Principal component-based image segmentation: a new approach to outline <i>in vitro</i> cell colonies

Arous, Delmon; Schrunner, Stefan; Hanson, Ingunn; Edin, Nina Frederike Jeppesen; Malinen, Eirik

doi:10.1080/21681163.2022.2035822

Cited by 3 publications

(3 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Automating the identification, segmentation, and counting of stained vitro cell colonies involve challenges with background noise and contaminations. Arous et al (2022) presented a machine learning procedure to amend these issues by characterizing, extracting, and segmenting inquired cell colonies using principal component analysis, k-means clustering, and a modified watershed segmentation algorithm to identify visible colonies automatically. The proposed segmentation algorithm was tested on two data sets: a T-47D (proprietary) cell colony and a bacteria (open source) data set.…”

Section: Other Image-processing Techniquesmentioning

confidence: 99%

Automated Bacteria Colony Counting using Hybrid Image Segmentation Algorithm and YOLOv5 Transfer Learning Model

Galope,

Lisondra,

Nanual

2024

International Conference on Innovative Practices in Management, Engineering &Amp; Social Sciences

View full text Add to dashboard Cite

Bacteria colony counting is a critical process in microbiology research, but manual colony counting remains tedious and error-prone, motivating the need for automation. This study aimed to develop an automated system for accurately detecting and counting E. coli colonies on agar plates. The research objectives were achieved by generating a dataset of E. coli colony images, developing a hybrid image segmentation algorithm, and training a YOLOv5 transfer learning model. The dataset was created by capturing images of E. coli colonies on agar plates under controlled conditions. The cultured agar plates were generated by Davao Oriental State University (DOrSU). A novel framework for a hybrid image segmentation algorithm combining Watershed and Falling-Ball was developed to address the challenge of accurately segmenting colonies from complex backgrounds. The algorithm utilized the output of the Watershed Algorithm to create a binary mask, which the Falling-Ball Algorithm further refined to improve edge detection and fill gaps. The YOLOv5 transfer learning model was trained using the generated dataset to detect and count E. coli colonies. The model achieved a detection accuracy of up to 75%, providing a reliable automated solution for colony counting. Performance evaluation metrics such as precision, recall, and mAP_0.5 were utilized to assess the model's performance. However, training the model using the dataset that underwent the framework could not proceed due to its resource-intensive requirements.

show abstract

Section: Other Image-processing Techniquesmentioning

confidence: 99%

Automated Bacteria Colony Counting using Hybrid Image Segmentation Algorithm and YOLOv5 Transfer Learning Model

Galope,

Lisondra,

Nanual

2024

International Conference on Innovative Practices in Management, Engineering &Amp; Social Sciences

View full text Add to dashboard Cite

show abstract

“…A novel machine learning approach using principal component-based watershed method for image segmentation was developed to identify individual colonies in a cell flask (Arous et al 2022). This automated algorithm was applied to scanned images of the cell culture flasks containing fixed and stained A549 cell colonies to locate the centroid of viable colonies.…”

Section: Colony Segmentation and Localizationmentioning

confidence: 99%

“…The colony segmentation procedure was previously developed and validated for sparsely seeded and unirradiated cells (Arous et al 2022). As the current work included densely seeded and irradiated cells, it was necessary to conduct a new, supervised calibration of the segmentation procedure.…”

Section: Colony Segmentation and Localizationmentioning

confidence: 99%

2D mapping of radiation dose and clonogenic survival for accurate assessment of in vitro X-ray GRID irradiation effects

Arous¹,

Lie²,

Håland³

et al. 2023

Phys. Med. Biol.

Self Cite

View full text Add to dashboard Cite

Spatially fractionated radiation therapy (SFRT or GRID) is an approach to deliver high local radiation doses in an ’on-off’ pattern. To better appraise the radiobiological effects from GRID, a framework to link local radiation dose to clonogenic survival needs to be developed. A549 lung cancer cells were irradiated in T25 cm2 flasks using 220 kV X-rays with an open field or through a tungsten GRID collimator with periodical 5 mm openings and 10 mm blockings. Delivered nominal doses were 2, 5, and 10 Gy. A novel approach for image segmentation was used to locate the centroid of surviving colonies in scanned images of the cell flasks. GafchromicTM film dosimetry (GFD) and FLUKA Monte Carlo (MC) simulations were employed to map the dose at each surviving colony centroid. Fitting the linear-quadratic (LQ) function to clonogenic survival data for open field irradiation, the expected survival level at a given dose level was calculated. The expected survival levels were then mapped together with the observed levels in the GRID-irradiated flasks. GFD and FLUKA MC gave similar dose distributions, with a mean peak-to-valley dose ratio of about 5. LQ-parameters for open field irradiation gave α = 0.24 ± 0.02 Gy-1 and β = 0.019 ± 0.002 Gy-2. The mean relative percentage deviation (RPD) between observed and predicted survival in the (peak ; valley) dose regions was (4.6 ; 3.1) %, (26.6 ; -1.0) %, and (129.8 ; -2.3) % for 2, 5 and 10 Gy, respectively. In conclusion, a framework for mapping of surviving colonies following GRID irradiation together with predicted survival levels from homogeneous irradiation was presented. For the given cell line, our findings indicate that GRID irradiation causes reduced survival in the peak regions compared to an open field configuration.

show abstract

Principal component-based image segmentation: a new approach to outline in vitro cell colonies

Arous

Schrunner

Hanson

et al. 2022

Computer Methods in Biomechanics and Biomedical Engineering: Im

View full text Add to dashboard Cite

Principal component-based image segmentation: a new approach to outline in vitro cell colonies

Cited by 3 publications

References 37 publications

Automated Bacteria Colony Counting using Hybrid Image Segmentation Algorithm and YOLOv5 Transfer Learning Model

Automated Bacteria Colony Counting using Hybrid Image Segmentation Algorithm and YOLOv5 Transfer Learning Model

2D mapping of radiation dose and clonogenic survival for accurate assessment of in vitro X-ray GRID irradiation effects

Principal component-based image segmentation: a new approach to outline in vitro cell colonies

Contact Info

Product

Resources

About