Vet Clin Path Professor: Dr. Candice Chu

Vet Clin Path Professor: Dr. Candice Chu 👩🏻‍⚕️ Candice Chu, DVM, PhD, DACVP 臺灣人
🤖 Veterinary AI Educator & Researcher
🇺🇲 Assistant Professor at Texas A&M University

Our new study, “Artificial intelligence–based detection of small and intermediate-to-large lymphocytes for cytologic dia...
08/31/2026

Our new study, “Artificial intelligence–based detection of small and intermediate-to-large lymphocytes for cytologic diagnosis of canine lymphoma,” is now published in Veterinary Pathology!

Rather than relying on whole-slide scanners or specialized digital pathology equipment, we explored whether AI could extract diagnostically useful information from images acquired with cell phones.

In this study, we developed an AI-based approach to detect and classify small and intermediate-to-large lymphocytes in canine cytology images, with the potential to support cytologic assessment of lymphoma in the future.



👉My lab is hiring a thesis-master student and a board-eligible or board-certified pathologist who is interested in digital pathology. Please send me an email if you are interested. 👈

Save the date: the AVMA Convention is bringing a full six-hour, AI-focused continuing education session to veterinarians...
08/14/2026

Save the date: the AVMA Convention is bringing a full six-hour, AI-focused continuing education session to veterinarians in 2027.

I just spent two days with the AVMA Task Force on Emerging Technologies and Innovation at our fall in-person meeting, and planning this symposium was a big part of it. Six hours is a real commitment of program time — enough to go beyond ”AI is coming“ and actually work through what practitioners need: AI literacy resources, a framework for ethical AI use in veterinary practice, and closer collaboration with the academic researchers who build and validate these tools.

📅 AVMA Convention 2027 · July 16–20, 2027
📍 Minneapolis Convention Center, Minneapolis, MN

Veterinarians don‘t need to become machine learning engineers. They need to know what to ask and what to trust. That’s what we‘re building toward.

Sneaking AI into your work? Good luck hiding it from now on.Anthropic updated its help docs today (8/11) and officially ...
08/11/2026

Sneaking AI into your work? Good luck hiding it from now on.

Anthropic updated its help docs today (8/11) and officially confirmed that Claude will embed machine-readable marks in the content it generates. Which means AI-edited text and images are about to get a lot easier to spot.

This comes out of Anthropic's commitment to the EU AI Act's Article 50(2) Code of Practice on transparency, but it applies worldwide.

There are two marking methods:

📝 Text watermarks: An invisible watermark baked into the text itself. It doesn't change the meaning, quality, or readability. And because it lives in the text rather than in a file, it survives copy and paste, and can survive some editing too. Every Claude product and interface will produce the “watermark.”

🖼️ File provenance metadata: Supported file types (.svg, .png, .jpg) get signed provenance metadata following the C2PA standard, which also lets you tell whether the file has been edited by AI.

Future Claude models launched in the EU on or after August 2, 2026 support marking from day one. Anything released before that is still inside the transition period the regulation allows, and Anthropic says it's working on backfilling support. We don’t know if the current Claude output is already watermarked.

Note that:

A detected mark only means the content may have been processed by Claude. Proofreading, translation, and summarizing all carry the mark too. It does not mean Claude wrote it.

And no mark detected does not mean no AI. Heavy rewriting, translation, very short passages, or a file conversion that strips the metadata will all kill the signal.

The technical docs for detecting these marks haven't been published yet, so nobody can build detection into their workflow today. But a Claude AI detector feels like a matter of time.

How Claude mark AI generated content: https://lnkd.in/gwSSweVa

So what about ChatGPT?

OpenAI hasn't announced any verifiable watermark for regular ChatGPT text. However, images from ChatGPT, Codex, and the API already carry both C2PA metadata and an invisible SynthID watermark. SynthID is embedded in the image signal itself, so even if the metadata gets stripped, it can survive certain edits and conversions.

OpenAI already published a public verification tool where you upload an image and check for AI tracing: https://lnkd.in/gbZCdw-Y

Poland 🇵🇱 I am coming~Any veterinary ophthalmologists heading to Warsaw next year for the European College of Veterinary...
08/09/2026

Poland 🇵🇱 I am coming~

Any veterinary ophthalmologists heading to Warsaw next year for the European College of Veterinary Ophthalmologists (ECVO) meeting?

I’ve been invited as the Guest Speaker on AI, and I’ll be giving a featured lecture on day one of the conference plus joining a panel discussion.

The organizers are expecting around 500 in-person attendees and 200 online.

Time to start planning a Poland-Czech-Vienna trip 😆

AI Cytology Called It “Inflammation.” Histopathology Called It a “Malignant Tumor.” An Oregon Veterinary Hospital Is Now...
08/01/2026

AI Cytology Called It “Inflammation.” Histopathology Called It a “Malignant Tumor.” An Oregon Veterinary Hospital Is Now Suing Zoetis



(Important note: The following summary is based on reporting by The Oregonian/OregonLive and The Independent. The reports describe allegations made in the complaint, which represent only one side of the dispute. The case has not yet been tried, and the court has made no findings of fact. According to the reports, Zoetis had not responded to media inquiries at the time of publication.)

[News Summary]

Columbia Veterinary Hospital in The Dalles, Oregon, filed a lawsuit against Zoetis in Wasco County last Thursday, alleging fraud and violations of Oregon’s Unlawful Trade Practices Act. The hospital is seeking at least $50,000 in damages.

According to the complaint, an 11-year-old Belgian Tervuren was evaluated for a rapidly enlarging mass on the neck. In October 2025, the hospital submitted a cytology sample for analysis using Zoetis’s Vetscan Imagyst platform. The complaint alleges that the AI system detected mast cells and spindle cells but ultimately classified the lesion as inflammatory rather than neoplastic.

Based on that interpretation, the veterinarian surgically removed the mass. Subsequent histopathologic evaluation by the Oregon State University Veterinary Diagnostic Laboratory reportedly identified the lesion as an invasive malignant tumor.

The plaintiff alleges that the dog’s death was directly related to the need for a second invasive surgery, performed within a very short period, to achieve complete tumor removal. The complaint further argues that the second surgery would not have been necessary if Imagyst had initially identified the lesion as malignant, as the plaintiff believes Zoetis’s marketing led users to expect.

The lawsuit claims that Zoetis promoted Imagyst as “the world’s most capable veterinary AI analyzer,” leading the hospital to believe that the system would provide diagnostic results comparable to those of an expert.

The complaint also alleges that Zoetis modified the original AI interpretation report after being notified of the alleged misclassification. According to the filing, Zoetis requested a conference call after the hospital reported the case. During that call, company representatives allegedly apologized, acknowledged fault, and stated that other Imagyst users had experienced similar incidents.

[My Perspective]

Cytology is not, and has never been, 100% accurate. Mast cells can also be present in small numbers within inflammatory lesions. Without reviewing the original cytology slides, I cannot determine whether the interpretation was unreasonable or whether the diagnostic limitations were adequately communicated.

That said, if an incorrect cytologic interpretation alone is sufficient grounds for taking legal actions, the concern would extend beyond AI systems. Even board-certified veterinary clinical pathologists could potentially face similar claims. The difference is that few specialists would describe themselves as “the world’s most capable veterinary clinical pathologist.”

Regardless of the eventual legal outcome, this dispute highlights a meaningful gap between how companies market veterinary AI services and how veterinarians understand the capabilities, limitations, and appropriate clinical use of those systems.

This is why teaching AI literacy to veterinary professionals is so important!

Sources:

The Oregonian/OregonLive. “Oregon veterinary hospital sues tech company, claims AI misdiagnosis led to dog’s death.” July 2026.

The Independent. “Oregon vet sues AI company saying misdiagnosis of cancer tumor led to 11-year-old dog’s death.” July 2026.

[ChatGPT for Academic Researchers]ChatGPT is now offering a free 1-year pro membership for faculty and postdoctoral rese...
07/30/2026

[ChatGPT for Academic Researchers]

ChatGPT is now offering a free 1-year pro membership for faculty and postdoctoral researchers in eligible scientific fields at eligible institutions (Texas A&M University is on the list).

Applicants must provide a paper posted on arXiv, bioRxiv, or ChemRxiv within the past three years that lists them as an author, along with a brief description of their intended research use.

Once approved, the faculty can add up to 4 verified students or collaborators from the same institution. Content you share in your ChatGPT for Academic Researchers workspace is not used to train OpenAI’s models by default.

I have submitted my application, and it is currently under review. Not sure how long it will take.

Submit your application today!

New website alert! Research, publications, veterinary AI education resources, VetClinPathGPT, and speaking topics, all i...
07/26/2026

New website alert!

Research, publications, veterinary AI education resources, VetClinPathGPT, and speaking topics, all in one place.

Take a look and tell me what is missing. I would love to hear from you!

Celebrating the first publication of my PhD student, Yumi Chang, in Veterinary Pathology. 🎉“Applying artificial intellig...
07/16/2026

Celebrating the first publication of my PhD student, Yumi Chang, in Veterinary Pathology. 🎉

“Applying artificial intelligence to automate clinical data extraction from veterinary electronic health records: A practical guidance and comparative scoping review”

Veterinary EHRs hold enormous research value, but almost all of it is locked inside unstructured free text.

Turning that text into structured, analyzable data is still slow, manual, and standard-free. In this new open-access review, we asked: how are AI tools actually being used to automate clinical data extraction from unstructured EHRs in veterinary and human medicine?

👉 We screened 5,796 original research papers across PubMed, CAB Abstracts, Web of Science, and ACL Anthology

👉 Included 54 studies (23 veterinary, 31 human) on AI-based clinical data extraction

👉 Compared where the two fields diverge: veterinary work leans on large supervised datasets, while human medicine increasingly uses prompt-based LLMs that need far fewer annotated examples

In this article, we provided:

🎯 A step-by-step framework covering data preparation, platform and privacy decisions, and prompt engineering, plus a worked case study extracting 62 variables from canine pancreatitis records (Fig. 5)

🎯 An annotated prompt template built from the components that actually improved performance in the literature (Table 6)

This article is written specifically to help veterinary researchers adopt these tools without heavy coding or infrastructure.

Getting ready for   ! American Veterinary Medical Association (AVMA)Don’t miss out my AI talk “Less pAIn, More gAIn: A S...
07/11/2026

Getting ready for !

American Veterinary Medical Association (AVMA)

Don’t miss out my AI talk “Less pAIn, More gAIn: A Starter Guide to AI Literacy for Veterinary Professionals” on July 12 (Sunday) 1:00PM at Room 257. ❤️

I’m speaking at AVMA Convention 2026 in Anaheim!If AI feels more like pAIn than gAIn right now, this session is for you....
07/07/2026

I’m speaking at AVMA Convention 2026 in Anaheim!

If AI feels more like pAIn than gAIn right now, this session is for you.

Less pAIn, More gAIn: A Starter Guide to AI Literacy for Veterinary Professionals (hosted by IDEXX)

🗓️ Sunday, July 12
🕐 1:00 - 1:50 PM PDT
📍 Room 257, Anaheim Convention Center

What you’ll walk away with:
✅ What AI literacy actually means for veterinary professionals (no coding required)
✅ How to critically evaluate AI tools before bringing them into your practice
✅ Practical first steps to turn AI anxiety into AI confidence

Whether you’re AI-curious or AI-skeptical, come join the conversation.

See you in Anaheim! 🌴

Address

660 Raymond Stotzer Pkwy
College Station, TX
77843

Alerts

Be the first to know and let us send you an email when Vet Clin Path Professor: Dr. Candice Chu posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Shortcuts

Share

Category