Pamphlet style options

BCMG AI Services and Advertising

Five static HTML directions for a biospatial AI services pamphlet. Each version uses local assets from the surrounding projects and keeps the content focused on pathology, tissue imaging, IHC, mIF, hemepath, wound measurement, and WSI workflows.

01 Clinical One-Pager Clean, medical-services positioning with evidence blocks. 02 Executive Dark Premium, image-heavy narrative for sponsor conversations. 03 Services Grid Direct menu of offerings and project examples. 04 Foldout Brochure Three-panel print-style pamphlet layout. 05 Technical Catalog Capability matrix for technical buyers and collaborators.

Local project sources consulted include BRAF thyroid, MSI colorectal/endometrial, AML/hemepath, MET NSCLC, PH WSI, TROP2/IHC segmentation, FOLR1/TROP2 dashboard work, mIF pseudo-H&E and CD68 salvage, wound tracker, staining lab, BCMG AI curation, BCMG Path Foundation, and SegMen.

Biospatial AI services

BCMG AI Services

Computational pathology, tissue-image intelligence, and human-in-the-loop biomarker workflows for teams working with whole-slide images, IHC, multiplex IF, cytology, flow data, and serial wound imaging.

Whole-slide classifiers
IHC scoring calibration
mIF to pseudo-H&E
Pathologist review tools
Biospatial measurement

Example WSI classifier pipeline: tiling, foundation encoders, multi-instance learning, and biomarker prediction.

0.912BRAF CV AUROC

0.877MSI-H holdout AUROC

0.956AML NPM1 CV AUROC

17/18PH WSI LOO calls

Representative internal research/prototype runs, not clinical performance claims.

Core Services

WSI AI Models

BRAF V600E thyroid, colorectal MSI-H, endometrial MSI pilots, MET NSCLC, pulmonary hypertension WSI, and hematopathology molecular-screen prototypes.

Biomarker Review

TROP2 and FOLR1 workflows, ROI adjustment, dot placement, H-score support, DAB deconvolution, segmentation calibration, and case-screening infrastructure.

Data Operations

Read-only curation of WSI, crop, cytology, hemepath, FCS, text, and molecular folders with static reporting and traceable project ledgers.

Positioning

BCMG AI bridges pathology judgment and production AI: pathologist-supervised datasets, calibrated visual review, rigorous QC, and lightweight local tools that can move from research notebooks to operational workflows.

Project Examples

$TROP2 IHC segmentation comparison$

TROP2 IHC segmentation and DAB scoring with HoVer-Net, SAM, and pathologist feedback.

mIF salvage: registered pseudo-CD68, lineage-negative gating, and per-cell audit panels.

Metric wound tracking with fiducial registration, SAM3-style zones, and serial tissue maps.

Pulmonary hypertension WSI attention region

Pulmonary hypertension WSI attention review with explainable model focus regions.

BCMG AI Services and Advertising

Biospatial intelligence for pathology programs.

BCMG AI builds the practical layer between tissue, pathologist review, machine learning, and sponsor-ready deliverables. The work spans WSI foundation models, IHC quantification, multiplex image rescue, hemepath data fusion, and serial tissue measurement.

452CRC MSI slides staged

93BRAF holdout slides

7,025CD68 audit cells

64TROP2 WSI review cases

Predict

BRAF V600E, MSI-H, MET amplification, PH WSI, and AML NPM1 prototypes using tile embeddings, ABMIL, fusion encoders, and calibrated holdouts.

Quantify

TROP2/FOLR1 IHC workflows, DAB separation, cell segmentation, H-score review, ROI correction, and pathologist-rated segmentation loops.

Rescue

mIF-to-pseudo-H&E, brightfield-to-mIF registration, pseudo-CD68 recovery, threshold audits, QuPath-ready exports, and failed-project salvage.

$TROP2 segmentation comparison$

Whole-slide AI

Classifier development and benchmarking

BRAF thyroid, colorectal MSI-H, MET NSCLC, PH WSI, endometrial MSI pilots, and hemepath molecular screens using UNI, GigaPath, Phikon-v2, and ABMIL/fusion variants.

$IHC segmentation$

IHC operations

Biomarker scoring and review systems

TROP2 and FOLR1 review, ROI adjustment, dot placement, DAB deconvolution, cell segmentation, membrane scoring, H-score support, and batch feedback capture.

mIF rescue

Cross-modality reconstruction

Pseudo-H&E from 8-channel Opal/Vectra images, brightfield IHC to mIF registration, CD68 regional recovery, global threshold audits, and QuPath-ready export.

Biospatial measurement

Serial tissue measurement

Metric wound capture with a 100 x 100 mm fiducial, homography QC, wound-zone segmentation, tissue-change heatmaps, and JSON/report outputs.

QC and normalization

Stain, artifact, and scanner hygiene

Ink artifact filtering, color correction, white-balance models, mask QC, attention-map review, and reportable QC flags for WSI pipelines.

Data foundation

Curation and traceable reporting

Read-only local crawler/indexer for WSI, crops, cytology, hemepath, FCS, text, and molecular folders, plus append-only ledgers and serialized HTML reports.

Delivery formats

Static HTML reports, review dashboards, model summaries, audit panels, curated manifests, CSV/JSON exports, QuPath-ready images, and lightweight local apps for pathologist-in-the-loop iteration.

BCMG AI

Biospatial AI services for pathology and tissue programs.

From whole-slide biomarker models to mIF rescue and pathologist-guided IHC scoring, BCMG AI turns complex tissue data into reviewable, traceable, and sponsor-ready outputs.

Representative research and operations examples: BRAF, MSI, AML NPM1, MET NSCLC, PH WSI, TROP2, FOLR1, mIF CD68 recovery, and wound tracking.

What we build

Services

AI model workflows

Tile extraction, foundation embeddings, ABMIL, fusion modeling, holdout analysis, attention maps, and concise performance reports.

IHC and biomarker review

Case screening, ROI correction, dot placement, H-score support, DAB scoring, membrane quantification, and interactive segmentation evaluation.

Image rescue

Pseudo-H&E generation, cross-modality registration, CD68 imputation, stain deconvolution, QC montages, and QuPath-ready export.

Data infrastructure

Read-only local indexing, clean manifests, curation ledgers, static reports, and reproducible project handoffs.

Proof points

Selected work

BRAF thyroid fusion model: CV AUROC 0.912, 93-slide holdout.
Colorectal MSI-H: full 452-slide cohort, GigaPath holdout AUROC 0.877.
AML NPM1: 102 bags, CV AUROC 0.956, OOF AUROC 0.899.
PH WSI: 18-case LOO run, AUROC 0.9625, 17/18 calls.
mIF CD68 salvage: 7,025 audited cells across 4 regions.

Research/prototype values are presented as internal evidence, not clinical claims.

Technical capability catalog

BCMG AI Services and Advertising

A biospatial AI portfolio organized by data type, model pathway, review interface, and deliverable. Built around local, reproducible workflows that keep pathologist oversight close to the data.

Capability

Inputs

Methods

Outputs

WSI classifier programs

SVS/NDPI/OME-TIFF, labels, splits, feature stores

UNI, GigaPath, Phikon-v2, ABMIL, fusion encoders, holdouts

AUROC/AUPRC reports, attention maps, predictions, manifests

IHC biomarker scoring

TROP2, FOLR1, DAB IHC crops, WSI review tables

Macenko DAB, HoVer-Net, SAM, InstanSeg, pathologist ratings

H-score support, overlays, ROI/dot review, batch summaries

mIF and image-analysis salvage

Vectra/Opal QPTIFF, IHC NDPI/SVS, QuPath exports

Beer-Lambert pseudo-H&E, affine registration, global gating

Pseudo-H&E BigTIFF, pseudo-CD68 fields, cell tables, audit HTML

Biospatial measurement

Phone photos, fiducial cards, serial wound captures

Homography QC, SAM-style segmentation, tissue-zone maps

Area, length, width, perimeter, change heatmaps, JSON reports

mIF reconstruction

$TROP2 segmentation$

IHC segmentation

Metric wound report

Stain normalization