hemepath.ai

What It Is

HemeAI Lab is an AI and computational pathology lab under Memorial Sloan Kettering Cancer Center, focused on hematopathology. Its goal is to improve diagnosis and treatment outcomes for patients with leukemia and other blood and bone marrow disorders through real-world datasets, deep learning models, and clinically deployable software. The core tools shown on the site include DeepHeme, HemeSeg, DeepHemeClinical, HemeTeacher, HemeLabel, HemeParser, and HemeRegion.

Core Capabilities

Based on the site’s description, DeepHeme is its flagship model: a general-purpose, multi-center, pathologist-level cell image classification algorithm for human bone marrow aspirates. It can distinguish 23 cell classes and generalize across institutions. HemeSeg handles cell segmentation in digitized bone marrow slides; HemeRegion detects and classifies regions of interest; HemeLabel is used for rapid annotation of bone marrow and peripheral blood images; and HemeParser extracts information from clinical hematopathology reports. Overall, the toolchain covers multiple stages, including data annotation, image recognition, clinical information extraction, and teaching/training.

Pricing and Trial

The site offers a DeepHeme Demo, allowing users to test the algorithm with sample region images from UCSF and MSK, as well as upload their own region images. However, the page does not disclose any free quota, trial limits, commercial pricing, procurement process, or payment methods. At this stage, it looks more like a research and clinical-translation showcase than a mature SaaS product page.

Pros and Cons

Its strengths lie in its highly specialized use case, its backing by MSKCC and collaborators such as UCSF, UC Berkeley, and Harvard, and its explicit emphasis on cross-institution generalization. The toolchain also goes beyond a single classifier, covering annotation, segmentation, report parsing, and education. The main drawback is that publicly available information is incomplete: there is no clear explanation of API availability, PACS/LIS/EHR integration, data privacy policies, rules for handling uploaded images, regulatory approval status, or detailed performance metrics. Its boundaries for real-world clinical deployment still need further verification.

Who It’s For and Access from China

It is better suited for hematopathology departments, cancer centers, computational pathology research teams, medical AI developers, and educational institutions for evaluation and research purposes. Access from China is currently unknown. The site is in English, and there is no visible Chinese interface, Chinese report processing, or local payment information. For use in a Chinese clinical environment, users would need to carefully verify network accessibility, cross-border data compliance, hospital system integration capabilities, and domestic medical device regulatory requirements.