Biology subjects

Toth, T.

Publications and source records attributed to Toth, T..

8 recordsLinked to original sources

bioRxiv2025

Microenvironmental information significantly improves the recognition of cell types in human lung cancer patients

Accurate single-cell phenotypic classification in histopathological tissue sections is essential for understanding tumor behavior, identifying potential therapeutic targets, and improving prognostic assessments in cancer research. In this study, we applied deep learning to classify lung cancer cell phenotypes in hematoxylin and eosin-stained tissue sections. Using 11 whole slide images from 11 patients, we annotated nearly 20,000 cells into seven distinct phenotypes for training and validation. We used a fisheye transformation technique, which modifies images to mimic fisheye camera effects in order to incorporate cellular microenvironment information to enhance deep learning models. We evaluated its effectiveness on lung cancer tissue sections, optimizing transformation parameters and assessing classification performance through multiple cross-validation strategies. Our results demonstrate that the transformation significantly improves classification accuracy, approaching human level performance, particularly for phenotypes that rely on subtle morphological differences. The approach enhances model generalizability across patient samples, highlighting the importance of integrating spatial context in computational pathology. These findings suggest that incorporating adaptive image transformations can significantly improve automated histopathological analysis, with potential implications for more robust and clinically applicable AI-driven diagnostics.

bioinformatics↗

bioRxiv2024

HCS-3DX, a next-generation AI-driven automated 3D high-content screening system

Self-organized three-dimensional (3D) cell cultures, collectively called 3D-oids, include spheroids, organoids and other co-culture models. Systematic evaluation of these models forms a critical new generation of high-content screening (HCS) systems for patient-specific drug analysis and cancer research. However, the standardisation of working with 3D-oids remains challenging and lacks convincing implementation. This study develops and tests HCS-3DX, a next-generation system revolutionising HCS analysis in 3D imaging and image evaluation. HCS-3DX is based on three main components: an automated Artificial Intelligence (AI)-driven micromanipulator for 3D-oid selection, an HCS foil multiwell plate for optimised imaging, and image-based AI software for single-cell data analysis. We validated HCS-3DX directly on 3D tumour models, including tumour-stroma co-cultures. Our data demonstrate that HCS-3DX achieves a resolution that overcomes the limitations of current systems and reliably and effectively performs 3D HCS at the single-cell level. Its application will enhance the accuracy and efficiency of drug screening processes, support personalised medicine approaches, and facilitate more detailed investigations into cellular behaviour within 3D structures.

bioinformatics↗

bioRxiv2024

When the pen is mightier than the sword: semi-automatic 2 and 3D image labelling

Data is the driving engine of learning-based algorithms, the creation of which fundamentally determines the performance, accuracy, generalizability and quality of any model or method trained on it. When only skilled or trained personnel can create reliable annotations, assisted software solutions are desirable to reduce the time and effort the expert must spend on labelling. Herein is proposed an automated annotation helper software package in napari that offers multiple methods to assist the annotator in creating object-based labels on 2D or 3D images.

bioinformatics↗

bioRxiv2022

Show me your neighbour and I tell what you are: fisheye transformation for deep learning-based single-cell phenotyping

Recently we have concluded that image-based features derived from the microenvironment have an enormous impact on successfully determining the class of an object1. Here we demonstrate that deep learning-based phenotypic analysis of cells with a properly chosen microenvironment-size provides results comparable to our earlier neighbourhood-based methods that utilise hand-crafted image features. We hypothesised that treating cells with equal weight, regardless of their position within the cellular microenvironment, is suboptimal, and direct neighbours have a larger impact on the phenotype of the cell-of-interest than cells in its larger proximity. Hence we present a novel approach that (1) considers the fully featured view of the cell-of-interest, (2) includes the neighbourhood and (3) gives lesser weight to cells that are far from the cell. To achieve this, we present a transformation similar to those characteristic for fisheye cameras. Such a transformation satisfies all the above defined criteria, with a fast rate of transform for any images. Using the proposed transformation with proper settings we could significantly increase the accuracy of single-cell phenotyping, both in case of cell culture and tissue-based microscopy images. The range of potential applications of the proposed method goes beyond microscopy, as we present improved results on the iWildCam 2020 dataset containing images of wild animals.

bioinformatics↗

bioRxiv2019

3D-Cell-Annotator: an open-source active surface tool for single cell segmentation in 3D microscopy images

SummarySegmentation of single cells in microscopy images is one of the major challenges in computational biology. It is the first step of most bioimage analysis tasks, and essential to create training sets for more advanced deep learning approaches. Here, we propose 3D-Cell-Annotator to solve this task using 3D active surfaces together with shape descriptors as prior information in a fully- and semi-automated fashion. The software uses the convenient 3D interface of the widely used Medical Imaging Interaction Toolkit (MITK). Results on 3D biological structures (e.g. spheroids, organoids, embryos) show that the precision of the segmentation reaches the level of a human expert.\n\nAvailability and implementation3D-Cell-Annotator is implemented in CUDA/C++ as a patch for the segmentation module of MITK. The 3D-Cell-Annotator enabled MITK distribution can be downloaded at: www.3D-cell-annotator.org. It works under Windows 64-bit systems and recent Linux distributions even on a consumer level laptop with a CUDA-enabled video card using recent NVIDIA drivers.\n\nContactsfilippo.piccinini@irst.emr.it and horvath.peter@brc.mta.hu

bioinformatics↗

bioRxiv2019

A deep learning framework for nucleus segmentation using image style transfer

Single cell segmentation is typically one of the first and most crucial tasks of image-based cellular analysis. We present a deep learning approach aiming towards a truly general method for localizing nuclei across a diverse range of assays and light microscopy modalities. We outperform the 739 methods submitted to the 2018 Data Science Bowl on images representing a variety of realistic conditions, some of which were not represented in the training data. The key to our approach is to adapt our model to unseen and unlabeled data using image style transfer to generate augmented training samples. This allows the model to recognize nuclei in new and different experiments without requiring expert annotations.

bioinformatics↗

bioRxiv2018

Unraveling intra- and intersegmental neuronal connectivity between central pattern generating networks in a multi-legged locomotor system

Animal walking results from a complex interplay of central pattern generating networks (CPGs), local sensory signals expressing position, velocity and forces generated in the legs, and coordinating signals between neighboring ones. In the stick insect, in particular, intra- and intersegmental coordination is conveyed by these sensory signals. The rhythmic activity of the CPGs, hence of the legs, can be modified by the aforementioned sensory signals. However, the precise nature of the interaction between the CPGs and these sensory signals has remained largely unknown. Experimental methods aiming at finding out details of these interactions, often apply the muscarinic acetylcholine receptor agonist, pilocarpine in order to induce rhythmic activity in the CPGs, hence in the motoneurons of the segmental ganglia. Using this general approach, we removed the influence of sensory signals and investigated the putative connections between CPGs associated with the coxa-trochanter (CTr)-joint in the different segments (legs) in more detail. The experimental data underwent phase-difference analysis and Dynamic Causal Modelling (DCM). These methods can uncover the underlying coupling structure and strength between pairs of segmental ganglia (CPGs). We set up different coupling schemes (models) for DCM and compared them using Bayesian Model Selection (BMS). Models with contralateral connections in each segment and ipsilateral connections on both sides, as well as the coupling from the meta- to the ipsilateral prothoracic ganglion were preferred by BMS to all other types of models tested. Moreover, the intrasegmental coupling strength in the mesothoracic ganglion was the strongest and most stable in all three ganglia.

neuroscience↗

bioRxiv2017

Show me your neighbours, and I’ll tell you what you are – cellular microenvironment matters

To answer major questions of cell biology, it is essential to understand cellular complexity. Modern automated microscopes produce vast amounts of images routinely, making manual analysis nearly impossible. Due to their efficiency, machine learning-based analysis software have become essential tools to perform single-cell-level phenotypic analysis of large imaging datasets. However, an important limitation of such methods is that they do not use the information gained from the cellular micro- and macroenvironment: the algorithmic decision is based solely on the local properties of the cell of interest. Here, we present how various microenvironmental features contribute to identifying a cell and how such additional information can improve single-cell-level phenotypic image analysis. The proposed methodology was tested for different sizes of Euclidean and nearest neighbour-based cellular environments both on tissue sections and cell cultures. Our experimental data verify that the microenvironment of a cell largely determines its entity. This effect was found to be especially strong for established tissues, while it was somewhat weaker in the case of cell cultures. Our analysis shows that combining local cellular features with the properties of the cell's microenvironment significantly improves the accuracy of machine learning-based phenotyping.

bioinformatics↗