Determining the scale of image patches using a deep learning approach

Otálora, Sebastian; Perdómo, Oscar; Atzori, Manfredo; Andersson, Mats; Jacobsson, Ludwig; Hedlund, Martin; Müller, Henning

doi:10.1109/isbi.2018.8363703

Cited by 5 publications

(5 citation statements)

References 9 publications

(17 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several new and personalized deep learning models have been proposed by researchers to classify histo-pathological images of cancers of the following organs: bone [2][3], brain [4][5], Breast [6][7] [8]), Colon [9][10] [11][12], Lung [13], Gastric [14], Pancreas [15] and Prostate [16].…”

Section: Discussionmentioning

confidence: 99%

Classification of histological images of thyroid nodules based on a combination of Deep Features and Machine Learning

BELLAL,

SAIM,

BENAHMED

et al. 2023

Med. Technol. J.

View full text Add to dashboard Cite

Background: Thyroid nodules are a prevalent worldwide disease with complex pathological types. They can be classified as either benign or malignant. This paper presents a tool for automatically classifying histological images of thyroid nodules, with a focus on papillary carcinoma and follicular adenoma. Methods: In this work, two pre-trained Convolutional Neural Network (CNN) architectures, VGG16 and VGG19, are used to extract deep features. Then, a principal component analysis was used to reduce the dimensionality of the vectors. Then, three machine learning algorithms (Support Vector Machine, K-Nearest Neighbor, and Random Forest) were used for classification. These investigations were applied to our database collection, Results: The proposed investigations have been applied to our private database collection with a total of 112 histological images. The highest results were obtained by the VGG16 transfer deep feature and the SVM classifier with an accuracy rate equal to 100%.

show abstract

Section: Discussionmentioning

confidence: 99%

Classification of histological images of thyroid nodules based on a combination of Deep Features and Machine Learning

BELLAL,

SAIM,

BENAHMED

et al. 2023

Med. Technol. J.

View full text Add to dashboard Cite

show abstract

“…The scale detector tool is a CNN trained to estimate the magnification level of a given patch or image. This task has been explored in the past Otálora et al (2018a), Otálora et al (2018b) in the prostate and breast tissue types. Similar approaches have been recently extended to different organs in the TCGA repository Zaveri et al (2020).…”

Section: Scale Detectormentioning

confidence: 99%

“…CNNs can identify abnormalities in tissues, but the information and the features related to the abnormalities are not the same for each scale representation (Jimenez-del Toro et al, 2017). Therefore, the proper scale must be selected to train CNNs (Gecer et al, 2018;Otálora et al, 2018b). Unfortunately, scale information is not always available into images.…”

Section: Introductionmentioning

confidence: 99%

Multi_Scale_Tools: A Python Library to Exploit Multi-Scale Whole Slide Images

Marini

Otálora

Podareanu

et al. 2021

Front. Comput. Sci.

Self Cite

View full text Add to dashboard Cite

Algorithms proposed in computational pathology can allow to automatically analyze digitized tissue samples of histopathological images to help diagnosing diseases. Tissue samples are scanned at a high-resolution and usually saved as images with several magnification levels, namely whole slide images (WSIs). Convolutional neural networks (CNNs) represent the state-of-the-art computer vision methods targeting the analysis of histopathology images, aiming for detection, classification and segmentation. However, the development of CNNs that work with multi-scale images such as WSIs is still an open challenge. The image characteristics and the CNN properties impose architecture designs that are not trivial. Therefore, single scale CNN architectures are still often used. This paper presents Multi_Scale_Tools, a library aiming to facilitate exploiting the multi-scale structure of WSIs. Multi_Scale_Tools currently include four components: a pre-processing component, a scale detector, a multi-scale CNN for classification and a multi-scale CNN for segmentation of the images. The pre-processing component includes methods to extract patches at several magnification levels. The scale detector allows to identify the magnification level of images that do not contain this information, such as images from the scientific literature. The multi-scale CNNs are trained combining features and predictions that originate from different magnification levels. The components are developed using private datasets, including colon and breast cancer tissue samples. They are tested on private and public external data sources, such as The Cancer Genome Atlas (TCGA). The results of the library demonstrate its effectiveness and applicability. The scale detector accurately predicts multiple levels of image magnification and generalizes well to independent external data. The multi-scale CNNs outperform the single-magnification CNN for both classification and segmentation tasks. The code is developed in Python and it will be made publicly available upon publication. It aims to be easy to use and easy to be improved with additional functions.

show abstract

“…using the Keras 18 Deep Learning library with the TensorFlow 19 backend in order to train a model for automatically detecting the scale of histopathology image patches from the biomedical literature. It is inspired by [26] and works by predicting the average area occupied by cell nuclei within the patch to determine its magnification level. The code runs both on CPU-only systems, as well as hosts configured for GPU computation.…”

Section: B Computational Taskmentioning

confidence: 99%

Distributed Container-Based Evaluation Platform for Private/Large Datasets

Eggel

Schaer

Müller

2018

2018 17th International Symposium on Parallel and Distributed Computing (ISPDC)

Self Cite

View full text Add to dashboard Cite

The rise of big data and artificial intelligence techniques such as deep learning has lead to an exponential increase in stored data in various fields, including medical imaging, genetics and financial trading. Sharing these increasing amounts of data for research is challenging, as privacy risks increase with the increased size of data. Physically moving very large datasets to researchers is inconvenient, as download or sending physical hard disks are not optimal. Research on sensitive data is often not possible, as sharing is not legal. The popularity of containerbased technologies such as Docker has revolutionized the way applications are deployed, due to their self-sufficient, lightweight and portable nature. In this paper, we propose a novel distributed platform using containers for simple execution and evaluation of research applications on the data owner's infrastructure, bringing the algorithms to the data. This approach avoids the cumbersome transfer of large datasets and can help circumventing problems linked to non-shareable data by providing a sandboxed execution environment with read-only access to the data. At no point the data leave the data owner's site, giving researchers access to their evaluation results, only, and not the data themselves. The presented proof-of-concept confirms the feasibility of a distributed container-based evaluation platform for large and/or sensitive data. This has several advantages, including execution of code instead of submission of result files and availability of otherwise inaccessible data. The container architecture allows for minimal computational overhead, no software dependency management on the infrastructure, distributed runtime environment and isolation of processes from the underlying host system. A version addressing various identified architectural and security-related challenges has the potential to be deployed in a production setting and therefore allows researchers to gain insights from previously inaccessible data. One goal is to target hospitals with increasingly strong local infrastructure for storage and computation, needed for artificial intelligence based decision support (genetics and imaging).

show abstract

Determining the scale of image patches using a deep learning approach

Cited by 5 publications

References 9 publications

Classification of histological images of thyroid nodules based on a combination of Deep Features and Machine Learning

Classification of histological images of thyroid nodules based on a combination of Deep Features and Machine Learning

Multi_Scale_Tools: A Python Library to Exploit Multi-Scale Whole Slide Images

Distributed Container-Based Evaluation Platform for Private/Large Datasets

Contact Info

Product

Resources

About