Hybrid Graph Mamba Framework for Histopathological Image Classification
Journal of Trends in Computer Science and Smart Technology · 24.04.2026 · 8 (2) · 304-323
DOI: 10.36548/jtcsst.2026.2.006 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Trends in Computer Science and Smart Technology, online erschienen am 24.04.2026. Verfasst von Dave Machnaim und 4 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 35 Literaturangaben.
Abstract (Originalsprache)
Digital pathology has already been shown to be useful in disease diagnosis; however, the computational power required to use high-resolution digital pathology is substantial, making it unrealistic in many clinical laboratories due to limitations in hardware capabilities. Accordingly, this paper presents a new architecture, MFLA-Graph-Mamba-NCA, which addresses these limitations. The main point of MFLA-Graph-Mamba-NCA is to develop a lightweight model for tissue classification while maintaining speed and optimal accuracy. Four complementary learning paradigms are incorporated into the architecture: a morphological frontend, a frequency-domain branch based on Functional Kernel Adaptive Networks (FunKAN), feature stabilization over time using Liquid Neural Ordinary Differential Equations, and global sequence modelling using a Graph-based Mamba module. The model was tested on a harmonized benchmark of 125,000 images, based on the NCT-CRC-HE-100K and LC25000 datasets, and divided into 12 different biological classes to expressly avoid cross-dataset lab leakage. It has been shown to provide 99.89%accuracy and a Macro: F1 score of 99.78%. These techniques were condensed to 1.46 million parameters, forming a hybridized model that can be deployed to standard clinical workstations.
Zitieren
APA
Machnaim, D., Jaison, S., Sundar, N., D., N. N., & Midhun, S. (2026). Hybrid Graph Mamba Framework for Histopathological Image Classification. Journal of Trends in Computer Science and Smart Technology, 8(2), 304-323. https://doi.org/10.36548/jtcsst.2026.2.006
Vancouver
Machnaim D, Jaison S, Sundar N, D. NN, Midhun S. Hybrid Graph Mamba Framework for Histopathological Image Classification. Journal of Trends in Computer Science and Smart Technology. 2026;8(2):304-323. doi:10.36548/jtcsst.2026.2.006
BibTeX
@article{Machnaim2026,
author = {Machnaim, Dave and Jaison, Shawn and Sundar, Naveen and D., Narmadha Naveen and Midhun, Sai},
title = {Hybrid Graph Mamba Framework for Histopathological Image Classification},
journal = {Journal of Trends in Computer Science and Smart Technology},
year = {2026},
volume = {8},
number = {2},
pages = {304--323},
doi = {10.36548/jtcsst.2026.2.006}
}
EUVIMED