HemBLIP: A Vision-Language Model for Interpretable Leukemia Cell Morphology Analysis

TL;DR

HemBLIP is a novel vision-language model that provides interpretable descriptions of leukemia cell morphology, improving diagnostic transparency and accuracy in blood cell analysis.

Contribution

We developed HemBLIP, a new vision-language model for blood cell analysis, with a specialized dataset and efficient training methods, enhancing interpretability and performance.

Findings

HemBLIP outperforms existing models in caption quality and morphological accuracy.

LoRA adaptation reduces computational cost while maintaining high performance.

The model demonstrates potential for transparent hematological diagnostics.

Abstract

Microscopic evaluation of white blood cell morphology is central to leukemia diagnosis, yet current deep learning models often act as black boxes, limiting clinical trust and adoption. We introduce HemBLIP, a vision language model designed to generate interpretable, morphology aware descriptions of peripheral blood cells. Using a newly constructed dataset of 14k healthy and leukemic cells paired with expert-derived attribute captions, we adapt a general-purpose VLM via both full fine-tuning and LoRA based parameter efficient training, and benchmark against the biomedical foundation model MedGEMMA. HemBLIP achieves higher caption quality and morphological accuracy, while LoRA adaptation provides further gains with significantly reduced computational cost. These results highlight the promise of vision language models for transparent and scalable hematological diagnostics.