An Annotation-free Whole-slide Training Approach to Pathological Classification of Lung Cancer Types by Deep Neural Network

2020 ◽  
Author(s):  
Chi-Long Chen ◽  
Chi-Chung Chen ◽  
Wei-Hsiang Yu ◽  
Szu-Hua Chen ◽  
Yu-Chan Chang ◽  
...  
Keyword(s):  
Lung Cancer ◽  
Digital Pathology ◽  
Memory Consumption ◽  
Cancer Dataset ◽  
Lung Cancers ◽  
Training Approach ◽  
Cancer Types ◽  
Critical Regions ◽  
Whole Slide Images

Abstract Deep learning for digital pathology is hindered by the extremely high spatial resolution of whole slide images (WSIs). Most studies adopt patch-based methods which, however, require well annotated data for training. These are typically done by laboriously free-hand contouring on the WSI by experts. To both alleviate annotation burdens of experts and enjoy benefits from scaling up amounts of data, we develop a whole-slide training method for entire WSIs to classify types of lung cancers using slide-level diagnoses. Our method leverages unified memory to offload the excessive amount of memory consumption to host memory to train a classifier by entire hundreds-of-million-pixels slides. Experiments were conducted on the lung cancer dataset which contains 9,662 digital slides with various main types. The results showed that the proposed method can achieve an AUC of 0.950 and 0.924 for adenocarcinoma and squamous cell carcinoma on a separate testing set respectively. Furthermore, critical regions highlighted by the class activation map (CAM) technique of our model reveals a high correspondence to cancerous areas annotated by pathologists.

scholarly journals Automated Classification of Lung Cancer Types from Cytological Images Using Deep Convolutional Neural Networks

2017 ◽  
Vol 2017 ◽  
pp. 1-6 ◽  
Author(s):  
Atsushi Teramoto ◽  
Tetsuya Tsukamoto ◽  
Yuka Kiriyama ◽  
Hiroshi Fujita
Keyword(s):  
Lung Cancer ◽  
Cell Carcinoma ◽  
Graphics Processing Unit ◽  
Processing Unit ◽  
Automated Classification ◽  
Deep Convolutional Neural Networks ◽  
Lung Cancers ◽  
Microscopic Images ◽  
Cancer Types

Lung cancer is a leading cause of death worldwide. Currently, in differential diagnosis of lung cancer, accurate classification of cancer types (adenocarcinoma, squamous cell carcinoma, and small cell carcinoma) is required. However, improving the accuracy and stability of diagnosis is challenging. In this study, we developed an automated classification scheme for lung cancers presented in microscopic images using a deep convolutional neural network (DCNN), which is a major deep learning technique. The DCNN used for classification consists of three convolutional layers, three pooling layers, and two fully connected layers. In evaluation experiments conducted, the DCNN was trained using our original database with a graphics processing unit. Microscopic images were first cropped and resampled to obtain images with resolution of 256 × 256 pixels and, to prevent overfitting, collected images were augmented via rotation, flipping, and filtering. The probabilities of three types of cancers were estimated using the developed scheme and its classification accuracy was evaluated using threefold cross validation. In the results obtained, approximately 71% of the images were classified correctly, which is on par with the accuracy of cytotechnologists and pathologists. Thus, the developed scheme is useful for classification of lung cancers from microscopic images.

New microRNA-based diagnostic test for lung cancer classification.

2012 ◽  
Vol 30 (15_suppl) ◽  
pp. 10528-10528
Author(s):  
Ranit Aharonov ◽  
Gila Lithwick Yanai ◽  
Hila Benjamin ◽  
Mats Olot Sanden ◽  
Marluce Bibbo ◽  
...  
Keyword(s):  
Lung Cancer ◽  
Cell Carcinoma ◽  
Expression Profiles ◽  
Lung Tumors ◽  
Small Cell ◽  
Diagnostic Methods ◽  
Expression Levels ◽  
Qrt Pcr ◽  
Cancer Types

10528 Background: Lung cancer is the leading cause of cancer deaths in the US. Treatment options are determined by tumor subtyping, for which there is lack of standardized, objective, and highly accurate techniques. In 20%-30% of cases significant limitations of tumor quantity and quality prevent full classification of the tumor using traditional diagnostic methods. Using microRNA microarray data generated from over a hundred formalin-fixed, paraffin-embedded (FFPE) primary lung cancer samples, we have identified microRNA expression profiles that differ significantly for the main lung cancer types. Based on these findings, we have developed and validated a microRNA-based qRT-PCR assay that differentiates primary lung cancers into four types: squamous cell carcinoma, non-squamous non-small cell lung cancer, carcinoid and small cell carcinoma. Methods: Over 700 primary tumor samples from different histological types of lung cancer were collected. Samples included FFPE blocks from resection or biopsies and cell blocks from cytology specimens including fine needle aspiration, bronchial brushing and bronchial washing. High-quality RNA was extracted from the samples using proprietary protocols. Expression levels of potential microRNA biomarkers were profiled using microarrays followed by a sensitive and specific qRT-PCR platform. An assay for lung tumors classification using 8 microRNAs on qRT-PCR was developed based on data from 261 samples. This assay was validated on an independent blinded set of 451 cytological and pathological samples. Results: Using the expression levels of 8 microRNAs measured in qRT-PCR, accurate classification of the primary lung tumors into the four main cancer types is obtained. The microRNA-based assay reached an accuracy of 94%. Moreover, cytological samples composed over 50% of the validation set and reached an accuracy of 95%. Conclusions: We present here a new microRNA-based assay for the classification of the four main types of lung cancer based only on the expression of 8 microRNAs. This assay displays very high levels of accuracy for both pathological and cytological samples. The assay comprises a standardized, well-tested and objective tool which can assist physicians in the diagnosis of lung cancer.

scholarly journals Automated communication system for detection of lung cancer using catastrophe features

2020 ◽  
Vol 53 (3-4) ◽  
pp. 184-190
Author(s):  
Ramaiah Arun ◽  
Shanmugasundaram Singaravelan
Keyword(s):  
Lung Cancer ◽  
Support Vector ◽  
Color Spaces ◽  
Lung Cancers ◽  
Clustering Techniques ◽  
Early Stages ◽  
Fuzzy C Means Clustering ◽  
Glandular Cells ◽  
Medical Advances

One of the biggest challenges the world face today is the mortality due to Cancer. One in four of all diagnosed cancers involve the lung cancer, where the mortality rate is high, even after so much of technical and medical advances. Most lung cancer cases are diagnosed either in the third or fourth stage, when the disease is not treatable. The main reason for the highest mortality, due to lung cancer is because of non availability of prescreening system which can analyze the cancer cells at early stages. So it is necessary to develop a prescreening system which helps doctors to find and detect lung cancer at early stages. Out of all various types of lung cancers, adenocarcinoma is increasing at an alarming rate. The reason is mainly attributed to the increased rate of smoking - both active and passive. In the present work, a system for the classification of lung glandular cells for early detection of Cancer using multiple color spaces is developed. For segmentation, various clustering techniques like K-Means clustering and Fuzzy C-Means clustering on various Color spaces such as HSV, CIELAB, CIEXYy and CIELUV are used. Features are Extracted and classified using Support Vector Machine (SVM).

scholarly journals Towards the 9th Edition of the Tumour, Node and Metastasis Classification of Lung Cancer. A Historical Appraisal and Future Perspectives

2019 ◽  
Vol 1 (2) ◽  
Author(s):  
Ramón Rami-Porta
Keyword(s):  
Lung Cancer ◽  
International Association ◽  
Tumour Size ◽  
Primary Tumour ◽  
Cancer Control ◽  
Regional Lymph Nodes ◽  
Lung Cancers ◽  
Lung Cancer Patients ◽  
Pathologic Staging

Since 1966 the classification of anatomic extent of lung cancer, based on the primary tumour (T), the loco-regional lymph nodes (N) and the metastases (M) has been used in the management of lung cancer patients. Developed by Pierre Denoix, it was adopted by the Union for International Cancer Control and the American Joint Committee on Cancer. Clifton Mountain revised the second through the sixth editions based on a North American database of more than 5000 patients. For the seventh and the eighth editions, the International Association for the Study of Lung Cancer (IASLC) collected international databases of around 100,000 patients worldwide that allowed the introduction of innovations in both editions, namely the subdivision of the T and M categories based on tumour size and on the location and number of metastases, respectively. The revisions also showed the prognostic relevance of the quantification of nodal disease, and proposed recommendations on how to measure tumour size for solid lung cancers, part-solid adenocarcinomas, and for lung cancers removed after induction therapy. Despite the innovations, prognosis based on the anatomic extent is limited, because prognosis depends on factors related to the tumour, the patient and the environment. For the 9th edition, these factors, especially genetic biomarkers, will be combined in prognostic groups to refine prognosis at clinical and pathologic staging. To achieve this challenging objective, international cooperation is essential, and the IASLC Staging and Prognostic Factors Committee counts on it for the development of the 9th edition due to be published in 2024.

scholarly journals An annotation-free whole-slide training approach to pathological classification of lung cancer types using deep learning

2021 ◽  
Vol 12 (1) ◽  
Author(s):  
Chi-Long Chen ◽  
Chi-Chung Chen ◽  
Wei-Hsiang Yu ◽  
Szu-Hua Chen ◽  
Yu-Chan Chang ◽  
...  
Keyword(s):  
Lung Cancer ◽  
Deep Learning ◽  
Characteristic Curve ◽  
Digital Pathology ◽  
Classification Performance ◽  
Multiple Instance Learning ◽  
Data Set ◽  
Memory Mechanism ◽  
Strong Localization ◽  
Training Approach

AbstractDeep learning for digital pathology is hindered by the extremely high spatial resolution of whole-slide images (WSIs). Most studies have employed patch-based methods, which often require detailed annotation of image patches. This typically involves laborious free-hand contouring on WSIs. To alleviate the burden of such contouring and obtain benefits from scaling up training with numerous WSIs, we develop a method for training neural networks on entire WSIs using only slide-level diagnoses. Our method leverages the unified memory mechanism to overcome the memory constraint of compute accelerators. Experiments conducted on a data set of 9662 lung cancer WSIs reveal that the proposed method achieves areas under the receiver operating characteristic curve of 0.9594 and 0.9414 for adenocarcinoma and squamous cell carcinoma classification on the testing set, respectively. Furthermore, the method demonstrates higher classification performance than multiple-instance learning as well as strong localization results for small lesions through class activation mapping.

Towards the 9th Edition of the Tumour, Node and Metastasis Classification of Lung Cancer. A Historical Appraisal and Future Perspectives.

2019 ◽  
Author(s):  
Ramón RAMI-PORTA
Keyword(s):  
Lung Cancer ◽  
International Association ◽  
Tumour Size ◽  
Primary Tumour ◽  
Cancer Control ◽  
Regional Lymph Nodes ◽  
Lung Cancers ◽  
Lung Cancer Patients ◽  
Pathologic Staging

Since 1966 the classification of anatomic extent of lung cancer, based on the primary tumour (T), the loco-regional lymph nodes (N) and the metastases (M) has been used in the management of lung cancer patients. Developed by Pierre Denoix, it was adopted by the Union for International Cancer Control and the American Joint Committee on Cancer. Clifton Mountain revised the second through the sixth editions based on a North American database of more than 5000 patients. For the seventh and the eighth editions, the International Association for the Study of Lung Cancer (IASLC) collected international databases of around 100,000 patients worldwide that allowed the introduction of innovations in both editions, namely the subdivision of the T and M categories based on tumour size and on the location and number of metastases, respectively. The revisions also showed the prognostic relevance of the quantification of nodal disease, and proposed recommendations on how to measure tumour size for solid lung cancers, part-solid adenocarcinomas, and for lung cancers removed after induction therapy. Despite the innovations, prognosis based on the anatomic extent is limited, because prognosis depends on factors related to the tumour, the patient and the environment. For the 9th edition, these factors, especially genetic biomarkers, will be combined in prognostic groups to refine prognosis at clinical and pathologic staging. To achieve this challenging objective, international cooperation is essential, and the IASLC Staging and Prognostic Factors Committee counts on it for the development of the 9th edition due to be published in 2024.

scholarly journals A deep learning model for the classification of indeterminate lung carcinoma in biopsy whole slide images

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Fahdi Kanavati ◽  
Gouji Toyokawa ◽  
Seiya Momosaki ◽  
Hiroaki Takeoka ◽  
Masaki Okamoto ◽  
...  
Keyword(s):  
Lung Cancer ◽  
Deep Learning ◽  
Area Under The Curve ◽  
Learning Model ◽  
Primary Sources ◽  
Independent Test ◽  
Hematoxylin And Eosin ◽  
Deep Learning Model ◽  
Whole Slide Images

AbstractThe differentiation between major histological types of lung cancer, such as adenocarcinoma (ADC), squamous cell carcinoma (SCC), and small-cell lung cancer (SCLC) is of crucial importance for determining optimum cancer treatment. Hematoxylin and Eosin (H&E)-stained slides of small transbronchial lung biopsy (TBLB) are one of the primary sources for making a diagnosis; however, a subset of cases present a challenge for pathologists to diagnose from H&E-stained slides alone, and these either require further immunohistochemistry or are deferred to surgical resection for definitive diagnosis. We trained a deep learning model to classify H&E-stained Whole Slide Images of TBLB specimens into ADC, SCC, SCLC, and non-neoplastic using a training set of 579 WSIs. The trained model was capable of classifying an independent test set of 83 challenging indeterminate cases with a receiver operator curve area under the curve (AUC) of 0.99. We further evaluated the model on four independent test sets—one TBLB and three surgical, with combined total of 2407 WSIs—demonstrating highly promising results with AUCs ranging from 0.94 to 0.99.

scholarly journals Impact of Recent Developments in Lung Cancer on the Practice of Pathology

2016 ◽  
Vol 140 (4) ◽  
pp. 322-325 ◽  
Author(s):  
Philip T. Cagle ◽  
Timothy C. Allen ◽  
Eric H. Bernicker ◽  
Yimin Ge ◽  
Abida Haque ◽  
...  
Keyword(s):  
Lung Cancer ◽  
World Health ◽  
World Health Organization Classification ◽  
Lung Cancers ◽  
The Us ◽  
Recent Developments ◽  
Health Organization ◽  
Key Events ◽  
Low Dose Computed Tomography

Landmark events in the field of lung cancer in the past year have the potential to significantly alter the practice of pathology. Three key events are (1) approval of payment for low-dose computed tomography screening for lung cancer, (2) publication of an extensively revised World Health Organization classification of lung cancers, and (3) approval of immunohistochemistry based companion diagnostics by the US Food and Drug Administration. We briefly review these milestones in the context of their impact on the practice of pathology.

Sign in / Sign up

Export Citation Format

Share Document