforia’s Chief Product Officer and the episode’s focus on foundation models in digital pathology.
01:05 — From radiology AI to digital pathology Panu describes his background in medical device development, radiology, oncology solutions, and clinical AI implementation.
06:29 — Foundation models versus convolutional neural networks What makes foundation models more context-aware, robust, and generalizable across image datasets.
07:07 — Image-only and multimodal foundation models Why these two categories offer different capabilities and potential clinical uses.
07:50 — A foundation model is a platform, not a finished solution The underlying model may understand image features, but it still needs a task, interface, and clinical workflow.
08:34 — Foundation model, adapter, and task-specific head How these components work together to create an application for grading, mitotic counting, or another pathology task.
09:57 — The cost of larger models Why increased robustness must be balanced against computational demands, inference speed, and affordability.
11:12 — Pathologists won’t wait for AI Why even short delays can interrupt the clinical workflow.
11:46 — Running AI in the background A workflow in which slides are scanned, analyzed automatically, and added to the worklist with results ready for review.
12:16 — Combining foundation models with curated annotations How smaller task-specific datasets and adapter technology can produce practical pathology models.
15:24 — Generalizability across scanners, laboratories, and populations How foundation models may make adaptation to new domains more manageable.
16:36 — Two datasets, two sources of potential bias The regulatory questions created by an underlying foundation model and a separate controlled annotated dataset.
20:46 — Making foundation models accessible Why a platform and user interface are necessary for pathologists and researchers who don’t work directly with code.
21:57 — Testing foundation models in Aiforia Create How researchers can compare a CNN with supported foundation models in the same no-code environment.
25:45 — How foundation models are selected Quality, licensing, model size, annotated data, and the requirements of the intended use.
26:57 — Training cost versus inference cost Why a more expensive training iteration may still reduce total development costs if fewer iterations are needed.
32:23 — Pathologists are visual reviewers The importance of segmentation quality and showing exactly what the model identified.
36:26 — The potential of multimodal AI Combining pathology images with text, genomic information, molecular data, and clinical outcomes.
37:35 — Keeping multimodal AI inside a controlled environment Privacy, security, regulatory oversight, and the risks of moving clinical information into consumer AI tools.
43:24 — Explainability in clinical pathology AI Using semantic segmentation, object detection, instance segmentation, and annotated ground truth to show how results were calculated.
47:43 — Moving toward predictive and prognostic models How established digital workflows could allow pathologists to contribute more information about likely outcomes.
49:45 — Research and clinical collaboration with Aiforia How interested researchers and laboratories can connect through the Aiforia website.
Listen to the full discussion to understand what foundation models can add to digital pathology—and what still has to happen before they become practical, trusted clinical tools.