Skip to content
path/blood-smear

Peripheral Blood Smear

Peripheral blood smear read: classified white cells, differential percentages and a blast alert.

Coming soonWSIDigital Pathology

This model is not available to run yet. Vote to show interest and get an email when it opens.

Overview

What it does

Peripheral Blood Smear reads the nucleated cells of a stained blood film. Smear review follows an abnormal complete blood count or an analyser flag, and its most important job is safety: not missing blasts or abnormal lymphocytes among otherwise ordinary cells.

The run locates and classifies each nucleated cell as neutrophil, lymphocyte, monocyte, eosinophil, basophil, immature granulocyte, blast, erythroblast or atypical cell, and turns the counts into a differential in percent. Every cell appears in a gallery grouped by class, so the reviewer sees the cells behind each number. When blasts or abnormal lymphocytes are found, the result raises an alert with "refer for haematology review" guidance.

Intended use

Decision support for the laboratory morphology review: a pre-classified differential with every cell in view, and a safety net that puts suspicious smears first. Red-cell and platelet morphology are not graded by this model.

Who it is for

Haematology laboratory scientists and technologists, haematologists, clinical pathologists, and labs without a digital morphology analyser. Developers use the run to add a structured differential and an abnormal-cell alert to a laboratory information system.

Inputs and protocol

Accepted input

Two kinds of input describe the same smear:

  • Whole-slide image of the monolayer of a Wright- or May-Grünwald-Giemsa-stained smear, as DICOM VL Whole Slide Microscopy Image or a scanner file (SVS, NDPI, MRXS, CZI, OME-TIFF and others).
  • Single-cell images exported by a digital morphology analyser, as JPEG, PNG or TIFF crops of individual cells.

Requirements

Requirement Why
Romanowsky-type stain (Wright, MGG) Granules, nuclear chromatin and cytoplasm colour separate the cell classes
Readable monolayer on whole slides Cells are classified where they lie singly, not in thick or feathered areas
Pixel spacing on whole slides Cell size helps separate blasts from lymphocytes and monocytes

Optional context

The full blood count, analyser flags, age and the clinical question can be passed in clinical_context; the guidance reads the differential against them.

Outputs and standards

The result

Section Content
WBC differential Class of every nucleated cell, differential percentages, cell gallery by class and the blast or abnormal-cell alert
Draft report Opt-in (options.report): an English narrative that interprets the result for the reader (findings, impression, limitations); every number is checked against the findings

Counts and percentages sit in measurements; each gallery cell keeps a link to its position on the slide or to its source image.

Standards

  • Morphology: ICSH 2015 recommendations for the standardisation of nomenclature and grading of peripheral blood cell morphological features.
  • DICOM: classified cells as Microscopy Bulk Simple Annotations on whole slides, the differential in SR TID 1500, an Encapsulated PDF.
  • Interoperability: FHIR R4 DiagnosticReport with one LOINC-coded Observation per cell class.

The blast alert

The alert names the cells classed as blasts or abnormal lymphocytes, shows them in their own gallery and carries the "refer for haematology review" guidance.

Works with

When blasts are flagged, Bone Marrow Aspirate (path/bone-marrow) reads the follow-up aspirate with a full myelogram and blast percentage.

Result sections

One run returns every section its input supports.

  1. WBC differentialwbc-differential
  2. Draft reportreport
Explore all Digital Pathology models

Prostate Biopsy Pathology

Prostate core biopsy read: cancer per core, Gleason and ISUP grade, % pattern 4, tumour length.

Pelvis4 sections
WSI· wsi

Any-Organ H&E Slide Analysis

Organ-agnostic H&E read: slide QC, tumour map, nuclei and mitoses, primary-site ranking, embeddings.

Whole body8 sections
WSI· wsi

Breast Pathology

Breast case read from H&E, IHC and node slides: type, grade, biomarker scores, TILs and pN.

Breast8 sections
WSI· wsi

Colorectal Pathology

Colorectal polyp and cancer read: adenoma or serrated histology, dysplasia, MSI, tissue map.

GI4 sections
WSI· wsi

Lung Cancer Pathology

Lung biopsy and resection read: NSCLC subtype, IASLC grade, PD-L1 TPS and EGFR mutation probability.

Chest4 sections
WSI· wsi

Bone Marrow Aspirate

Marrow aspirate read: cell-by-cell myelogram, blast percentage, M:E ratio and dysplasia flags.

Whole body2 sections
WSI· wsi