What this course is, and what it is built on

This is an educational demonstration. It is not a medical device, it is not registered as one anywhere, and nothing it shows may be used to diagnose, screen or manage a patient. The patches are anonymised research images from a published dataset, and the categories beside them are the expert labels the dataset carries, not a second opinion on a living person.

The model is a logistic regression fitted in 1993 on 569 patients measured with a different camera at an unknown pixel scale. Its call on a modern scan is an illustration of what nuclear measurements carry, and no more than that.

Each objective is testable inside the app, which is why the Progress page can tell you whether the course worked.

1. 1Say what an adequate breast FNA smear is, and recognise one that is not.
2. 2Read the four things a cytologist reads first at low power: cellularity, cohesion and architecture, the second population, and the background itself.
3. 3Read nuclei at high power, and say which of those features a measurement can capture and which it cannot.
4. 4Place a patch in C1 to C5, and say the risk of malignancy and the management that follows.
5. 5Name the common benign entities from their pattern, and the malignant ones this dataset holds, using the source laboratory's own report lines as the answer key.
6. 6Screen a whole slide at low power, go to high power in the right places, and reach a category.
7. 7Explain what the OncoScan model measures, why its call can disagree with a cytologist, and when to trust which.

Objectives 2 to 5 sit at ACGME Cytopathology Milestones 2.0, Medical Knowledge 1: Diagnosis. Levels 1 and 2, which ask a trainee to correctly describe cytomorphology and to provide differential diagnosis; locate and categorize cells as normal, reactive, or neoplastic. Objective 6 is the screening skill cytology programmes measure. Objective 7 is the one this course adds.

Fine needle aspiration is 10 to 15 percent of the CT(ASCP) examination, and breast is one of its twelve named sites, per the ASCP Board of Certification, CT(ASCP) examination content guideline. This course is not an examination preparation product and is not affiliated with the ASCP or the ACGME.

CNB should be regarded as a complementary rather than replacement test.

Inside a session the next item comes from the family you are weakest at. A family is one kind of item on one kind of case, such as a five way call on a Papanicolaou malignant patch. Its priority rises when your accuracy there is lower, when your correct answers there are slower, and when longer has passed since you last saw it, and the same family never comes twice in a row.

A family counts as mastered after three correct answers in a row, each inside the target response time, and then drops to about one trial in ten so it is maintained rather than drilled. Speed is scored because fluent reading, not just correct reading, is what separates an experienced screener from a student.

Across days, every family is a card in an open spaced repetition scheduler (ts-fsrs, at a target retention of 0.9). A miss brings the card back soon, a slow correct answer brings it back later, and three fluent answers in a row push it well out. That is all the due today line on the home page is: the cards whose date has come.

Doing a third of the course is a normal state here, not a failure. The trial that inspired this schedule found real gains at a mean completion of 38.6 percent.

Every patch, and every picture in the explainer that is not separately credited, comes from the Multi-Center Breast FNAC Cytology Dataset (Patil, Jain and Sethi), Zenodo record 20763900, CC BY 4.0. It holds 7,393 expert labelled patches from 470 slides of 321 patients, gathered at 13 study sites across India, and it is released under CC BY 4.0.

The cases in this build come from 14 of those sites, credited here by name and nowhere else. A site name never appears on a case, so no hospital can be read as a league table.

• All India Institute of Medical Sciences, Raebareli
• Amrita Hospital, Kochi
• Dr. Balasaheb Vikhe Patil Rural Medical College, Loni
• Government Institute of Medical Sciences,Noida
• Government Medical College and Hospital,Chandigarh
• Government Medical College, Vizianagaram (GMCV)
• Lokmanya Tilak MMC, Mumbai
• Lokmanya Tilak MMC,Mumbai
• Mandya Institute of Medical Sciences, Mandya
• NEIGRIHMS, Meghalaya
• Pandit Bhagwat Dayal Sharma Post Graduate Institute of Medical Sciences,Rohtak
• Rajendra Institute of Medical Sciences,Ranchi
• Regional Institute of Medical Sciences, Imphal
• Sawai Man Singh Medical College,Jaipur

The criteria this course quotes, the evidence its design follows, and the data it runs on.

• Field AS et al. The International Academy of Cytology Yokohama System for Reporting Breast Fine-Needle Aspiration Biopsy Cytopathology. Acta Cytologica 2019;63:257-273.
• Multi-Center Breast FNAC Cytology Dataset (Patil, Jain and Sethi), Zenodo record 20763900, CC BY 4.0.
• Street WN, Wolberg WH, Mangasarian OL. Nuclear feature extraction for breast tumor diagnosis. 1993. Wisconsin Diagnostic Breast Cancer dataset, UCI Machine Learning Repository.
• Modified Masood scoring system for breast fine needle aspiration cytology (six parameters scored 1 to 4).
• ACGME Cytopathology Milestones 2.0, Medical Knowledge 1: Diagnosis.
• ASCP Board of Certification, CT(ASCP) examination content guideline.
• Van Es SL et al. Cytopathology whole slide images and adaptive tutorials for senior medical students: a randomized crossover trial. Diagnostic Pathology 2016.
• Krasne S, Kellman PJ et al. Applying perceptual and adaptive learning techniques for teaching introductory histopathology. Journal of Pathology Informatics 2022.
• SpacED POCUS: a randomized controlled trial of an adaptive spaced education POCUS curriculum for medical students. POCUS Journal 2025.
• Eye tracking in cytotechnology education: visualizing students becoming experts. Journal of the American Society of Cytopathology 2019.